{ "version": "1.8.0", "slug": "special-needs-trust-able-disbursement-compliance-desk", "title": "BenefitGuard — Special Needs Trust & ABLE Disbursement Compliance Desk", "vertical": "Elder & disability-benefits compliance / special needs trust (SNT) and ABLE account disbursement compliance for professional fiduciaries", "seed": { "s": "special-needs-trust-able-disbursement-compliance-desk", "t": "BenefitGuard — Special Needs Trust & ABLE Disbursement Compliance Desk", "v": "Elder & disability-benefits compliance / special needs trust (SNT) and ABLE account disbursement compliance for professional fiduciaries", "r": "Medium-high", "m": "M" }, "product": { "slug": "special-needs-trust-able-disbursement-compliance-desk", "project_name": "BenefitGuard", "project_type": "done-for-you, per-disbursement special needs trust (SNT) and ABLE account compliance desk", "vertical": "Elder & disability-benefits compliance / special needs trust (SNT) and ABLE account disbursement compliance for professional fiduciaries", "audience": "professional fiduciaries (independent professional trustees, corporate trust-company trust officers, elder-law-firm trust-administration practices) administering 15+ active SNTs and/or ABLE-linked accounts (primary ICP); pooled-trust nonprofits (Arc chapters, similar d4C operators) with large sub-account portfolios (expansion ICP); settlement-planning firms establishing first-party SNTs from injury settlements (expansion ICP)", "geography": "United States — federal SSI/Medicaid rules (SSA POMS) apply nationally; state Medicaid-supplement variation is codified state by state as pilot fiduciaries onboard, starting with the founder's home state plus federal baseline", "scale_expectation": "5-8 professional fiduciary or pooled-trust accounts for the founding pilot cohort, capped explicitly to protect analyst/attorney-bench bandwidth; a hardening checkpoint after 5 pilots, a second after 10 pilots, and a pause-and-measure checkpoint after 20 pilots until cost-per-determination, rework rate, escalation rate, and cycle time are formally measured — continuous per-transaction demand, not a single filing deadline", "core_workflows": [ "Disbursement Request intake and completeness check", "Bright-Line Check + Rule Matrix classification and Determination Memo drafting", "100% Benefits-Compliance Analyst review, with automatic Exception Queue escalation", "CELA/licensed elder-law attorney sign-off on novel, high-value, or Certified Opinion requests", "Portfolio Monitoring Retainer tracking and Quarterly Portfolio Review" ], "discovery": { "one_line_purpose": "On BenefitGuard, a professional fiduciary sends one Disbursement Request — the amount, purpose, payee, and the beneficiary's current SSI/SSDI/Medicaid status and state — and gets back, within one business day (rush: 4 hours), a dated, cited Determination Memo: approve, approve-with-conditions, or do-not-disburse, with a Benefit-Impact Projection and, where relevant, a Restructuring Alternative. Every AI-drafted determination is reviewed by a Benefits-Compliance Analyst before delivery, and a licensed elder-law attorney signs any novel, high-risk, or Certified Opinion determination.", "primary_users": [ "professional fiduciary / trust officer / trust-department principal who personally bears fiduciary liability for a wrongful distribution", "Benefits-Compliance Analyst (the trained human who reviews every AI-drafted determination and approves every delivery)", "CELA / licensed elder-law attorney (external bench, engaged for escalated and Certified Opinion determinations)" ], "jobs_to_be_done": [ "When a beneficiary or family requests a disbursement, tell me whether paying it puts their SSI or Medicaid at risk — and if it's risky, tell me exactly how to structure it so it isn't — without me becoming an amateur benefits-law expert for every one of my trusts", "Stop losing sleep over whether the 'harmless' rent payment I approved last month just quietly triggered an SSI suspension I won't discover until the beneficiary's next redetermination", "Be the fiduciary whose files are always audit-ready, not the one explaining to a court or a state bar why a beneficiary lost benefits on my watch", "Make sure I'm applying this year's rules and this beneficiary's actual state's Medicaid-supplement variant — not stale guidance from before the 2026 ABLE age expansion changed who I'm now serving" ], "value_prop": "BenefitGuard turns a proposed SNT or ABLE disbursement into a dated, cited, audit-ready Determination Memo — the fiduciary sends the request; a Benefits-Compliance Analyst reviews every AI-drafted determination before delivery; a licensed elder-law attorney signs any novel, high-value, or Certified Opinion case. BenefitGuard is paid per disbursement request or a flat monthly retainer, never hourly.", "competitors": [ "True Link Financial — spending-control prepaid cards plus an attorney-facing compliance guide the trustee must configure and monitor themselves, not a per-transaction determination service", "Full-service professional fiduciary/corporate trustee firms (e.g., CPT Institute-style) — take on the entire trusteeship for AUM-plus-itemized fees, not just the disbursement-compliance-check function", "Elder-law one-time consultation at trust drafting — legal advice at trust creation, billed hourly, not structured for ongoing per-transaction review", "The status quo: a fiduciary checking a disbursement against memory or a general-purpose AI chatbot with no maintained jurisdiction-specific rule matrix, no audit trail, and no licensed sign-off" ], "differentiation": "Done-for-you, per-disbursement compliance research, not a tool the fiduciary must operate: a maintained, versioned Rule Matrix (federal POMS + state Medicaid-supplement + ABLE QDE guidance) with citation-level traceability, a Benefits-Compliance Analyst's sign-off on 100% of determinations, a licensed attorney's Certified Opinion tier, per-request/subscription pricing with a free diagnostic entry point, and a positioning built around the fiduciary remaining trustee of record — never surrendering the trusteeship relationship.", "positioning_statement": "For professional fiduciaries who personally carry the liability for every dollar a special needs trust or ABLE account pays out, BenefitGuard is the done-for-you disbursement compliance desk that turns a proposed expenditure into a dated, cited Determination Memo, a Benefit-Impact Projection, and — where relevant — a Restructuring Alternative, with a licensed attorney's sign-off available on demand — unlike a spending-control card the fiduciary must configure themselves, or a corporate trustee relationship that requires surrendering the entire trusteeship." }, "assumptions": [ { "id": "special-needs-trust-able-disbursement-compliance-desk-a1", "statement": "Professional fiduciaries will pay a flat per-request or retainer fee for a documented, cited disbursement determination once a free diagnostic check demonstrates the format and rigor on a real request.", "confidence": "medium", "impact_if_wrong": "severe", "revisit_trigger": "First 5-8 pilot fiduciaries' free-diagnostic-to-paid conversion." }, { "id": "special-needs-trust-able-disbursement-compliance-desk-a2", "statement": "A 1-business-day standard / 4-hour rush turnaround is achievable from fiduciary-supplied request details and benefit-status confirmation alone, without a live SSA/Medicaid data integration.", "confidence": "medium", "impact_if_wrong": "high", "revisit_trigger": "First 10-15 completed determinations, manually-assisted." }, { "id": "special-needs-trust-able-disbursement-compliance-desk-a3", "statement": "Fiduciaries tolerate AI-assisted classification and drafting when a Benefits-Compliance Analyst's review is explicit and mandatory on every determination before delivery.", "confidence": "medium", "impact_if_wrong": "high", "revisit_trigger": "First pilot-cohort feedback survey round." }, { "id": "special-needs-trust-able-disbursement-compliance-desk-a4", "statement": "A state-by-state Rule Matrix build-out (starting with the founder's home state plus federal POMS baseline) can expand fast enough to match pilot-cohort geographic footprint without blocking onboarding.", "confidence": "medium", "impact_if_wrong": "medium", "revisit_trigger": "First 3-5 states added at Day 30." }, { "id": "special-needs-trust-able-disbursement-compliance-desk-a5", "statement": "The Certified Opinion tier ($350 flat) is economical against actual attorney time on novel/high-risk cases.", "confidence": "low", "impact_if_wrong": "medium", "revisit_trigger": "First 5 Certified Opinions delivered — track actual attorney minutes per case." } ], "unknowns": [ { "id": "special-needs-trust-able-disbursement-compliance-desk-u1", "question": "What is the true free-diagnostic-to-paid conversion rate among warm, qualified professional fiduciaries?", "blocks": "Pricing hypothesis; scaling decision past the 5-8-account pilot cap.", "resolution_path": "5-8-account pilot cohort with an explicit free-diagnostic conversation, measured at the Day-90 retrospective." }, { "id": "special-needs-trust-able-disbursement-compliance-desk-u2", "question": "How much does Rule Matrix maturity actually reduce Benefits-Compliance Analyst review minutes by Day 90?", "blocks": "Unit-economics model; analyst-throughput assumptions.", "resolution_path": "Instrument analyst minutes per determination from pilot 1; compare to the 20min->10min target in ai-engine-spec.md." }, { "id": "special-needs-trust-able-disbursement-compliance-desk-u3", "question": "Will the pending SSI Savings Penalty Elimination Act (H.R.2540/S.1234, 119th Congress) pass and change the $2,000 resource limit this business's core compliance stakes partly rest on?", "blocks": "Long-term relevance of the resource-limit Bright-Line Check specifically (ISM and ABLE QDE compliance needs persist regardless).", "resolution_path": "Rule Matrix change-monitoring; re-verify each Congressional session per compliance-checklist.md." } ], "expert_panel": [ { "role": "Product Strategy", "key_concern": "Is the single-disbursement Determination Memo narrow enough to sell in weeks without a custom SSA/Medicaid data integration?", "recommendation": "Launch scoped to one disbursement request via email/upload intake, no live benefit-status API; expand only after the Portfolio Monitoring Retainer proves demand past 8 pilots.", "dissent": "May understate the always-on portfolio-monitoring story to prospects expecting continuous monitoring from day one." }, { "role": "Software Architecture", "key_concern": "Are we defaulting to a fiduciary-facing portal the pilot doesn't need yet?", "recommendation": "Single deployable modular monolith; email/upload-link intake and email/document delivery only at launch, no fiduciary-facing dashboard.", "dissent": "May read as under-built to a pooled-trust nonprofit partner expecting a branded dashboard." }, { "role": "Frontend / UX", "key_concern": "Is the citation chain (POMS/state-rule to Bright-Line Check to verdict) visible in every analyst-facing view?", "recommendation": "Every Determination Memo draft renders its cited rule, confidence score, and Bright-Line Check result in one glance before the analyst signs.", "dissent": "Adds density a newly hired analyst might call cluttered before onboarding." }, { "role": "Backend / Data", "key_concern": "Is the Audit Trail append-only and reconstructable years later for an SSA/Medicaid inquiry?", "recommendation": "Event-sourced audit log for every Disbursement Request lifecycle state transition; hash-chained Determination Memo snapshots retained per the engagement-letter-defined retention period.", "dissent": "More write-path complexity than plain CRUD; requires schema discipline as new states are onboarded." }, { "role": "AI / ML", "key_concern": "Are verdicts grounded in the Rule Matrix and never a free-form guess?", "recommendation": "Retrieval-first classification citing the specific POMS/state-rule provision and confidence score; deterministic Bright-Line Checks (resource limit, ISM formula, ABLE caps) in code, never the model; mandatory analyst gate before delivery.", "dissent": "Slower per-request than a fully automated tool; may frustrate a fiduciary expecting instant turnaround on every request." }, { "role": "Security", "key_concern": "Is one fiduciary's beneficiary financial/benefit-status data isolated from every other fiduciary's, and is the no-credential-collection rule enforced structurally?", "recommendation": "Row-level auth, per-fiduciary retrieval partitioning, log-scrubbing of beneficiary PII and benefit-status figures, and a hard code-level guard that blocks any workflow from ever storing an SSA `my Social Security` or ABLE portal credential.", "dissent": "None recorded this run." } ], "strategy": { "business_model": "Standard determination $45/rush $95 per Disbursement Request (never hourly); Portfolio Monitoring Retainer $60/trust/month (bundles up to 3 standard determinations); Certified Opinion (CELA sign-off) $350 flat; Quarterly Portfolio Review from $500 per 25 trusts/accounts; free Disbursement Compliance Check as the lead magnet", "revenue_streams": [ "Standard/rush per-Disbursement-Request determinations ($45/$95)", "Portfolio Monitoring Retainer ($60/trust/month, bundles up to 3 determinations)", "Certified Opinion — CELA sign-off (flat $350)", "Quarterly Portfolio Review (from $500 per 25 trusts/accounts)" ], "moat": [ "Rule Matrix — versioned, citation-anchored federal POMS + state Medicaid-supplement + ABLE QDE rule sets that compound with every state added and every determination adjudicated", "A Benefits-Compliance Analyst's 100%-review track record and a contracted CELA/elder-law attorney bench — operational persistence no self-serve tool replicates", "Audit Trail lock-in — the evidence of due diligence a fiduciary needs for their own liability protection, growing more valuable every determination on file", "Operated-outcome positioning against a spending-control card product the fiduciary would have to configure and monitor themselves" ], "gtm": [ "Founder-led educational content (10-post series on POMS/ISM/ABLE mechanics) anchored to the free Disbursement Compliance Check", "NAELA/state elder-law bar section and pooled-trust nonprofit outreach (warm shortlist of 3-5 targets)", "Answer-engine/AEO-optimized content answering exact buyer phrasing ('can my SNT pay for X', 'will this disbursement affect SSI')" ], "pricing_hypothesis": "Per-request pricing anchored just below CPT Institute's documented $35-75 disbursement-processing fee precedent, with the Portfolio Monitoring Retainer anchored to CPT Institute's $50+/month trustee-fee-inclusive-of-disbursement-processing figure — both re-confirmed directly against cptinstitute.org in this run's fresh research; the Certified Opinion priced as a flat attorney-work-product fee, Unverified for economics until real attorney-time data exists and validated in the first 5 delivered Certified Opinions.", "kill_criteria": [ "Fewer than 3 signed founding-pilot accounts after 30 days of qualified outreach", "Benefits-Compliance Analyst median review time exceeds the QC-gate targets after pilot account 5 (delivery-playbook.md)", "No Portfolio Monitoring Retainer conversion by Day 90", "A wrong determination contributes to a documented benefit-suspension incident — a zero-tolerance, business-ending signal" ] }, "security": { "stride": [ { "threat": "Spoofing", "scenario": "Attacker impersonates a Benefits-Compliance Analyst to approve a fabricated Determination Memo or release a Certified Opinion.", "mitigation": "SSO with MFA; analyst and attorney sign-offs bound to a cryptographic session claim, not a form field." }, { "threat": "Tampering", "scenario": "Historical Determination Memo or Audit Trail entries edited after the fact to hide an incorrect verdict.", "mitigation": "Append-only audit log; hash-chained Determination Memo snapshots; diff view on every analyst surface." }, { "threat": "Repudiation", "scenario": "A Benefits-Compliance Analyst or attorney denies approving a determination that later proves wrong.", "mitigation": "Signed attestations with server-side timestamp + reviewer identity; export bundle includes a signature manifest." }, { "threat": "Information Disclosure", "scenario": "Cross-fiduciary leak of one fiduciary's beneficiary financial/benefit-status data through shared indices, logs, or prompts.", "mitigation": "Fiduciary-scoped row-level auth; PII/financial-data scrubbing in logs; retrieval indices partitioned per fiduciary." }, { "threat": "Denial of Service", "scenario": "A runaway extraction job or a bulk portfolio-review batch exhausts shared workers during a quarterly-review volume spike.", "mitigation": "Per-tenant concurrency + budget caps; circuit breaker on model calls; degrade-gracefully queue." }, { "threat": "Elevation of Privilege", "scenario": "A standard operator acquires analyst-approval or attorney-sign-off capability via a workflow shortcut.", "mitigation": "Roles stored in a separate table; capability checks server-side; no client-only role checks; a hard code-level guard blocks any role from ever storing a beneficiary/family credential." } ], "privacy_posture": "Data-minimization by default; per-fiduciary isolation; a signed engagement letter on file before any request is worked; no secondary use or sale of beneficiary data; no SSA `my Social Security` or ABLE portal credential is ever collected, stored, or requested — a documented, hard-coded prohibition.", "compliance_targets": [ "SSA POMS trust/income/ISM/overpayment rules (SI 01120.200, SI 02301.205, SI 01130.740) codified with citation-level traceability", "IRC §529A (ABLE Act, as amended by the ABLE Age Adjustment Act, SECURE 2.0, effective Jan 1 2026) and IRS/SSA QDE guidance", "State Medicaid-supplement rule variation codified state by state, attorney-reviewed before any state is trusted for live determinations", "Unauthorized-practice-of-law boundary reviewed with counsel before the first paid engagement (see compliance-checklist.md)" ], "data_classifications": [ "Beneficiary benefit-status data (sensitive; SSI/SSDI/Medicaid enrollment)", "Disbursement request, trust, and financial data (sensitive financial)", "Trust governing-document and ABLE program terms (sensitive, third-party-sourced)", "Fiduciary contact / staff PII (standard)" ] }, "devops": { "ci_cd": "PR -> typecheck + unit + snapshot tests -> preview deploy -> main auto-deploys to a single production region; migrations gated on review.", "environments": [ "local", "preview (per-PR)", "staging (shared)", "production (single region + multi-AZ)" ], "observability": [ "Structured logs with fiduciary + request IDs", "RED metrics per workflow", "Error tracking with source maps", "Model-call spans with cost + latency", "Weekly SLO review, with a quarterly-review volume-spike capacity review" ], "testing_pyramid": [ "Unit tests on Bright-Line Check modules (resource limit, ISM formula, ABLE contribution caps)", "Component tests on the Benefits-Compliance Analyst exception-queue surfaces", "Contract tests on the Disbursement Request schema per Rule Matrix state entry", "End-to-end smoke test on the intake -> classify -> analyst sign-off -> delivery path" ], "accessibility_tests": [ "axe-core in CI on analyst-facing surfaces", "Keyboard-only walkthrough per workflow", "Prefers-reduced-motion honored" ], "performance_budget": "p95 workflow latency published per module; request extraction cold-path under 30s or shown as a background job." }, "accessibility_i18n_ethics": { "wcag_target": "AA", "locales": [ "en-US" ], "rtl_support": false, "ethical_risks": [ "A Determination Memo released without Benefits-Compliance Analyst approval", "A memo read as legal advice to a beneficiary or family rather than compliance research delivered to the fiduciary customer of record", "An error in a delivered determination attributed to AI rather than owned by the reviewing analyst" ], "ethical_guardrails": [ "Human Benefits-Compliance Analyst approval required before any Determination Memo ships — no exceptions", "Deterministic scope-of-review disclaimer attached to every memo distinguishing compliance research from individualized legal advice", "Engagement letter discloses AI-assisted drafting; no beneficiary/family credentials are ever requested or stored" ] }, "governance": { "ownership": [ { "area": "Product + roadmap", "owner": "Founder" }, { "area": "Architecture + platform", "owner": "Engineering Lead" }, { "area": "Determination + exception workflow", "owner": "Named Benefits-Compliance Analyst" }, { "area": "Compliance + attorney bench", "owner": "Contracted CELA / elder-law attorney" }, { "area": "Design system", "owner": "Design Lead" }, { "area": "SEO + content", "owner": "Content Lead" } ], "docs_required": [ "ADR log (checked in)", "Owner-action ledger", "Evidence register", "STRIDE threat model", "Runbook: incident, restore, breach notification", "Benefits-Compliance Analyst playbook + sign-off policy" ], "naming_conventions": [ "kebab-case slugs for blueprints and routes", "camelCase for TypeScript identifiers", "SCREAMING_SNAKE_CASE for environment variables", "Verb-first action names (e.g., escalate-to-attorney-bench)" ], "change_control": "ADR-per-major-decision; migrations require review; production deploys gated on green CI + owner-action ledger check." }, "risk_register": [ { "id": "special-needs-trust-able-disbursement-compliance-desk-r1", "risk": "A state Medicaid-supplement rule change breaks a codified determination path mid-year", "likelihood": "high", "impact": "medium", "mitigation": "Rule Matrix versioned per state; quarterly attorney-reviewed change-monitoring; any tracked source change triggers review of affected prior determinations", "contingency": "Emergency Rule Matrix patch within 7 days of a detected change; affected fiduciaries notified same week", "owner": "product" }, { "id": "special-needs-trust-able-disbursement-compliance-desk-r2", "risk": "AI misclassification or a low-confidence draft is delivered without adequate analyst review", "likelihood": "low", "impact": "high", "mitigation": "Mandatory 100% analyst review, no automated-delivery path; deterministic Bright-Line Check engine for the highest-consequence thresholds", "contingency": "Analyst-triggered rollback + regeneration; incident postmortem shared internally within 48 hours", "owner": "engineering" }, { "id": "special-needs-trust-able-disbursement-compliance-desk-r3", "risk": "Cross-fiduciary beneficiary financial/benefit-status data leak", "likelihood": "low", "impact": "severe", "mitigation": "Row-level auth + per-fiduciary retrieval indices + log-scrubbing", "contingency": "Breach-notification runbook; forced credential rotation; scoped fiduciary kill switch", "owner": "engineering" }, { "id": "special-needs-trust-able-disbursement-compliance-desk-r4", "risk": "Unauthorized-practice-of-law exposure on the non-certified tier", "likelihood": "medium", "impact": "high", "mitigation": "Counsel-reviewed engagement-letter scoping positioning standard determinations as compliance research delivered to the fiduciary, not individualized legal advice; defined escalation criteria route borderline legal-judgment cases to the attorney bench automatically", "contingency": "Route all engagements through attorney sign-off temporarily if a state bar objection arises", "owner": "owner" }, { "id": "special-needs-trust-able-disbursement-compliance-desk-r5", "risk": "True Link Financial or another well-funded competitor launches a directly competing documented-determination product", "likelihood": "low-medium", "impact": "medium", "mitigation": "Rule Matrix, attorney-sign-off tier, and Audit Trail are not a feature a card-product competitor can bolt on overnight", "contingency": "Reposition around the operational grind and liability-protection depth a card product leaves the fiduciary to build itself", "owner": "owner" } ], "roadmap": [ { "phase": "Phase 0 — Stand up the operating spine", "weeks": "Week 1 (Days 1-7)", "outcomes": [ "Rule Matrix v0 (founder's home state + federal POMS baseline) built and attorney-reviewed", "Disbursement Compliance Check landing page + intake form live and tested", "Engagement-letter template drafted; CELA/elder-law attorney-bench relationship opened; 5 warm outreaches sent" ], "exit_criteria": [ "One full test determination processed end-to-end on sample data", "Counsel sign-off on the engagement-letter template", "Intake mailbox tested with one dummy submission" ], "kill_criteria": [ "Any operating-spine item not ready by Day 7 — outreach held even if it slips past Day 7" ] }, { "phase": "Phase 1 — Onboard the pilot cohort, run the first determinations", "weeks": "Weeks 2-4 (Days 8-30)", "outcomes": [ "First 3-5 paid pilot determinations delivered manually", "Attorney-bench agreement signed", "3 educational content pieces published", "Rule Matrix expanded to 3-5 states based on actual pilot footprint" ], "exit_criteria": [ "Pilot cap enforced at 5-8 accounts regardless of inbound demand", "Founder personally reviews every exception before delivery", "Exception queue and communications log stood up" ], "kill_criteria": [ "Fewer than 3 signed pilots after 30 days of qualified outreach", "No willingness-to-engage signal above the free-diagnostic offer" ] }, { "phase": "Phase 2 — Hold the cap, prove the model", "weeks": "Weeks 5-8 (Days 31-60)", "outcomes": [ "5-8-account pilot cap reached", "AI drafting layer introduced against a growing adjudicated-example library, 100% analyst review retained", "First Portfolio Monitoring Retainer conversions", "5-pilot hardening checkpoint completed" ], "exit_criteria": [ "1-business-day standard turnaround hit on every determination", "Zero benefit-suspension incidents traced to a delivered determination", "Certified Opinion delivered on at least 1 escalated case" ], "kill_criteria": [ "Pilot-cohort diagnostic-to-paid conversion stalls far below target", "Rework rate persistently above 5%" ] }, { "phase": "Phase 3 — Formalize SOPs, evaluate the 10-pilot checkpoint", "weeks": "Weeks 9-13 (Days 61-90)", "outcomes": [ "Second Benefits-Compliance Analyst certified independent of the founder", "Written SOPs, escalation matrix, and QA sampling rules formalized", "Quarterly Portfolio Review piloted with at least 1 account", "90-day retrospective completed; 10-pilot checkpoint evaluated with real data" ], "exit_criteria": [ "Analyst shadows Founder on at least 5 live determinations before unsupervised exception review", "10-pilot hardening checklist complete (intake spec, evidence requirements, QA checklists)", "Gross margin trending toward the blueprint's 50-55% target" ], "kill_criteria": [ "Escalation-to-attorney rate exceeds 30% with no improving trend", "Analyst review time not declining by pilot 15-20" ] } ], "metrics": { "north_star": "% of delivered determinations meeting the 1-business-day (rush: 4-hour) SLA with zero benefit-suspension incidents traced to a BenefitGuard-approved determination", "leading": [ "Founding-pilot engagement letters signed", "Free-diagnostic-to-paid conversion rate", "Benefits-Compliance Analyst review minutes per determination", "Rule Matrix state-coverage completeness", "Escalation rate to the attorney bench" ], "lagging": [ "Standard/rush determinations delivered", "Portfolio Monitoring Retainer accounts signed", "Pilot-to-retainer conversion (target 40-60%)", "Determinations lost to missed SLA (target: zero)" ], "guardrails": [ "Model spend per determination", "PII/financial/benefit-status data in logs (target: 0)", "Cross-fiduciary access attempts (target: 0)", "Escalation-to-outside-counsel rate outside the defined criteria (target: 0)" ] }, "executive_review": { "consensus": "Ship a thin, evidence-linked, analyst-gated vertical slice for BenefitGuard, wedged on the single-Disbursement-Request Determination Memo. Prefer a boring, single-region modular monolith. Do not launch past the founding pilot cohort until the Rule Matrix is codified from federal POMS plus the founder's home state and at least one attorney-signed Certified Opinion has resolved.", "dissent": "Founder and Counsel disagree on state-expansion tempo — Counsel recommends waiting for each new state's Medicaid-supplement rules to be fully attorney-reviewed before onboarding any fiduciary in that state, even inside the pilot cap.", "go_no_go": "conditional-go", "top_3_risks": [ "State Medicaid-supplement rule changes break a codified determination path mid-year", "AI misclassification delivered without adequate analyst review", "Unauthorized-practice-of-law exposure on the non-certified tier" ], "first_10_steps": [ "Stand up the intake mailbox, engagement-letter template, and Disbursement Compliance Check landing page", "Populate the owner-action ledger with entity, jurisdiction, contact inbox, privacy inbox", "Open founding-pilot intake (cap 5-8) with an explicit free-diagnostic conversation", "Write the home-state-first, multi-state-second sequencing and kill/pivot criteria into the launch plan", "Codify federal POMS (SI 01120.200, SI 02301.205, SI 01130.740) into the Rule Matrix v0 with citation anchors", "Wire the append-only audit log + hash-chained Determination Memo snapshots", "Add row-level auth + fiduciary-scoped retrieval indices", "Instrument diagnostic-request, determination-delivered, and analyst-sign-off events", "Publish the microsite noindexed until owner-action facts close", "Schedule the Day-90 kill/pivot review with the Founder" ] } }, "project_site": { "slug": "special-needs-trust-able-disbursement-compliance-desk", "app_name": "BenefitGuard disbursement compliance desk", "archetype": "compliance determination desk", "archetype_label": "special needs trust & ABLE disbursement compliance desk", "reader_role": "Professional Fiduciary / Trust Officer", "one_sentence_app": "BenefitGuard is a special needs trust and ABLE disbursement compliance desk for professional fiduciaries who need a dated, cited, audit-ready determination on every proposed distribution before it quietly costs a beneficiary their SSI or Medicaid.", "homepage_sequence": [ "scene", "workflow", "instrument", "offer", "proof", "qualification", "objections" ], "hero": { "frame_label": "Elder & disability-benefits compliance · special needs trust & ABLE disbursement compliance desk", "eyebrow": "Professional fiduciary or trust officer / a beneficiary or family just requested a disbursement and you're not sure it's safe to approve", "interface_title": "BenefitGuard disbursement compliance desk", "primary_panel_title": "Disbursement Request #BG-2026-000118 · ABLE education expense · Day 0 of 1-business-day SLA", "primary_panel_body": "Open determination: classify this request against the current Rule Matrix, run the Bright-Line Checks, and stage the Determination Memo for Benefits-Compliance Analyst release.", "side_panel_title": "Before this ships", "side_panel_items": [ "Beneficiary's current SSI/SSDI/Medicaid status and state confirmed", "Bright-Line Check (resource limit / ISM formula / ABLE cap) run and logged", "Benefits-Compliance Analyst has signed off on the verdict and citation trail" ], "status_metric": "REVIEW", "status_label": "analyst release gate active" }, "language": { "problem_heading": "The fiduciary's moment", "mechanism_heading": "Inside the BenefitGuard determination desk", "proof_heading": "Why this survives an SSA inquiry or a state Medicaid review", "offer_heading": "What leaves the room", "objection_heading": "The hard questions", "qualification_heading": "Who should not use this", "cta_close": "Get your free Disbursement Compliance Check — reviewed by a real Benefits-Compliance Analyst, cited to the current Rule Matrix." }, "modules": [ { "name": "Disbursement Compliance Check", "job": "Turns one proposed disbursement into a cited compliance verdict within one business day, at no cost for the first request.", "artifact": "free diagnostic determination" }, { "name": "Determination Memo", "job": "The dated, cited, audit-ready verdict — approve, approve-with-conditions, or do-not-disburse — with a Benefit-Impact Projection.", "artifact": "delivered determination memo" }, { "name": "Quarterly Portfolio Review", "job": "A full-portfolio sweep for retainer accounts catching drift a beneficiary's changed status or a state rule update introduces.", "artifact": "portfolio review report" } ], "checkpoints": [ { "label": "Bright-Line Check", "pass": "deterministic threshold clear, request proceeds to AI classification", "fail": "forces do-not-disburse or approve-with-conditions with a Restructuring Alternative" }, { "label": "confidence threshold", "pass": "at or above the analyst-review confidence band", "fail": "routed to the Exception Queue for attorney escalation regardless of dollar value" }, { "label": "Determination Memo release", "pass": "Benefits-Compliance Analyst signs (attorney co-signs on escalated/certified cases)", "fail": "stays in review" } ], "signature_scene": "You're the trust officer for a mid-sized professional fiduciary firm administering 40 special needs trusts and ABLE accounts across three states. A family just asked to withdraw $3,200 from a beneficiary's ABLE account to pay a landlord directly for six months of rent in advance. You know ABLE and SNT rules aren't the same, you know the ISM formula matters somewhere in this, and you have forty other trusts that need the same kind of check this month, with no in-house benefits-compliance specialist on staff. This page is built like the compliance desk that request needed before you had to guess." }, "ddd": { "slug": "special-needs-trust-able-disbursement-compliance-desk", "project_name": "BenefitGuard", "business_understanding": { "summary": "BenefitGuard — a professional fiduciary sends a proposed disbursement request and its supporting facts, and BenefitGuard classifies it against the current Rule Matrix, runs deterministic Bright-Line Checks, drafts a cited Determination Memo, routes any novel/high-value/low-confidence case to a licensed elder-law attorney, and delivers a dated, audit-ready verdict — approve, approve-with-conditions, or do-not-disburse — with a Benefit-Impact Projection and, where relevant, a Restructuring Alternative.", "customer_profile": "Professional fiduciaries (independent trustees, trust-company trust officers, elder-law firm trust departments) administering 15+ active SNTs and/or ABLE accounts; pooled-trust nonprofits with large sub-account portfolios; settlement-planning firms establishing first-party SNTs — none with an in-house benefits-compliance specialist reviewing every disbursement.", "customer_pain": "Direct cash to a beneficiary counts as unearned income; in-kind shelter payments trigger an SSI reduction via the ISM presumed-maximum-value formula; leaving a beneficiary holding more than $2,000 in countable resources at the wrong moment suspends SSI and creates a retroactive overpayment; ABLE accounts carry a parallel QDE compliance regime and a Medicaid-payback claim at death. Elder-law and settlement-planning practices describe these mistakes as common, good-faith trustee errors, not rare edge cases — a routine operational hazard most fiduciaries have no systematic check against.", "paid_outcome": "A dated, cited, audit-ready Determination Memo for a proposed disbursement, protecting the beneficiary's SSI/Medicaid eligibility and the fiduciary's own liability position — reconstructable on demand, defensible to SSA or a state Medicaid agency.", "value_creation": "AI compresses request extraction, rule-matrix classification, and first-pass determination drafting; deterministic code owns every Bright-Line Check and dollar calculation; humans own Benefits-Compliance Analyst sign-off, attorney escalation judgment, and anything that touches a novel legal question; the platform binds them with an append-only Audit Trail and per-fiduciary isolation.", "why_ai_native": "The classify-draft-review-sign loop is only economical with a bounded AI layer feeding deterministic rules. A pure-manual analyst hire cannot hit per-request pricing at the volumes a growing fiduciary portfolio generates; a pure-automated tool cannot hold the analyst approval gate, absorb genuinely novel fact patterns, or provide the licensed attorney sign-off that gives a fiduciary real liability protection.", "operational_risks": [ "Benefits-Compliance Analyst bottleneck during a synchronized portfolio-review volume spike", "Prompt injection via uploaded disbursement-request documents or trust governing documents", "Cross-fiduciary retrieval leak, including beneficiary benefit-status data", "Incomplete facts unobtainable at scale (a request stuck in evidence-gap status past the 1-business-day SLA)", "A determination read as individualized legal advice to a beneficiary rather than compliance research delivered to the fiduciary customer of record" ], "assumptions": [ "Professional fiduciaries will pay a per-request fee for evidence-linked, human-reviewed disbursement compliance research over continuing to decide on memory or a general chatbot", "The workflow can be completed from fiduciary-supplied facts and the Rule Matrix within the 1-business-day standard / 4-hour rush targets", "The vertical tolerates AI-assisted drafts when Benefits-Compliance Analyst approval is explicit and every delivery is authorized under a signed engagement letter", "Logical, per-fiduciary claim isolation (not dedicated infrastructure per fiduciary) is acceptable for the first cohort", "Demand persists as a continuous, per-transaction need, not a one-time event — unlike a single-deadline filing business" ], "validation_questions": [ "Which states are in scope for the founding cohort's Rule Matrix coverage?", "What is the actual free-diagnostic-to-paid conversion rate at launch pricing vs. standard pricing?", "Which existing systems (fiduciary case-management tools, email) must this integrate with day one?", "Is there a data-residency constraint for any pilot fiduciary's beneficiary data?", "When does a given state's Medicaid-supplement rule set stabilize enough to codify a Rule Matrix entry with confidence?" ] }, "domain_discovery": { "actors": [ { "actor": "Professional Fiduciary / Trust Officer (economic buyer)", "role": "Owns the disbursement-compliance engagement; supplies request facts and authorizes the determination process", "goals": [ "Approve only disbursements that don't jeopardize the beneficiary's SSI/Medicaid", "Avoid becoming an amateur benefits-law expert for every trust", "Have a defensible, audit-ready file if a distribution is later questioned" ], "decisions": [ "Whether to engage BenefitGuard", "Which facts/documents to supply per request", "Whether to sign the engagement letter" ], "pain_points": [ "No time to research POMS/state rules for every request", "Doesn't know which requests are genuinely risky vs. routine" ] }, { "actor": "Benefits-Compliance Analyst", "role": "Reviews AI-drafted determinations; signs off before any delivery; triages the Exception Queue", "goals": [ "Release only evidence-backed, correctly classified determinations", "Protect the beneficiary's benefits and the fiduciary's liability position", "Escalate genuinely ambiguous cases, not just dollar-threshold cases" ], "decisions": [ "Approve, edit, or escalate a drafted determination", "Decide whether an ambiguous case needs attorney review even below the dollar threshold" ], "pain_points": [ "Volume spikes during portfolio-review cycles", "Genuinely novel fact patterns requiring judgment beyond the Rule Matrix" ] }, { "actor": "CELA / Licensed Elder-Law Attorney (contracted bench)", "role": "Reviews and signs novel, high-value, or Certified Opinion determinations", "goals": [ "Resolve the escalated determination correctly and defensibly" ], "decisions": [ "Sign, request more facts, or decline to certify a determination" ], "pain_points": [ "Receiving an escalation without a complete fact and citation history" ] }, { "actor": "SSA / State Medicaid Agency (external)", "role": "The regulator whose rules govern every determination; may later question a distribution", "goals": [ "Correct benefit administration without improper distributions" ], "decisions": [ "Approve, question, or suspend benefits based on a beneficiary's actual disbursement history" ], "pain_points": [ "High volume of trust/ABLE cases across many fiduciaries and states" ] } ], "glossary": [ { "term": "Disbursement Compliance Check", "definition": "The free, 1-business-day diagnostic: a fiduciary's first real disbursement request in, a cited compliance verdict out.", "used_by": "Professional Fiduciary / Trust Officer", "context": "Intake", "example": "Disbursement Compliance Check on a $3,200 ABLE rent request", "notes": "Never called 'assessment', 'audit', or 'review'." }, { "term": "Disbursement Request", "definition": "The tracked unit of work: one proposed trust or ABLE expenditure submitted for review.", "used_by": "Benefits-Compliance Analyst, Fiduciary", "context": "Determination", "example": "Disbursement Request #BG-2026-000118, ABLE education expense, Day 0 of 1-business-day SLA", "notes": "Never called 'case', 'ticket', or 'claim'." }, { "term": "Determination Memo", "definition": "The dated, cited, audit-ready deliverable: approve / approve-with-conditions / do-not-disburse.", "used_by": "Benefits-Compliance Analyst, Fiduciary, Attorney bench", "context": "Delivery", "example": "Determination Memo #BG-2026-000118: approve-with-conditions, cites POMS SI 01120.200", "notes": "Never called 'report', 'summary', or bare 'opinion'." }, { "term": "Rule Matrix", "definition": "The maintained, versioned federal POMS + state Medicaid-supplement + ABLE QDE knowledge base every determination is classified against.", "used_by": "AI + rules, Benefits-Compliance Analyst, Attorney bench", "context": "Determination", "example": "Rule Matrix v2026.3 adds the Texas Medicaid-supplement variant", "notes": "Never called 'database' or bare 'rules engine' in customer-facing copy." } ], "decisions": [ { "decision": "Does this Disbursement Request pass its applicable Bright-Line Checks?", "who": "Deterministic code, no human judgment", "inputs": [ "[PLACEHOLDER] owner to complete" ], "rule": "Resource-limit, ISM presumed-maximum-value, and ABLE-cap checks run in code against the Rule Matrix's current thresholds; never left to model inference", "output": "Bright-Line Check pass/fail with a computed dollar impact where applicable", "risk": "A stale threshold value produces a systematically wrong check across every request until corrected" }, { "decision": "Is this determination ready to deliver, or does it need attorney escalation?", "who": "AI drafts and scores confidence, Benefits-Compliance Analyst confirms, escalation criteria route automatically", "inputs": "Drafted verdict, confidence score, dollar amount, novelty of fact pattern, Certified Opinion flag", "rule": "Any of: confidence below threshold, dollar amount above the state's high-value threshold, unrepresented fact pattern, or Certified Opinion requested — routes to the attorney bench regardless of how routine the case otherwise looks", "output": "Delivered Determination Memo or an escalated case with a logged reason", "risk": "Under-escalation delivers a determination that should have had attorney judgment; over-escalation burns attorney-bench capacity on routine cases" } ], "events": [ { "event": "DisbursementRequestSubmitted", "meaning": "A fiduciary's Disbursement Request and its supporting facts are received and ready for a completeness check", "trigger": "Fiduciary submits via secure portal or email", "downstream": [ "[PLACEHOLDER] owner to complete" ] }, { "event": "DeterminationDrafted", "meaning": "The AI engine has classified the request against the current Rule Matrix and produced a first-pass verdict, citations, and confidence score", "trigger": "Classification & drafting step completes", "downstream": "Enters the Benefits-Compliance Analyst review queue" } ] }, "subdomains": [ { "name": "Determination", "type": "core", "description": "Request classification against the Rule Matrix, Bright-Line Check execution, and Determination Memo drafting.", "reason": "This is the paid outcome's first gate — the verdict determines everything downstream.", "business_value": "Direct revenue driver; every priced determination starts here.", "recommendation": "Build in-house, deterministic-rules-first.", "ai_involvement": "Extraction, classification, drafting, confidence scoring", "human_involvement": "Benefits-Compliance Analyst review and approval of every draft", "risks": [ "Misclassified request causing a wrong verdict", "Incomplete facts causing an analyst-caught gap" ], "validation_questions": [ "What is the real analyst review time per determination at volume?" ] }, { "name": "Review & Escalation", "type": "core", "description": "Benefits-Compliance Analyst sign-off, Exception Queue triage, and attorney-bench escalation and Certified Opinion sign-off.", "reason": "Differentiates BenefitGuard from a fiduciary DIY-ing a chatbot check — this is the human-and-licensed-attorney discipline layer.", "business_value": "Drives the zero-benefit-suspension-incident north-star metric and the Certified Opinion revenue line.", "recommendation": "Build in-house; maintain the escalation matrix from day one.", "ai_involvement": "Escalation-triage scoring", "human_involvement": "Benefits-Compliance Analyst approves every determination; attorney bench signs escalated/certified cases", "risks": [ "Missed escalation on a genuinely novel case", "Attorney-bench bottleneck at volume" ], "validation_questions": [ "How often does an escalated case actually change the AI-drafted verdict?" ] }, { "name": "Portfolio Compliance", "type": "core", "description": "Portfolio Monitoring Retainer tracking, Quarterly Portfolio Review assembly, and drift detection across a fiduciary's full book.", "reason": "This is the human-chokepoint moat that produces the switching-cost lock-in the business is built on.", "business_value": "Directly drives renewal and expansion revenue.", "recommendation": "Build thin tooling (auto-assembly from the Audit Trail); keep the final review entirely human.", "ai_involvement": "Auto-assembles the review from the Audit Trail and flags drift patterns", "human_involvement": "Benefits-Compliance Analyst final review and spot-check before delivery", "risks": [ "A drift pattern is missed if review scope is incomplete", "Escalating too early or too late on a portfolio-level flag" ], "validation_questions": [ "What share of portfolio-level drift is actually preventable with a tighter review cadence?" ] }, { "name": "Rule Matrix", "type": "supporting", "description": "Maintained, per-jurisdiction versioned rule set (federal POMS, state Medicaid-supplement variants, ABLE QDE guidance) with citation anchors and change-monitoring.", "reason": "Necessary complement to Determination and Review but not itself the sold outcome.", "business_value": "Compounds accuracy and speed with every state added and every determination adjudicated.", "recommendation": "Build incrementally from real pilot-cohort state footprint.", "ai_involvement": "Assembles and diffs rule profiles from regulatory-source updates", "human_involvement": "Attorney reviews every new state entry and every diff before it's trusted for live determinations", "risks": [ "Stale rule applied after a regulatory update" ], "validation_questions": [ "Which states need re-verification most often as legislative sessions close?" ] }, { "name": "Referral & Outreach", "type": "generic", "description": "CRM, outbound sequencing, and content publishing.", "reason": "Not differentiating; off-the-shelf tools suffice at this stage.", "business_value": "Supports pipeline but not the core value proposition.", "recommendation": "Buy (lightweight CRM), do not build.", "ai_involvement": "Content drafting assistance", "human_involvement": "Founder-led outreach and content", "risks": [ "Low differentiation risk if under-invested" ], "validation_questions": [ "Which channel (NAELA, pooled-trust directory, outbound) actually converts at the assumed rate?" ] } ], "core_domain_analysis": { "primary_core": "Determination", "secondary_cores": [ "Review & Escalation", "Portfolio Compliance" ], "supporting_may_become_core": [ "Rule Matrix" ], "generic_do_not_distract": [ "Referral & Outreach", "CRM/pipeline tracking" ], "rationale": "The paid outcome is a correctly classified, analyst-reviewed, and (where needed) attorney-signed Determination Memo, so Determination is the primary core. Review & Escalation and Portfolio Compliance are secondary cores because the human-plus-attorney discipline and the Audit Trail/retainer lock-in are the moat that differentiates BenefitGuard from a self-serve chatbot check or a full corporate-trustee relationship. The Rule Matrix could become core as it matures into a licensable, multi-state data asset, but starts as supporting. Referral & Outreach uses off-the-shelf tooling deliberately to keep engineering focus on the determination-review-portfolio engine." }, "bounded_contexts": [ { "name": "Intake", "purpose": "Turn a fiduciary's submitted disbursement request and supporting facts into a normalized, completeness-checked DisbursementRequest record ready for classification.", "subdomain": "Determination (core)", "type": "core", "owned_language": [ "Disbursement Request", "Disbursement Compliance Check" ], "owns": [ "Disbursement request intake lifecycle", "Normalization state", "Required-fact completeness" ], "does_not_own": [ "Rule Matrix classification", "Analyst sign-off" ], "primary_actors": [ "Professional Fiduciary / Trust Officer", "Extraction Agent" ], "entities": [ "DisbursementRequest" ], "value_objects": [ "ConfidenceScore", "DocumentsReceived" ], "aggregates": [ "DisbursementRequestAggregate" ], "domain_services": [ "Request normalizer", "Completeness checker" ], "application_services": [ "SubmitDisbursementRequest", "NormalizeRequestData", "CheckIntakeCompleteness" ], "commands": [ "SubmitDisbursementRequest", "NormalizeRequestData" ], "domain_events": [ "disbursement.request_submitted", "request.data_normalized", "request.completeness_check_failed" ], "policies": [ "Auto-flag a missing required fact or document against the request type's evidence checklist" ], "specifications": [ "RequiredEvidenceCompletenessSpec" ], "invariants": [ "A DisbursementRequest cannot proceed to classification until its required-evidence checklist is satisfied or a named exception is logged" ], "ai_agents": [ "Extraction Agent" ], "human_roles": [ "Professional Fiduciary / Trust Officer" ], "data_owned": [ "DisbursementRequest" ], "inputs": [ "Disbursement description, amount, and payee", "Beneficiary benefit-status confirmation", "Trust/ABLE governing document" ], "outputs": [ "Normalized canonical disbursement-request record" ], "external_integrations": [ "Monitored intake mailbox", "Secure upload form" ], "risks": [ "A malformed submission silently mis-extracted", "Missing evidence for the request type goes unnoticed" ], "interfaces": [ "-> Determination (RequestDataNormalized)" ] }, { "name": "Determination", "purpose": "Run Bright-Line Checks, classify the request against the Rule Matrix, and draft the Determination Memo with citations and a confidence score.", "subdomain": "Determination (core)", "type": "core", "owned_language": [ "Determination Memo", "Benefit-Impact Projection", "Bright-Line Check", "Restructuring Alternative" ], "owns": [ "Classification engine", "Bright-Line Check execution", "Determination drafting" ], "does_not_own": [ "Analyst identity", "Billing" ], "primary_actors": [ "Benefits-Compliance Analyst", "Classification Agent" ], "entities": [ "DeterminationMemo" ], "value_objects": [ "ConfidenceScore", "BenefitImpactAmount" ], "aggregates": [ "DeterminationMemoAggregate" ], "domain_services": [ "Classification engine", "Bright-Line Check engine" ], "application_services": [ "RunBrightLineChecks", "ClassifyAndDraftDetermination" ], "commands": [ "RunBrightLineChecks", "ClassifyAndDraftDetermination" ], "domain_events": [ "brightline.check_completed", "determination.drafted", "determination.confidence_scored" ], "policies": [ "Auto-route confidence below threshold to the Exception Queue regardless of dollar value" ], "specifications": [ "RequiredEvidenceCompletenessSpec", "CitationRequiredSpec" ], "invariants": [ "A Bright-Line Check result is never overridden by AI classification", "No dollar figure in a Determination Memo is ever LLM-computed" ], "ai_agents": [ "Classification Agent", "Drafting Agent" ], "human_roles": [ "Benefits-Compliance Analyst" ], "data_owned": [ "DeterminationMemo" ], "inputs": [ "Normalized disbursement-request data", "Rule Matrix" ], "outputs": [ "Draft Determination Memo with citations and confidence score" ], "external_integrations": [ "Rule Matrix (internal)" ], "risks": [ "Misclassification against a stale Rule Matrix version", "An unsupported citation reaching the analyst queue" ], "interfaces": [ "-> Review & Escalation (DeterminationDrafted)" ] }, { "name": "Review & Escalation", "purpose": "Enforce 100% analyst review, triage the Exception Queue, and route novel/high-value/low-confidence/Certified Opinion cases to the attorney bench for sign-off.", "subdomain": "Review & Escalation (core)", "type": "core", "owned_language": [ "Exception Queue", "Certified Opinion", "Benefits-Compliance Analyst" ], "owns": [ "Review queue", "Escalation triggers", "Attorney sign-off lifecycle" ], "does_not_own": [ "Classification logic (upstream)" ], "primary_actors": [ "Benefits-Compliance Analyst", "CELA / Elder-Law Attorney" ], "entities": [ "DeterminationMemo", "EscalationCase" ], "value_objects": [ "EscalationReasonCode", "ReviewStatus" ], "aggregates": [ "EscalationCaseAggregate" ], "domain_services": [ "Escalation triage engine" ], "application_services": [ "ReviewDeterminationDraft", "EscalateToAttorneyBench", "SignDetermination", "DeliverDeterminationMemo" ], "commands": [ "EscalateToAttorneyBench", "SignDetermination", "DeliverDeterminationMemo" ], "domain_events": [ "determination.analyst_reviewed", "RequestEscalated", "DeterminationSigned", "DeterminationDelivered" ], "policies": [ "100% of drafted determinations require analyst review before delivery, no exceptions", "Every Certified Opinion requires attorney sign-off before delivery" ], "specifications": [ "EscalationCriteriaSpec" ], "invariants": [ "A Determination Memo cannot be delivered without at least one Benefits-Compliance Analyst sign-off", "A Certified Opinion cannot be delivered without a licensed attorney's signature" ], "ai_agents": [ "Escalation-Triage Agent" ], "human_roles": [ "Benefits-Compliance Analyst", "CELA / Elder-Law Attorney" ], "data_owned": [ "EscalationCase" ], "inputs": [ "Draft Determination Memo", "Escalation criteria thresholds" ], "outputs": [ "Analyst-approved or attorney-signed Determination Memo, delivered" ], "external_integrations": [ "Contracted attorney bench (e-signature)" ], "risks": [ "Under-escalation of a genuinely novel case", "Attorney-bench turnaround bottleneck at volume" ], "interfaces": [ "-> Portfolio Compliance (DeterminationDelivered)", "-> Rule Matrix (DeterminationOutcomeLogged)" ] }, { "name": "Portfolio Compliance", "purpose": "Track Portfolio Monitoring Retainer accounts, assemble the Quarterly Portfolio Review, and detect drift across a fiduciary's full trust/ABLE portfolio.", "subdomain": "Portfolio Compliance (core)", "type": "core", "owned_language": [ "Portfolio Monitoring Retainer", "Quarterly Portfolio Review", "Audit Trail" ], "owns": [ "Retainer lifecycle", "Quarterly review assembly", "Drift detection" ], "does_not_own": [ "Individual determination drafting (upstream)" ], "primary_actors": [ "Benefits-Compliance Analyst", "Portfolio Drift-Detection Agent" ], "entities": [ "PortfolioReview" ], "value_objects": [ "DriftScore" ], "aggregates": [ "PortfolioReviewAggregate" ], "domain_services": [ "Portfolio drift detector", "Review assembler" ], "application_services": [ "RunQuarterlyPortfolioReview", "TrackRetainerUsage" ], "commands": [ "RunQuarterlyPortfolioReview" ], "domain_events": [ "PortfolioDriftDetected", "portfolio_review.assembled" ], "policies": [ "Any portfolio-level pattern crossing a defined drift threshold is flagged even if each underlying determination looked fine individually" ], "specifications": [ "PortfolioReviewCompletenessSpec" ], "invariants": [ "A Quarterly Portfolio Review cannot be delivered without Benefits-Compliance Analyst final review" ], "ai_agents": [ "Portfolio Drift-Detection Agent" ], "human_roles": [ "Benefits-Compliance Analyst" ], "data_owned": [ "PortfolioReview" ], "inputs": [ "Full determination history for a fiduciary's portfolio" ], "outputs": [ "Delivered Quarterly Portfolio Review with flagged drift patterns" ], "external_integrations": [ "Fiduciary case-management export (as needed)" ], "risks": [ "A drift pattern missed if review scope excludes older determinations" ], "interfaces": [ "-> Rule Matrix (RuleMatrixVersionPublished consumed)" ] }, { "name": "Rule Matrix", "purpose": "Maintain the per-jurisdiction (federal POMS, state Medicaid-supplement, ABLE QDE) versioned rule set and citation anchors used by every other context.", "subdomain": "Rule Matrix (supporting)", "type": "supporting", "owned_language": [ "Rule Matrix", "Rule Matrix Version" ], "owns": [ "Jurisdiction rule sets", "Citation anchors", "Change-monitoring log" ], "does_not_own": [ "Determination decisions" ], "primary_actors": [ "CELA / Elder-Law Attorney" ], "entities": [ "RuleMatrixEntry" ], "value_objects": [ "RuleVersionTag", "CitationAnchor" ], "aggregates": [ "RuleMatrixVersionAggregate" ], "domain_services": [ "Rule set updater", "Change-diff detector" ], "application_services": [ "UpdateRuleMatrix" ], "commands": [ "UpdateRuleMatrix" ], "domain_events": [ "RuleMatrixVersionPublished" ], "policies": [ "Quarterly attorney-reviewed re-verification of every active jurisdiction's rules" ], "specifications": [ "RuleMatrixFreshnessSpec" ], "invariants": [ "Every rule entry carries a source citation and a last-confirmed date" ], "ai_agents": [], "human_roles": [ "CELA / Elder-Law Attorney" ], "data_owned": [ "RuleMatrixEntry" ], "inputs": [ "SSA POMS updates", "State Medicaid-supplement rule changes", "IRS/SSA ABLE QDE guidance updates" ], "outputs": [ "Updated per-jurisdiction rule profiles" ], "external_integrations": [], "risks": [ "Stale rule applied after a regulatory update" ], "interfaces": [ "<- Review & Escalation (DeterminationOutcomeLogged)", "<- Portfolio Compliance (PortfolioDriftDetected)" ] } ], "context_map": [ { "upstream": "Intake", "downstream": "Determination", "relationship": "Customer-Supplier", "integration_pattern": "Domain event (RequestDataNormalized) consumed by the classification engine", "translation_notes": "Canonical disbursement-request schema is the shared contract; no direct database coupling.", "pattern": "Customer-Supplier", "contract_type": "domain event", "data_exchanged": [ "[PLACEHOLDER] owner to complete" ], "events_exchanged": [ "Domain event (RequestDataNormalized) consumed by the classification engine" ], "ownership_boundary": "Downstream never writes to upstream's tables", "acl_notes": "Anti-corruption translation not needed — shared canonical schema", "business_reason": "Keeps each context's invariants enforceable independently", "failure_risks": [ "[PLACEHOLDER] owner to complete" ] }, { "upstream": "Determination", "downstream": "Review & Escalation", "relationship": "Customer-Supplier", "integration_pattern": "Domain event (DeterminationDrafted) triggers analyst review", "translation_notes": "Only classified, cited drafts cross the boundary.", "pattern": "Customer-Supplier", "contract_type": "domain event", "data_exchanged": "Canonical schema payload only", "events_exchanged": [ "Domain event (DeterminationDrafted) triggers analyst review" ], "ownership_boundary": "Downstream never writes to upstream's tables", "acl_notes": "Anti-corruption translation not needed — shared canonical schema", "business_reason": "Keeps each context's invariants enforceable independently", "failure_risks": "Downstream stalls if upstream event delivery fails" }, { "upstream": "Review & Escalation", "downstream": "Portfolio Compliance", "relationship": "Customer-Supplier", "integration_pattern": "Domain event (DeterminationDelivered) feeds the portfolio review", "translation_notes": "Only fully delivered determinations enter the portfolio history.", "pattern": "Customer-Supplier", "contract_type": "domain event", "data_exchanged": "Canonical schema payload only", "events_exchanged": [ "Domain event (DeterminationDelivered) feeds the portfolio review" ], "ownership_boundary": "Downstream never writes to upstream's tables", "acl_notes": "Anti-corruption translation not needed — shared canonical schema", "business_reason": "Keeps each context's invariants enforceable independently", "failure_risks": "Portfolio review stalls if a delivery event is dropped" }, { "upstream": "Rule Matrix", "downstream": "Determination", "relationship": "Shared Kernel (published language)", "integration_pattern": "Versioned rule-set read (thresholds, citation anchors, jurisdiction coverage)", "translation_notes": "Determination reads the current Rule Matrix version; never writes to it directly.", "pattern": "Conformist", "contract_type": "published schema", "data_exchanged": "Rule Matrix entries (thresholds, citations)", "events_exchanged": [ "RuleMatrixVersionPublished" ], "ownership_boundary": "Determination never writes to the Rule Matrix", "acl_notes": "None needed — Rule Matrix is the canonical source", "business_reason": "One place to update a regulatory change so every context stays consistent", "failure_risks": "Stale rule version used if the update event is missed" } ], "external_integrations": [ { "system": "Secure intake portal/email", "direction": "inbound", "pattern": "manual upload/email (cycle 1+), no live SSA/Medicaid API integration at launch", "risk": "Malformed or incomplete submission format", "owner_context": "Intake", "data_in": [ "[PLACEHOLDER] owner to complete" ], "data_out": [ "[PLACEHOLDER] owner to complete" ], "trigger": "Every new Disbursement Request", "internal_model": "Canonical disbursement-request schema", "acl_strategy": "Parser + LLM normalization layer translates any submission format to the canonical schema", "failure_strategy": "Malformed submission routes to the exception queue with a named parsing gap", "audit_need": "Every submission state observed is timestamped in the Audit Trail" }, { "system": "E-signature tool (attorney sign-off)", "direction": "outbound", "pattern": "signature request on every escalated/certified determination", "risk": "Signature not captured before delivery", "owner_context": "Review & Escalation", "data_in": "None", "data_out": "Determination Memo for signature", "trigger": "Escalation or Certified Opinion request", "internal_model": "EscalationCase", "acl_strategy": "Direct e-signature API integration", "failure_strategy": "Delivery blocked pending a captured signature", "audit_need": "Signature manifest retained with the delivered memo" }, { "system": "Lightweight CRM (pipeline/renewal tracking)", "direction": "bidirectional", "pattern": "off-the-shelf tool, not built in-house", "risk": "Low differentiation risk if under-invested", "owner_context": "Referral & Outreach", "data_in": "Prospect and account contact data", "data_out": "None sensitive", "trigger": "Outreach and renewal cadence", "internal_model": "N/A — external tool", "acl_strategy": "Standard CRM integration, no beneficiary data stored here", "failure_strategy": "Manual pipeline tracking fallback", "audit_need": "None — no sensitive data in this system" }, { "system": "OCR/document-extraction tool (near-term addition)", "direction": "inbound", "pattern": "receipt/invoice ingestion to support extraction", "risk": "OCR misread on a poor-quality scan", "owner_context": "Intake", "data_in": "Receipt/invoice images or PDFs", "data_out": "None", "trigger": "Disbursement request with supporting documents", "internal_model": "Canonical disbursement-request schema", "acl_strategy": "OCR output routed through the same confidence-scored extraction layer as manual entry", "failure_strategy": "Low-confidence OCR output routes to analyst confirmation", "audit_need": "Original document retained alongside the extracted record" } ], "event_storm": [ { "seq": 1, "command": "SubmitDisbursementRequest", "event": "disbursement.request_submitted", "actor": "Professional Fiduciary / Trust Officer", "context": "Intake", "aggregate": "DisbursementRequest", "policy": "Completeness checklist auto-enforced before any review starts", "downstream": "Normalization begins", "risk": "Incomplete request data blocks the compliance check" }, { "seq": 2, "command": "NormalizeRequestData", "event": "request.data_normalized", "actor": "Extraction Agent", "context": "Intake", "aggregate": "DisbursementRequest", "policy": "Confidence-scored field extraction; below-threshold fields flagged", "downstream": "Classification begins", "risk": "Malformed submission silently mis-extracted" }, { "seq": 3, "command": "RunBrightLineChecks", "event": "brightline.check_completed", "actor": "Deterministic rule engine", "context": "Determination", "aggregate": "DeterminationMemo", "policy": "Resource-limit/ISM/ABLE-cap checks run in code, never model-inferred", "downstream": "AI classification proceeds with the Bright-Line result as a hard constraint", "risk": "A stale threshold value produces a systematically wrong check" }, { "seq": 4, "command": "ClassifyAndDraftDetermination", "event": "determination.drafted", "actor": "Classification Agent / Drafting Agent", "context": "Determination", "aggregate": "DeterminationMemo", "policy": "Confidence below threshold forces Exception Queue routing", "downstream": "Analyst review queue", "risk": "Misclassification against a stale Rule Matrix version" }, { "seq": 5, "command": "ReviewDeterminationDraft / EscalateToAttorneyBench", "event": "determination.analyst_reviewed / RequestEscalated", "actor": "Benefits-Compliance Analyst", "context": "Review & Escalation", "aggregate": "EscalationCase", "policy": "100% human approval on every determination, no exceptions", "downstream": "Determination approved, or escalated to the attorney bench", "risk": "Under-escalation of a genuinely novel case" }, { "seq": 6, "command": "SignDetermination / DeliverDeterminationMemo", "event": "DeterminationSigned / DeterminationDelivered", "actor": "CELA / Elder-Law Attorney / Benefits-Compliance Analyst", "context": "Review & Escalation", "aggregate": "EscalationCase", "policy": "Certified Opinion requires attorney signature before delivery", "downstream": "Portfolio Compliance logs the delivered determination", "risk": "A signed-but-uncommunicated escalation delays delivery past the SLA" } ], "critical_path": [ "Intake -> Determination -> Review & Escalation (analyst or attorney sign-off) -> Portfolio Compliance, gated by a Benefits-Compliance Analyst sign-off on every delivery" ], "exception_flows": [ "A classification scores below the confidence threshold -> routed to the Exception Queue for attorney escalation before any memo is delivered", "A required fact (e.g., beneficiary's current benefit-status confirmation) is missing -> blocked from classification and logged as a named exception until the fiduciary supplies it" ], "escalation_flows": [ "A Bright-Line Check fails (resource limit, ISM formula, ABLE cap) -> forces a do-not-disburse or approve-with-conditions draft with a mandatory Restructuring Alternative attempt", "A dispute or novel legal question that could reach litigation -> hard stop at a documented Determination Memo, referred to the fiduciary's own counsel; BenefitGuard does not advocate legally" ], "retry_flows": [ "An incomplete request with no fiduciary response after 2 business days triggers an automated follow-up request, held for analyst visibility" ], "manual_override_flows": [ "A Benefits-Compliance Analyst may manually reclassify a request's purpose category when new facts change the correct Rule Matrix citation" ], "commands": [ { "name": "SubmitDisbursementRequest", "issued_by": "Professional Fiduciary / Trust Officer", "preconditions": [ "Engagement letter signed", "Request facts supplied" ], "aggregate": "DisbursementRequest", "success_event": "disbursement.request_submitted", "failure_event": "request.rejected_incomplete", "authorization": "Fiduciary-authorized submission only", "validation": "Completeness checklist", "audit": "Intake timestamp + submitted-fact manifest logged" }, { "name": "NormalizeRequestData", "issued_by": "Extraction Agent (analyst confirms below threshold)", "preconditions": [ "Request submitted" ], "aggregate": "DisbursementRequest", "success_event": "request.data_normalized", "failure_event": "request.extraction_below_threshold", "authorization": "System; analyst required below confidence threshold", "validation": "ConfidenceScore >= threshold or analyst sign-off", "audit": "Extraction rationale + confidence score logged" }, { "name": "RunBrightLineChecks", "issued_by": "Deterministic rule engine", "preconditions": [ "Request data normalized" ], "aggregate": "DeterminationMemo", "success_event": "brightline.check_completed", "failure_event": "brightline.check_blocked_missing_data", "authorization": "System only — no human or AI override", "validation": "Bright-Line thresholds from the current Rule Matrix version", "audit": "Check result + rule version logged per request" }, { "name": "ClassifyAndDraftDetermination", "issued_by": "Classification Agent / Drafting Agent", "preconditions": [ "Bright-Line Checks completed" ], "aggregate": "DeterminationMemo", "success_event": "determination.drafted", "failure_event": "determination.drafting_failed", "authorization": "System; analyst review required before delivery regardless of outcome", "validation": "CitationRequiredSpec — every claim must cite a source", "audit": "Draft + citations + confidence score logged" }, { "name": "EscalateToAttorneyBench", "issued_by": "Benefits-Compliance Analyst / automatic escalation rule", "preconditions": [ "Escalation criteria met (novel/high-value/low-confidence/certified)" ], "aggregate": "EscalationCase", "success_event": "RequestEscalated", "failure_event": "escalation.blocked_no_available_attorney", "authorization": "Analyst or automatic rule trigger", "validation": "EscalationCriteriaSpec", "audit": "Escalation reason + timestamp logged" }, { "name": "DeliverDeterminationMemo", "issued_by": "Benefits-Compliance Analyst (attorney co-signs on escalated/certified cases)", "preconditions": [ "Analyst sign-off recorded", "Attorney sign-off recorded if escalated" ], "aggregate": "EscalationCase", "success_event": "DeterminationDelivered", "failure_event": "delivery.blocked_missing_signoff", "authorization": "Analyst sign-off required; attorney sign-off required for escalated/certified cases", "validation": "Full citation + signature manifest present", "audit": "Delivered memo + signature manifest retained" } ], "policies": [ { "name": "Bright-Line resource-limit gate", "trigger": "A disbursement would leave the beneficiary holding more than $2,000 in countable resources", "condition": "Deterministic check against current countable resources plus the proposed disbursement's effect", "action": "Force a do-not-disburse or approve-with-conditions verdict with a mandatory Restructuring Alternative attempt", "context": "Determination", "ai_involvement": "None — deterministic code only", "human_approval": false }, { "name": "Low-confidence escalation", "trigger": "A determination is drafted", "condition": "ConfidenceScore below the defined threshold", "action": "Route automatically to the attorney bench regardless of dollar value", "context": "Determination", "ai_involvement": "Propose classification + confidence score", "human_approval": true }, { "name": "100%-analyst-review requirement", "trigger": "Any determination is drafted", "condition": "Always", "action": "No Determination Memo is delivered without a Benefits-Compliance Analyst's recorded sign-off", "context": "Review & Escalation", "ai_involvement": "Draft only", "human_approval": true }, { "name": "Certified Opinion attorney sign-off", "trigger": "A Certified Opinion is requested", "condition": "Always", "action": "Route to the attorney bench; block delivery until a licensed attorney signs", "context": "Review & Escalation", "ai_involvement": "Draft only", "human_approval": true } ], "aggregates": [ { "name": "DisbursementRequestAggregate", "root": "DisbursementRequest", "context": "Intake", "purpose": "Represents one proposed trust or ABLE expenditure and its supporting facts prior to classification.", "entities": [ "SupportingDocument" ], "value_objects": [ "ConfidenceScore", "DocumentsReceived" ], "invariants": [ "A DisbursementRequest cannot advance to Determination until required facts are present or exception-logged" ], "commands": [ "SubmitDisbursementRequest", "NormalizeRequestData" ], "events": [ "disbursement.request_submitted", "request.data_normalized" ], "repository": "DisbursementRequestRepository (per-fiduciary partitioned)", "transaction_boundary": "One request per commit" }, { "name": "DeterminationMemoAggregate", "root": "DeterminationMemo", "context": "Determination", "purpose": "Represents one drafted-through-delivered compliance verdict for a Disbursement Request, its citations, and its Benefit-Impact Projection.", "entities": [ "Citation" ], "value_objects": [ "ConfidenceScore", "BenefitImpactAmount" ], "invariants": [ "A Bright-Line Check result is never overridden by AI classification", "No dollar figure is ever LLM-computed" ], "commands": [ "RunBrightLineChecks", "ClassifyAndDraftDetermination" ], "events": [ "brightline.check_completed", "determination.drafted" ], "repository": "DeterminationMemoRepository (per-fiduciary partitioned)", "transaction_boundary": "One DeterminationMemo per commit" }, { "name": "EscalationCaseAggregate", "root": "EscalationCase", "context": "Review & Escalation", "purpose": "Represents one determination's review-and-sign-off lifecycle, including any attorney escalation.", "entities": [ "SignatureRecord" ], "value_objects": [ "EscalationReasonCode" ], "invariants": [ "A Determination Memo cannot be delivered without an analyst sign-off", "A Certified Opinion cannot be delivered without an attorney signature" ], "commands": [ "EscalateToAttorneyBench", "SignDetermination", "DeliverDeterminationMemo" ], "events": [ "RequestEscalated", "DeterminationSigned", "DeterminationDelivered" ], "repository": "EscalationCaseRepository (per-fiduciary partitioned)", "transaction_boundary": "One EscalationCase per commit" }, { "name": "PortfolioReviewAggregate", "root": "PortfolioReview", "context": "Portfolio Compliance", "purpose": "Represents one fiduciary's quarterly, full-portfolio compliance sweep and any detected drift.", "entities": [ "DriftFinding" ], "value_objects": [ "DriftScore" ], "invariants": [ "A Quarterly Portfolio Review cannot be delivered without analyst final review" ], "commands": [ "RunQuarterlyPortfolioReview" ], "events": [ "PortfolioDriftDetected", "portfolio_review.assembled" ], "repository": "PortfolioReviewRepository (per-fiduciary partitioned)", "transaction_boundary": "One PortfolioReview per fiduciary per quarter" }, { "name": "RuleMatrixVersionAggregate", "root": "RuleMatrixVersion", "context": "Rule Matrix", "purpose": "Represents one versioned snapshot of the federal/state/ABLE rule set with citation anchors.", "entities": [ "RuleEntry" ], "value_objects": [ "RuleVersionTag", "CitationAnchor" ], "invariants": [ "Every rule entry carries a source citation and a last-confirmed date" ], "commands": [ "UpdateRuleMatrix" ], "events": [ "RuleMatrixVersionPublished" ], "repository": "RuleMatrixRepository (shared, read by all fiduciaries)", "transaction_boundary": "One rule version per commit" } ], "ai_agents": [ { "name": "Extraction Agent", "context": "Intake", "responsibility": "Parses disbursement-request submissions of any format into the canonical schema.", "inputs": [ "Disbursement request description, amount, payee", "Supporting receipts/invoices" ], "outputs": [ "Normalized canonical records with a per-field confidence score" ], "tools": [ "Python parsers", "Frontier LLM API for unstructured-format extraction" ], "forbidden_actions": [ "Never stores or requests a beneficiary/family credential", "Never computes a final dollar figure used in a memo" ], "memory_scope": "Per-fiduciary, per-request; no cross-fiduciary memory", "retrieval_sources": [ "Rule Matrix field-mapping conventions" ], "validations": [ "Completeness checklist per request type" ], "confidence_scoring": "Per-field confidence score; below-threshold fields routed to analyst confirmation", "escalation_triggers": [ "Confidence below 70% on a required field" ], "human_approval": "Not required for normalization itself; required before any downstream delivery", "failure_modes": [ "Silent mis-extraction of a malformed submission", "OCR misread on a poor-quality scan" ], "audit_logs": [ "Intake timestamp + submitted-fact manifest + extraction confidence scores" ], "metrics": [ "Extraction accuracy vs. gold-standard sample", "Time to normalized output" ], "versioning": "Prompt + parser version pinned per Rule Matrix release" }, { "name": "Classification Agent", "context": "Determination", "responsibility": "Classifies the normalized request against the current Rule Matrix and computes a Benefit-Impact Projection.", "inputs": [ "Normalized disbursement-request record", "Rule Matrix" ], "outputs": [ "Classification with cited rationale and confidence score", "Benefit-Impact Projection" ], "tools": [ "Frontier LLM API", "Deterministic Bright-Line Check engine" ], "forbidden_actions": [ "Never overrides a Bright-Line Check result", "Never computes a dollar figure outside deterministic code" ], "memory_scope": "Per-fiduciary, per-request; prior determination history retained for pattern learning", "retrieval_sources": [ "Rule Matrix", "Prior determination history for the same fiduciary" ], "validations": [ "ConfidenceScore >= threshold or analyst sign-off" ], "confidence_scoring": "Fact-pattern-match confidence band against the adjudicated-example library", "escalation_triggers": [ "Confidence below 70%", "Fact pattern not represented in the adjudicated-example library" ], "human_approval": "Required below the confidence threshold and on every delivery", "failure_modes": [ "Misclassification against a stale Rule Matrix version", "Missed jurisdiction-specific nuance" ], "audit_logs": [ "Classification rationale + cited rule + confidence score per request" ], "metrics": [ "Auto-classification accuracy vs. gold-standard sample", "Escalation rate per 100 requests" ], "versioning": "Prompt + rules version pinned per Rule Matrix release" }, { "name": "Drafting Agent", "context": "Determination", "responsibility": "Assembles the Determination Memo narrative citing the classification result and, where applicable, a Restructuring Alternative.", "inputs": [ "Classification result with citations", "Rule Matrix citation set" ], "outputs": [ "Draft Determination Memo with cited evidence" ], "tools": [ "Frontier LLM API", "Citation-linked retrieval over the Rule Matrix" ], "forbidden_actions": [ "Never asserts a dollar amount not computed deterministically", "Never implies certainty about a future SSA/Medicaid finding" ], "memory_scope": "Per-fiduciary, per-determination", "retrieval_sources": [ "Rule Matrix citation anchors" ], "validations": [ "CitationRequiredSpec — every claim in the memo must cite a source" ], "confidence_scoring": "Memo flagged for analyst rewrite if a required citation is missing", "escalation_triggers": [ "Missing citation", "Dollar amount mismatch vs. deterministic computation" ], "human_approval": "Required before any memo reaches a fiduciary", "failure_modes": [ "Unsupported or overstated claim language" ], "audit_logs": [ "Draft memo + citations + analyst signature retained" ], "metrics": [ "Memo rejection rate", "Time from classification to drafted memo" ], "versioning": "Prompt version pinned per Rule Matrix release" }, { "name": "Escalation-Triage Agent", "context": "Review & Escalation", "responsibility": "Applies the escalation criteria in order and routes a determination to the attorney bench when triggered.", "inputs": [ "Drafted determination", "Confidence score", "Escalation criteria thresholds" ], "outputs": [ "Escalation decision with a logged reason" ], "tools": [ "Deterministic escalation-rule engine" ], "forbidden_actions": [ "Never suppresses a later-matching escalation reason once one is found", "Never de-escalates a case an analyst has flagged" ], "memory_scope": "Per-fiduciary, per-determination", "retrieval_sources": [ "Escalation criteria configuration" ], "validations": [ "EscalationCriteriaSpec" ], "confidence_scoring": "N/A — deterministic rule application", "escalation_triggers": [ "Certified Opinion requested", "Confidence below threshold", "High-value threshold exceeded", "Novel fact pattern", "Governing-document change implicated" ], "human_approval": "Escalation itself is automatic; the resulting attorney review is always human", "failure_modes": [ "Escalation-rule misconfiguration causing under- or over-escalation" ], "audit_logs": [ "Escalation reason + timestamp logged per case" ], "metrics": [ "Escalation rate", "Escalation-reason distribution" ], "versioning": "Escalation criteria versioned per Rule Matrix release" }, { "name": "Portfolio Drift-Detection Agent", "context": "Portfolio Compliance", "responsibility": "Scans a fiduciary's full determination history for the Quarterly Portfolio Review and flags aggregate patterns that look risky only in combination.", "inputs": [ "Full-portfolio determination history" ], "outputs": [ "Draft drift findings with a completeness score" ], "tools": [ "Deterministic pattern-detection rules", "Frontier LLM API for narrative summaries" ], "forbidden_actions": [ "Never issues a final portfolio verdict without analyst review", "Never omits a required review section without a logged, named gap" ], "memory_scope": "Per-fiduciary, per-review-cycle", "retrieval_sources": [ "Full determination history for the fiduciary" ], "validations": [ "PortfolioReviewCompletenessSpec" ], "confidence_scoring": "Drift score per detected pattern", "escalation_triggers": [ "Any high-severity drift pattern detected" ], "human_approval": "Required — final review and spot-check before delivery", "failure_modes": [ "A drift pattern missed if review scope excludes older determinations" ], "audit_logs": [ "Review manifest signed by the Benefits-Compliance Analyst" ], "metrics": [ "Review completeness score", "Review-cycle turnaround time" ], "versioning": "Review template versioned per program-year" } ], "prompt_chain_map": [ "Extraction Agent (parse) -> Classification Agent (classify + score, after deterministic Bright-Line Checks) -> Drafting Agent (memo, only after classification) -> Benefits-Compliance Analyst gate -> Escalation-Triage Agent (parallel, routes if triggered)", "Full determination history -> Portfolio Drift-Detection Agent (draft findings) -> Benefits-Compliance Analyst final review -> delivered Quarterly Portfolio Review" ], "rag_map": [ "Rule Matrix (versioned federal POMS + state Medicaid-supplement + ABLE QDE guidance with citation anchors) retrieved by the Classification Agent and the Drafting Agent for every determination", "Prior determination history retrieved by the Portfolio Drift-Detection Agent, restricted to the requesting fiduciary's own portfolio only" ], "ai_evaluation": [ "Gold-standard adjudicated determination examples used for regression testing on every prompt or Rule Matrix version change", "Monthly red-team run: seeded error requests (ambiguous ISM cases, borderline resource-limit cases, mixed SNT/ABLE fact patterns) verify the engine catches known failure patterns", "5-10% sampled re-verification of standard-tier determinations against a manually re-checked baseline" ], "hallucination_controls": [ "Retrieval-first classification and drafting — every claim cites a specific POMS/state-rule/ABLE QDE citation", "Deterministic Bright-Line Checks and all dollar arithmetic in code, never the model", "Confidence-scored classification with mandatory analyst confirmation below threshold", "CitationRequiredSpec rejects any memo missing a required citation", "Deterministic disclaimer injection prevents any memo from implying certainty about a future SSA/Medicaid finding" ], "human_in_the_loop_plan": [ "Classification gate: no determination below the confidence threshold proceeds without analyst confirmation", "Delivery gate: every Determination Memo is human-approved before it reaches a fiduciary", "Certification gate: every Certified Opinion and every escalated determination is attorney-signed", "Dual review on any Certified Opinion above the defined high-value threshold" ], "ai_audit_plan": [ "Every AI-drafted determination and every AI-proposed classification is logged with its confidence score and cited source before analyst review", "Quarterly gold-standard regression test against the current Rule Matrix version", "Annual external review of the hallucination-control suite ahead of each state's legislative session" ], "prompt_versioning": "Prompts + schemas + the Rule Matrix are model-agnostic and version-pinned together; a frontier-model swap is a config change, not a rebuild (see ai-engine-spec.md Model Portability).", "human_roles": [ { "role": "Benefits-Compliance Analyst", "responsibilities": [ "Review every AI-drafted determination before delivery", "Approve every escalation decision", "Sign off on every Quarterly Portfolio Review" ], "contexts": [ "Determination", "Review & Escalation", "Portfolio Compliance" ], "decisions_owned": [ "Determination confirmation or edit", "Escalation triage below automatic thresholds", "Portfolio drift-finding release" ], "ai_support": [ "[PLACEHOLDER] owner to complete" ], "approval_authority": "Sole approver for all standard-tier determinations; escalated/certified cases require an additional attorney signature", "escalation_authority": "Can escalate any case to the attorney bench; cannot give legal advice", "quality_metrics": [ "Review turnaround time", "Rework rate on analyst-approved determinations (<5% target)" ], "workload_risks": [ "Portfolio-review volume spikes across many fiduciaries at once" ] }, { "role": "CELA / Elder-Law Attorney (contracted bench)", "responsibilities": [ "Review and sign escalated determinations", "Review and sign every Certified Opinion", "Quarterly attorney-reviewed Rule Matrix red-team audit" ], "contexts": [ "Review & Escalation", "Rule Matrix" ], "decisions_owned": [ "Final legal determination on novel/high-risk fact patterns", "Rule Matrix entry approval for a new jurisdiction" ], "ai_support": "AI-prepared draft memo and citation trail shortens attorney review time; the sign-off itself never automates", "approval_authority": "Required sign-off on every escalated and certified determination", "escalation_authority": "Final authority on any litigation-adjacent boundary question", "quality_metrics": [ "Attorney turnaround time", "Zero unauthorized-practice-of-law incidents" ], "workload_risks": [ "Attorney-bench bottleneck as escalated/certified volume grows" ] }, { "role": "Founder", "responsibilities": [ "Own program-expertise content and sales", "Certify the first Benefits-Compliance Analyst hires", "Hold the 5/10/20-pilot build-before-scale checkpoints" ], "contexts": [ "Rule Matrix", "Referral & Outreach" ], "decisions_owned": [ "Pilot-cohort admission (5-8-account cap)", "State-expansion sequencing" ], "ai_support": "AI drafts content and outreach copy for founder review", "approval_authority": "Final sign-off on pilot admission and state-expansion timing", "escalation_authority": "Escalates key-person risk items to Counsel", "quality_metrics": [ "Pilot-to-retainer conversion", "First-renewal retention" ], "workload_risks": [ "Key-person dependency on founder for sales and program expertise" ] } ], "human_review_checkpoints": [ "Every determination below the confidence threshold", "Every Determination Memo before delivery (100%, no exceptions)", "Every Certified Opinion and every escalated determination (attorney sign-off)", "Final Quarterly Portfolio Review before delivery to a retainer fiduciary" ], "escalation_matrix": [ "Confidence below threshold, high-value amount, novel fact pattern, or Certified Opinion requested -> automatic routing to the attorney bench", "A dispute or fact pattern trending toward litigation -> documented Determination Memo handed to the fiduciary's own counsel; BenefitGuard steps back from legal advocacy" ], "manual_override_rules": [ "A Benefits-Compliance Analyst may manually reclassify a request's purpose category when new facts change the correct Rule Matrix citation", "A Benefits-Compliance Analyst may manually route a below-threshold case for a second analyst opinion before attorney escalation" ], "separation_of_duties": [ "The analyst who drafts an escalation reason is never the same person who signs the resulting Certified Opinion (that is always the attorney bench)" ], "quality_control_workflow": [ "Classification gate -> delivery gate -> certification gate (the three permanent human-in-the-loop gates)", "5-10% random re-verification of standard-tier determinations", "Monthly seeded-error red-team runs against known failure patterns" ], "data_objects": [ { "name": "DisbursementRequest", "meaning": "One proposed trust or ABLE expenditure submitted for review.", "owner_context": "Intake", "writers": [ "Extraction Agent", "Professional Fiduciary" ], "readers": [ "Determination", "Review & Escalation", "Portfolio Compliance" ], "source_of_truth": "Intake context", "retention": "Engagement-letter-defined retention period", "privacy": "Sensitive; beneficiary benefit-status heightened", "audit": "Every state transition timestamped in the append-only audit log" }, { "name": "DeterminationMemo", "meaning": "One dated, cited compliance verdict for a Disbursement Request.", "owner_context": "Determination", "writers": [ "Classification Agent", "Drafting Agent", "Benefits-Compliance Analyst" ], "readers": [ "Review & Escalation", "Portfolio Compliance", "Fiduciary (via delivery)" ], "source_of_truth": "Determination context", "retention": "Engagement-letter-defined retention period", "privacy": "Sensitive financial + benefit-status data", "audit": "Hash-chained snapshot per delivered memo" }, { "name": "RuleMatrixEntry", "meaning": "One versioned rule (POMS provision, state Medicaid-supplement rule, or ABLE QDE guidance) with a citation anchor.", "owner_context": "Rule Matrix", "writers": [ "CELA / Elder-Law Attorney" ], "readers": [ "Determination", "Portfolio Compliance" ], "source_of_truth": "Rule Matrix context", "retention": "Indefinite, versioned", "privacy": "Not sensitive — regulatory reference data", "audit": "Change log with source citation for every update" } ], "read_models": [ "Exception Queue view (per analyst, severity-tiered)", "Attorney-bench queue (per attorney, escalation-reason tagged)", "Portfolio dashboard (per fiduciary, all active retainer determinations)", "Determinations-delivered / SLA-hit scorecard (per fiduciary and aggregate)" ], "reporting_models": [ "Quarterly Portfolio Review summary", "Annual Rule Matrix currency report" ], "data_duplication_notes": [ "Request data is normalized once at Intake and never re-parsed downstream — Determination, Review & Escalation, and Portfolio Compliance all read the same canonical record", "Rule Matrix citations are written once and referenced (not copied) by Determination and Portfolio Compliance" ], "data_retention": [ "Beneficiary benefit-status and disbursement data retained for the engagement-letter-defined retention period", "Audit Trail retained for the same period, extended under an active legal hold", "Data deleted per the offboarding SOP on engagement termination unless a legal hold applies" ], "data_quality_risks": [ "Inconsistent submission formats across different fiduciary case-management tools", "Beneficiary benefit-status data going stale between the submission date and the review date", "A trust governing document on file becoming outdated after an amendment the fiduciary didn't flag" ], "use_cases": [ { "name": "Standard determination cycle", "actor": "Benefits-Compliance Analyst", "context": "Intake / Determination / Review & Escalation", "goal": "A complete Disbursement Request reviewed and delivered within the 1-business-day SLA", "preconditions": [ "Fiduciary onboarded with a signed engagement letter", "Rule Matrix covers the beneficiary's state" ], "main_flow": [ "Fiduciary submits the request", "Bright-Line Checks and classification run", "Analyst reviews and approves the draft", "Determination Memo delivered" ], "alternative_flows": [ "A below-threshold classification routes to the Exception Queue before delivery" ], "business_rules": [ "No memo is delivered without an analyst-approved sign-off" ], "ai_role": "Drafts the classification, Benefit-Impact Projection, and memo narrative", "human_role": "Analyst reviews the draft and approves delivery", "commands": [ "ClassifyAndDraftDetermination", "DeliverDeterminationMemo" ], "events": [ "determination.drafted" ], "aggregates": [ "DeterminationMemoAggregate" ], "success": "Memo delivered on time with a fully cited verdict", "failure_handling": "A missed SLA triggers a root-cause review of the review-queue capacity", "audit": "Full cycle logged in the append-only audit trail" }, { "name": "Escalated / Certified Opinion cycle", "actor": "CELA / Elder-Law Attorney", "context": "Review & Escalation", "goal": "A novel, high-value, or Certified-Opinion-requested determination reviewed and signed by a licensed attorney", "preconditions": [ "Escalation criteria triggered or Certified Opinion requested" ], "main_flow": [ "Escalation-Triage Agent routes the case with a logged reason", "Attorney reviews the AI-prepared draft and citation trail", "Attorney signs or requests more facts", "Determination Memo delivered with the attorney's signature" ], "alternative_flows": [ "Attorney requests additional facts before signing, extending the SLA with fiduciary notice" ], "business_rules": [ "No Certified Opinion is delivered without an attorney signature" ], "ai_role": "Prepares the draft memo and citation trail to shorten attorney review time", "human_role": "Attorney makes the final legal determination and signs", "commands": [ "EscalateToAttorneyBench", "SignDetermination" ], "events": [ "RequestEscalated", "DeterminationSigned" ], "aggregates": [ "EscalationCaseAggregate" ], "success": "Certified Opinion delivered with a complete signature manifest", "failure_handling": "An attorney-bench bottleneck triggers a review of bench capacity and turnaround-time KPIs", "audit": "Attorney name, bar number, and signed memo retained" } ], "invariants": [ { "invariant": "A Determination Memo cannot be delivered without at least one Benefits-Compliance Analyst sign-off", "context": "Review & Escalation", "aggregate": "EscalationCaseAggregate", "why": "Prevents an unreviewed or wrong determination from reaching a fiduciary and affecting a real disbursement decision", "enforcement": "Application-service-level guard; no delivery command executes without a recorded analyst approval event" }, { "invariant": "A Certified Opinion cannot be delivered without a licensed attorney's signature", "context": "Review & Escalation", "aggregate": "EscalationCaseAggregate", "why": "The Certified Opinion tier is a genuine licensed-attorney work product by definition", "enforcement": "Delivery command requires a recorded attorney signature event for any case flagged certified" }, { "invariant": "A Bright-Line Check result is never overridden by AI classification", "context": "Determination", "aggregate": "DeterminationMemoAggregate", "why": "The highest-consequence thresholds (resource limit, ISM formula, ABLE cap) must be deterministic and auditable, never a model inference", "enforcement": "Deterministic code runs before and independently of the classification model; the model cannot alter a Bright-Line Check output" }, { "invariant": "No beneficiary's or family member's SSA `my Social Security` or ABLE program portal credential is ever collected, stored, or used by BenefitGuard", "context": "Intake", "aggregate": "DisbursementRequestAggregate", "why": "BenefitGuard is never the beneficiary's representative and must not create an unauthorized-access or credential-custody risk", "enforcement": "Schema-level rejection of any credential-shaped field; code-level guard blocks the workflow" }, { "invariant": "A DisbursementRequest cannot enter the review queue until intake completeness is satisfied or a named exception is logged", "context": "Intake", "aggregate": "DisbursementRequestAggregate", "why": "Prevents wasted analyst time on requests missing the facts needed for a defensible determination", "enforcement": "State-machine guard on the DisbursementRequest lifecycle" } ], "architecture": { "style": "Modular monolith, single production region", "why": "A 5-8-account pilot cohort does not justify microservices; module boundaries mirror the bounded contexts so a future split is cheap.", "rejected_alternatives": [ "Microservices from day one (premature operational overhead for a pilot cohort)", "No-code workflow tool (can't enforce the deterministic Bright-Line Check and citation invariants)" ], "backend_modules": [ "intake", "determination", "review-escalation", "portfolio-compliance", "rule-matrix" ], "frontend_modules": [ "analyst exception-queue console", "internal reporting console (no fiduciary-facing portal at launch)" ], "api_boundaries": [ "Internal-only APIs between backend modules", "No public API surface at launch (email/upload-link intake and email/document delivery only)" ], "database_strategy": "Single Postgres instance, per-fiduciary row-level partitioning; no per-fiduciary dedicated infrastructure at pilot scale.", "event_bus": "In-process domain event dispatch at pilot scale; a lightweight external event bus is deferred until multi-analyst concurrency requires it.", "queue": "Background job queue for extraction and classification batches; per-tenant concurrency caps.", "workflow_engine": "Deterministic rules engine (code) for Bright-Line Checks and completeness gates; no general-purpose workflow engine needed at this scale.", "ai_orchestration": "Frontier LLM API called from typed service functions per agent (Extraction, Classification, Drafting, Escalation-Triage, Portfolio Drift-Detection); no autonomous multi-step agent loops.", "rag_layer": "Retrieval over the versioned Rule Matrix, citation-linked, not free-form.", "file_storage": "Encrypted object storage, per-fiduciary access-controlled buckets, for disbursement-request documents and Determination Memos.", "authn_authz": "SSO with MFA for internal staff; role-based capability checks enforced server-side (Benefits-Compliance Analyst, Attorney bench, Founder).", "admin_dashboard": "Internal-only; no fiduciary-facing admin surface at launch.", "client_portal": "None at launch — the fiduciary never operates software; deliverables are documents by email.", "operator_dashboard": "Benefits-Compliance Analyst exception-queue console with severity tiers and SLA timers.", "observability": "Structured logs, RED metrics per workflow, model-call cost/latency spans.", "audit_logging": "Append-only, hash-chained snapshots per Determination Memo; every request state observed is timestamped.", "deployment": "Single-region, multi-AZ production; preview environment per PR." }, "module_structure": { "tree": "src/{intake,determination,review-escalation,portfolio-compliance,rule-matrix}/{domain,application,infra}", "modules": [ { "name": "intake", "purpose": "Disbursement request normalization and completeness checking", "owned_domain": [ "DisbursementRequest" ], "application_services": [ "SubmitDisbursementRequest", "NormalizeRequestData" ], "infra_adapters": [ "Email/upload intake handler", "Object storage adapter", "LLM extraction client" ], "public_interfaces": [ "RequestDataNormalized event" ], "forbidden_deps": [ "Must not depend on review-escalation or portfolio-compliance internals" ] }, { "name": "determination", "purpose": "Bright-Line Checks, classification, and memo drafting", "owned_domain": [ "DeterminationMemo (draft)" ], "application_services": [ "RunBrightLineChecks", "ClassifyAndDraftDetermination" ], "infra_adapters": [ "LLM classification client", "Rule Matrix reader" ], "public_interfaces": [ "DeterminationDrafted event" ], "forbidden_deps": [ "Must not write to intake's tables" ] }, { "name": "review-escalation", "purpose": "Analyst review, attorney escalation, and delivery", "owned_domain": [ "EscalationCase" ], "application_services": [ "EscalateToAttorneyBench", "SignDetermination", "DeliverDeterminationMemo" ], "infra_adapters": [ "Email send adapter", "E-signature adapter" ], "public_interfaces": [ "DeterminationDelivered event" ], "forbidden_deps": [ "Must not perform classification logic" ] }, { "name": "portfolio-compliance", "purpose": "Retainer tracking, quarterly review assembly, drift detection", "owned_domain": [ "PortfolioReview" ], "application_services": [ "RunQuarterlyPortfolioReview" ], "infra_adapters": [ "Report template renderer", "Object storage adapter" ], "public_interfaces": [ "PortfolioDriftDetected event" ], "forbidden_deps": [ "Must not draft individual determinations directly" ] }, { "name": "rule-matrix", "purpose": "Versioned per-jurisdiction rule sets with citation anchors", "owned_domain": [ "RuleMatrixEntry" ], "application_services": [ "UpdateRuleMatrix" ], "infra_adapters": [ "Regulatory-change diff detector" ], "public_interfaces": [ "RuleMatrixVersionPublished event" ], "forbidden_deps": [ "Must not depend on any fiduciary-specific data" ] } ], "dependency_rules": [ "Downstream contexts consume upstream domain events; no direct cross-module table access", "rule-matrix has no dependency on any other module" ] }, "security_governance": { "controls": [ { "risk": "A wrong or unreviewed determination reaches a fiduciary", "context": "Determination", "impact": "high", "control": "Confidence-scored classification + mandatory analyst confirmation below threshold + 5-10% sampled re-verification", "audit": "Classification rationale + confidence score logged per request" }, { "risk": "Cross-fiduciary data leak, including beneficiary benefit-status records", "context": "Intake / Determination", "impact": "severe", "control": "Row-level auth + per-fiduciary retrieval partitioning", "audit": "Access logs reviewed weekly" }, { "risk": "A beneficiary/family credential is collected in violation of the documented prohibition", "context": "Intake", "impact": "severe", "control": "Schema-level rejection of any credential-shaped field; code-level guard", "audit": "Automated schema-validation test in CI" } ], "ai_governance": [ "Every AI-drafted output is confidence-scored and cited", "No dollar figure is ever LLM-computed", "Analyst approval required before any delivery" ], "prompt_injection_defense": [ "Uploaded disbursement-request/trust-document content treated as untrusted data, never as instructions", "Output schema validation on every agent response before it reaches a human queue" ], "sensitive_data_handling": [ "Beneficiary benefit-status and financial data classified sensitive with least-privilege access", "Per-fiduciary segregation with no shared indices across fiduciaries", "Deletion SLA enforced on offboarding" ], "access_control_matrix": [ { "role": "Benefits-Compliance Analyst", "context": "Determination / Review & Escalation / Portfolio Compliance", "capabilities": [ "Approve classifications below threshold", "Release Determination Memos", "Sign off on Quarterly Portfolio Reviews" ] }, { "role": "CELA / Elder-Law Attorney", "context": "Review & Escalation / Rule Matrix", "capabilities": [ "Sign escalated and Certified Opinion determinations", "Approve new Rule Matrix jurisdiction entries" ] }, { "role": "Founder", "context": "Rule Matrix / Referral & Outreach", "capabilities": [ "Admit pilot fiduciaries", "Approve state-expansion sequencing" ] } ], "audit_log_requirements": [ "Every request state observed, every determination drafted, every rule applied is timestamped per request", "Hash-chained snapshots for every delivered Determination Memo" ] }, "observability": { "metrics": [ { "metric": "Auto-classification accuracy", "type": "gauge", "context": "Determination", "why": "Tracks how much of the classification workload runs correctly without analyst correction", "target": "improving toward 85%+ by day 90", "alert_threshold": "<50% sustained for 2 weeks" }, { "metric": "Benefit-suspension incidents traced to a delivered determination", "type": "counter", "context": "Review & Escalation", "why": "The zero-tolerance quality target", "target": "0", "alert_threshold": ">0 triggers immediate root-cause review" }, { "metric": "Determination cycle time", "type": "histogram", "context": "Review & Escalation", "why": "SLA is 1 business day standard / 4 hours rush", "target": "<=1 business day standard", "alert_threshold": ">1.5 business days" } ], "dashboards": [ "Analyst exception-queue dashboard (severity-tiered, SLA timers)", "Aggregate determinations-delivered / SLA-hit scorecard" ], "audit_reports": [ "Monthly audit-trail completeness report per fiduciary" ], "quality_review_reports": [ "Monthly seeded-error red-team result summary" ], "ai_evaluation_reports": [ "Quarterly gold-standard regression report" ], "client_outcome_reports": [ "Determination Memo per request", "Quarterly Portfolio Review per retainer fiduciary" ] }, "testing_strategy": { "tests": [ { "type": "unit", "validates": "Bright-Line Check math (resource limit, ISM formula, ABLE cap)", "context": "Determination", "example": "A $500 direct rent payment computes the correct ISM presumed-maximum-value reduction" }, { "type": "component", "validates": "Exception-queue routing on below-threshold classifications", "context": "Determination", "example": "A 60%-confidence classification routes to attorney escalation, never auto-delivers" }, { "type": "e2e", "validates": "Full intake -> classification -> analyst sign-off -> delivery path", "context": "cross-context", "example": "A seeded sample request produces a correctly delivered Determination Memo end-to-end" } ], "critical_domain_rules": [ "SSI resource-limit Bright-Line Check", "ISM presumed-maximum-value formula", "Low-confidence attorney-escalation rule" ], "ai_eval_dataset": [ "Gold-standard adjudicated determination examples", "Seeded error requests for red-team runs (ambiguous ISM cases, mixed SNT/ABLE fact patterns)" ], "regression_plan": "Every Rule Matrix version bump re-runs the full gold-standard regression suite before the new rule set goes live for any fiduciary.", "contract_testing_plan": "Contract tests validate the canonical disbursement-request schema against every supported submission format sampled from the pilot cohort.", "manual_qa_checklist": [ "Analyst spot-checks 5-10% of standard-tier determinations each cycle", "Monthly red-team seeded-error run reviewed by the founder", "Quarterly review reviewed against the completeness checklist before delivery" ] }, "mvp_roadmap": [ { "phase": "MVP — Disbursement Determination Cycle", "goal": "Prove the classify-review-deliver loop manually-assisted for the first 3-5 pilot fiduciaries", "features": [ "Disbursement Compliance Check intake + readout", "Standard determination delivery", "Exception Queue escalation" ], "contexts": [ "Intake", "Determination", "Review & Escalation" ], "ai_needs": [ "Extraction Agent", "Classification Agent", "Drafting Agent" ], "human_workflows": [ "Founder reviews every exception and every delivery personally for the first 5 pilots" ], "data_needs": [ "Federal POMS + founder's home-state Rule Matrix v0" ], "integrations": [ "Email/upload intake, secure email delivery" ], "risks": [ "Manual founder bottleneck at volume" ], "exit_criteria": [ "5 pilot fiduciaries with at least one completed determination each, zero SLA misses" ] } ], "scaling_roadmap": [ { "stage": "3-5 -> 5-8 pilot accounts", "trigger": "5-pilot hardening checkpoint passed (intake spec, evidence requirements, QA checklists standardized)", "architecture_change": "Introduce AI drafting layer against a growing adjudicated-example library; certify a second Benefits-Compliance Analyst", "operational_change": "Exception-queue tiers and escalation-criteria checklists formalized", "risk": "Per-fiduciary snowflake logic that can't be ruled is a stop-the-line signal" }, { "stage": "8 -> 10+ accounts (multi-state wave)", "trigger": "Gross margin trending toward the 50-55% target and zero benefit-suspension incidents in the trailing quarter", "architecture_change": "Additional state Rule Matrix entries codified and attorney-reviewed", "operational_change": "Pause new onboarding at the 20-pilot checkpoint until cost-per-determination, rework rate, escalation rate, and cycle time are measured", "risk": "New-state Medicaid-supplement rule volatility during a program's first year of coverage" } ], "risk_register": [ { "risk": "A state Medicaid-supplement rule change breaks a codified determination path mid-year", "likelihood": "high", "impact": "medium", "signal": "A tracked regulatory source change is detected outside a scheduled review window", "mitigation": "Rule Matrix versioned per jurisdiction with change-monitoring", "owner": "product", "context": "Rule Matrix" }, { "risk": "Attorney-bench bottleneck as escalated/certified volume grows", "likelihood": "medium", "impact": "medium", "signal": "Escalated-case turnaround time exceeds the KPI target", "mitigation": "Build a bench of 2-3 contracted attorneys across relevant jurisdictions before scaling past the pilot cohort", "owner": "owner", "context": "Review & Escalation" }, { "risk": "Fuzzy classification error routes a request to the wrong verdict", "likelihood": "low", "impact": "high", "signal": "Sampled re-verification finds a mismatch", "mitigation": "Confidence thresholds + mandatory analyst confirmation below threshold + deterministic Bright-Line Checks", "owner": "engineering", "context": "Determination" } ], "adrs": [ { "id": "ADR-001", "decision": "Launch as a single-region modular monolith, not microservices", "status": "accepted", "context": "A 5-8-account pilot cap does not justify microservice operational overhead", "options": [ "Modular monolith", "Microservices", "No-code workflow tool" ], "chosen": "Modular monolith", "business_reason": "Fastest path to the first 3-5 pilot determinations without a platform team", "technical_reason": "Module boundaries mirror bounded contexts, keeping a future split cheap", "tradeoffs": [ "[PLACEHOLDER] owner to complete" ], "risks": [ "[PLACEHOLDER] owner to complete" ], "revisit_trigger": "Sustained load beyond the analyst-throughput target", "why": "Matches the actual pilot scale", "consequences": "Single deploy unit, simpler ops, requires discipline on module boundaries", "reversal": "Extract a module to a service if its load profile diverges sharply", "revisit_when": "Post-20-pilot pause-and-measure checkpoint" }, { "id": "ADR-002", "decision": "No fiduciary-facing portal at launch; delivery is email/document only", "status": "accepted", "context": "The blueprint's sales thesis is that the fiduciary never operates software", "options": [ "Build a fiduciary-facing dashboard", "Email/document delivery only", "White-labeled third-party portal" ], "chosen": "Email/document delivery only", "business_reason": "Matches the 'we sell a determination, not a dashboard' positioning", "technical_reason": "Removes an entire authn/authz surface at pilot scale", "tradeoffs": "No self-serve visibility for the fiduciary between determinations", "risks": "A fiduciary used to a dashboard may perceive the service as less real-time", "revisit_trigger": "Multiple pilot fiduciaries request live status visibility", "why": "Keeps the MVP thin and evidence-linked", "consequences": "Delivered memos and readout calls carry the full trust burden", "reversal": "Add a lightweight read-only status view if demand is proven", "revisit_when": "After the 10-pilot hardening checkpoint" } ], "self_audit": { "scores": [ { "category": "Domain clarity", "score": 5, "weakness": "None material", "improvement": "Keep the Rule Matrix citation discipline as new states are added" }, { "category": "Human-chokepoint discipline", "score": 5, "weakness": "None material", "improvement": "Continue dual review on high-value Certified Opinions as volume grows" }, { "category": "Regulatory-boundary clarity", "score": 4, "weakness": "Unauthorized-practice-of-law boundary requires counsel review before the first paid engagement", "improvement": "Complete counsel review before onboarding pilot 1" } ], "weakest_parts": [ "Certified Opinion pricing economics are Unverified until real attorney-time data exists", "Total national SNT count is Unverified — sized only directionally" ], "biggest_assumptions": [ "Free-diagnostic-to-paid conversion reaches a viable rate among warm professional fiduciaries", "A 1-business-day determination can run from fiduciary-supplied facts alone without a live SSA/Medicaid data integration" ], "highest_risk_decisions": [ "Offering the Certified Opinion at a flat $350 fee before real attorney-time data validates the economics", "Onboarding a fiduciary in a new state before that state's Medicaid-supplement rules are fully attorney-reviewed" ], "needs_domain_expert": [ "[PLACEHOLDER] owner to complete" ], "needs_legal": [ "[PLACEHOLDER] owner to complete" ], "needs_prototype": [ "[PLACEHOLDER] owner to complete" ], "validate_before_prod": [ "[PLACEHOLDER] owner to complete" ] }, "final_recommendations": [ "Launch the Disbursement Determination pilot immediately: codify federal POMS plus the founder's home state into the Rule Matrix, ship the free Disbursement Compliance Check, and recruit 5-8 founding fiduciary accounts", "Hold multi-state Rule Matrix expansion as the timed second act once the home-state rule library and pilot cap are proven", "Do not build a fiduciary-facing portal or software product beyond the internal engine until the 20-pilot checkpoint proves margin", "Keep the Benefits-Compliance Analyst approval gate permanent on every delivered determination regardless of automation maturity", "Treat the Certified Opinion tier as a licensed-attorney-gated revenue line with pricing validated against real attorney time, never assumed permanent at the initial anchor price" ], "extensions": { "service_business_reality_check": { "is_service_business": true, "paid_outcome_clear": true, "workflow_present": true, "ai_native_fit_score": 4.5, "red_flags": [] }, "ai_native_fit": { "score": 4.5, "why": "Mechanical, rule-driven workflow (request classification, Bright-Line Check application, memo drafting) ideally suited to AI with a clean, permanent human chokepoint on every delivered determination.", "disqualifiers": [] }, "domain_evidence_register": [ { "claim": "ABLE accounts totaled 213,855 nationally as of Q2 2025, up 18.3% YoY from 180,748 (Q2 2024)", "evidence_type": "industry association data", "source": "NAST Q2 2025 National ABLE Data, via PlanSponsor", "strength": "high", "gaps": "None — re-confirmed live during this build" }, { "claim": "SSI's $2,000 individual countable-resource limit is unchanged for decades", "evidence_type": "primary government source", "source": "SSA POMS SI 01120.200", "strength": "high", "gaps": "A pending bipartisan bill (H.R.2540/S.1234) proposes raising it — not yet law as of this build" } ], "assumption_register": [ { "assumption": "Professional fiduciaries will pay a per-request fee over continuing to decide on memory or a general chatbot", "impact_if_wrong": "severe", "how_to_validate": "First 5-8 pilot fiduciaries' free-diagnostic-to-paid conversion", "blocking": true }, { "assumption": "A 1-business-day determination can run from fiduciary-supplied facts alone without a live SSA/Medicaid data integration", "impact_if_wrong": "high", "how_to_validate": "First 10-15 completed determinations, manually-assisted", "blocking": false } ], "language_conflict_map": [ { "term": "account", "meaning_a": "A beneficiary's ABLE account (the state program's object)", "context_a": "ABLE program terminology", "meaning_b": "A BenefitGuard client relationship (a 'fiduciary account')", "context_b": "Internal sales/ops usage", "resolution": "Always qualify: 'ABLE account' for the program object; 'fiduciary engagement' for the client relationship — never bare 'account' in client-facing copy" } ], "build_buy_integrate": [ { "subdomain": "Determination", "decision": "build", "reason": "Core paid outcome; no vendor sells this per-transaction determination service" }, { "subdomain": "Referral & Outreach", "decision": "buy", "reason": "Lightweight CRM is not differentiating" } ], "core_protection_strategy": [ "[PLACEHOLDER] owner to complete" ], "boundary_stress_tests": [ { "scenario": "True Link Financial or a similar fintech ships a documented, per-transaction determination product", "contexts_touched": [ "Determination", "Review & Escalation" ], "breaks_if": "BenefitGuard were positioned as a card/tool rather than an operated, attorney-backed outcome", "verdict": "Holds — the fiduciary still needs a named analyst and licensed attorney standing behind the determination, not just a report screen" } ], "unresolved_ownership": [ { "concept": "Who owns a beneficiary-level 'benefit-status change' event that affects multiple pending requests at once", "candidates": [ "Determination (per-request)", "a new BeneficiaryStatusChange aggregate" ], "recommendation": "Model a BeneficiaryStatusChange as a future aggregate once pilot volume shows status-change events affecting multiple pending requests are common" } ], "published_language_contracts": [ { "producer": "Rule Matrix", "consumer": "Determination, Portfolio Compliance", "contract": "Versioned RuleMatrixEntry schema with citation anchors", "versioning": "Semantic version bump on every regulatory-source diff" } ], "shared_kernel_warnings": [ "Rule Matrix is a shared kernel across every fiduciary on a given jurisdiction — a bad rule-entry edit affects every client simultaneously; changes require the same review rigor as a production migration" ], "aggregate_stress_tests": [ { "aggregate": "DeterminationMemoAggregate", "scenario": "A beneficiary's benefit status changes mid-review (e.g., a Medicaid redetermination completes while a request is pending)", "invariant_at_risk": "The memo's Benefit-Impact Projection reflects the beneficiary's status at classification time, not at delivery time", "verdict": "Holds if the status-change date is captured and the memo is re-classified before delivery; flagged as a status-change exception for analyst review either way" } ], "agent_stress_tests": [ { "agent": "Classification Agent", "scenario": "A disbursement request plausibly qualifies under either the SNT ISM rules or the ABLE QDE rules depending on account type", "failure_mode": "Could propose a plausible but wrong rule-set classification if account type isn't explicit", "guardrail": "Account-type field is required at intake; ambiguous or missing account type forces analyst confirmation before classification", "verdict": "Holds with the guardrail; would fail without the required account-type field" } ], "regulated_domain_handling": [ { "regime": "Unauthorized practice of law", "applies_because": "Standard determinations are compliance research, but the boundary with individualized legal advice must be actively managed", "controls": [ "Engagement-letter scoping positioning standard determinations as research delivered to the fiduciary customer of record", "Escalation criteria route any borderline legal-judgment case to the attorney bench automatically" ], "evidence_required": [ "[PLACEHOLDER] owner to complete" ] }, { "regime": "Beneficiary data privacy (financial + benefit-status)", "applies_because": "BenefitGuard processes beneficiary financial and means-tested-benefit-status data under fiduciary authorization", "controls": [ "Signed engagement letter with confidentiality and deletion terms", "No beneficiary/family credential ever collected" ], "evidence_required": "Signed engagement letter per fiduciary before any request is worked" } ], "unit_economics": { "price_model": "Per-Disbursement-Request flat fee + Portfolio Monitoring Retainer subscription + Certified Opinion flat fee", "unit_of_value": "one delivered Determination Memo", "gross_margin_pct": 53, "cost_drivers": [ "Model inference (classification, drafting): $1-2/determination at day-90 maturity", "Benefits-Compliance Analyst review time (~10-20 min @ loaded rate): $8-10/determination", "QA sampling amortized + delivery tooling: $1.50-2.50/determination" ], "breakeven_note": "COGS ~$20-24/standard determination at day 90 vs. a $52.50 blended standard/rush price — roughly 53-55% gross margin target, consistent with the blueprint's explicitly stated 50-55% figure (see financial-model.csv header notes for the reconciliation against the blueprint's separately itemized, lower COGS line-items)." }, "margin_leakage_map": [ { "leakage": "Per-fiduciary snowflake case-management integration requests", "cause": "Pilot fiduciaries requesting one-off bespoke reporting outside the standard Determination Memo template", "impact": "Analyst minutes balloon past the day-90 target, eroding gross margin", "mitigation": "Product-vs-custom rule enforced from pilot 1; decline or separately price bespoke requests" } ], "slop_findings": [ { "pattern": "Generic 'AI-powered platform' framing", "status": "avoided", "note": "All copy anchors to specific mechanics (the $2,000 resource limit, the ISM formula, named POMS sections) rather than generic AI claims" } ], "drift_checks": [ { "stage": "DNA vs. landing page vs. playbook lexicon", "status": "checked", "findings": [ "[PLACEHOLDER] owner to complete" ] } ], "gates": [ { "id": "gate-1", "title": "Human-chokepoint completeness", "passed": true, "checks": [ { "name": "Every delivered determination has an analyst-approval invariant", "ok": true, "evidence": "ddd.invariants entry + Review & Escalation bounded-context invariants" }, { "name": "Every Certified Opinion has an attorney-sign-off gate", "ok": true, "evidence": "ddd.policies 'Certified Opinion attorney sign-off'" }, { "name": "No beneficiary/family credential handling anywhere in the workflow", "ok": true, "evidence": "ddd.invariants 'No beneficiary...credential' + security_governance controls" } ] } ], "contradiction_scan": [ { "id": "cs-1", "severity": "none", "message": "No contradiction found between the blueprint's licensing boundary and the DDD security_governance/compliance sections.", "refs": [ "product.security.compliance_targets", "ddd.security_governance.controls" ] } ], "rubric": { "categories": [ { "category": "Domain modeling completeness", "score": 5, "min": 3, "passed": true }, { "category": "Human-in-the-loop rigor", "score": 5, "min": 3, "passed": true }, { "category": "Regulatory-boundary clarity", "score": 4, "min": 3, "passed": true }, { "category": "Evidence traceability", "score": 4, "min": 3, "passed": true }, { "category": "Anti-genericness (lexicon specificity)", "score": 5, "min": 3, "passed": true } ], "pass": true, "average": 4.6 }, "foundry_package": { "version": "1.0", "checksum": "n/a-generated-locally", "counts": { "subdomains": 5, "bounded_contexts": 5, "aggregates": 5, "events": 6, "commands": 6, "policies": 4, "ai_agents": 5, "invariants": 5, "integrations": 4, "adrs": 2 }, "subset": { "subdomains": [ "Determination", "Review & Escalation", "Portfolio Compliance", "Rule Matrix", "Referral & Outreach" ], "bounded_contexts": [ "Intake", "Determination", "Review & Escalation", "Portfolio Compliance", "Rule Matrix" ], "aggregates": [ "DisbursementRequestAggregate", "DeterminationMemoAggregate", "EscalationCaseAggregate", "PortfolioReviewAggregate", "RuleMatrixVersionAggregate" ], "events": [ "DisbursementRequestSubmitted", "DeterminationDrafted", "RequestEscalated", "DeterminationSigned", "DeterminationDelivered", "RuleMatrixVersionPublished" ], "commands": [ "SubmitDisbursementRequest", "RunBrightLineChecks", "ClassifyAndDraftDetermination", "EscalateToAttorneyBench", "SignDetermination", "DeliverDeterminationMemo" ], "policies": [ "Bright-Line resource-limit gate", "Low-confidence escalation", "100%-analyst-review requirement", "Certified Opinion attorney sign-off" ], "ai_agents": [ "Extraction Agent", "Classification Agent", "Drafting Agent", "Escalation-Triage Agent", "Portfolio Drift-Detection Agent" ], "invariants": [ "Determination Memo analyst sign-off required", "Certified Opinion attorney signature required", "Bright-Line Check never overridden by AI", "No beneficiary/family credential handling", "Intake completeness required before review" ], "integrations": [ "Secure intake portal/email", "E-signature tool (attorney sign-off)", "Lightweight CRM", "OCR/document-extraction tool" ], "adrs": [ "ADR-001", "ADR-002" ] } } } }, "architecture": { "slug": "special-needs-trust-able-disbursement-compliance-desk", "archetypes": [ "compliance determination desk", "elder & disability-benefits advisory-adjacent service" ], "archetype_impact": "Drives a request-in / cited-memo-out shape with a permanent human-plus-attorney approval gate before any delivery, rather than a self-serve SaaS shape.", "personality": [ "Quiet, evidence-first, never alarmist", "Citation-literal (POMS sections and dollar amounts stated exactly, never vague)", "Fiduciary-branded internally — BenefitGuard is the fiduciary's own compliance desk, not a family-facing product", "Audit-grade precision in every citation" ], "forces_ranked": [ { "force": "Regulatory/evidence fidelity", "why": "Every citation and dollar figure must be deterministic and sourced — a wrong one costs a beneficiary's benefits and the fiduciary's liability position" }, { "force": "Human-plus-attorney chokepoint discipline", "why": "The business's entire moat rests on the analyst-approval and attorney-sign-off gates holding on every delivery" }, { "force": "Time-to-first-pilot", "why": "5-8-account pilot cap must be reachable in 90 days without a platform build" }, { "force": "Multi-jurisdiction extensibility", "why": "Federal POMS today, state-by-state Medicaid-supplement rules next — the Rule Matrix must generalize across jurisdictions cleanly" } ], "tradeoffs": [ "Single-region modular monolith trades some scalability headroom for speed-to-pilot and lower ops burden", "No fiduciary-facing portal at launch trades self-serve visibility for a thinner, faster-to-ship MVP" ], "quality_scenarios": [ { "attribute": "Correctness", "assumption": "A determination is drafted by the Classification Agent", "target": "0 benefit-suspension incidents traced to a delivered determination in the trailing quarter" }, { "attribute": "Availability", "assumption": "A synchronized portfolio-review volume spike across the full pilot cohort", "target": "Determination delivery on time (1-business-day SLA) for 100% of standard requests" } ], "options": [ { "style": "Modular monolith", "complexity": "low", "cost": "low", "ops_burden": "low", "team_fit": "high", "scaling_path": "Extract a module to a service if its load profile diverges", "security_impact": "Simpler perimeter, single auth boundary", "fits_here": true, "fits_when": "Pilot-to-early-scale (up to a few dozen fiduciaries per analyst)", "wrong_here": false, "recommended": true }, { "style": "Microservices", "complexity": "high", "cost": "high", "ops_burden": "high", "team_fit": "low at this stage", "scaling_path": "N/A yet", "security_impact": "More surface area, more inter-service auth", "fits_here": false, "fits_when": "Post-pilot multi-analyst scale", "wrong_here": true, "recommended": false }, { "style": "No-code workflow tool", "complexity": "low", "cost": "low", "ops_burden": "low", "team_fit": "medium", "scaling_path": "Would require a rebuild to add deterministic Bright-Line Checks and citation invariants", "security_impact": "Vendor-dependent data handling", "fits_here": false, "fits_when": "Never for this business's compliance-grade evidence requirements", "wrong_here": true, "recommended": false } ], "chosen_style": "Modular monolith, single production region", "chosen_rationale": "Matches the actual 5-8-account pilot scale and keeps the deterministic-rules and human-plus-attorney chokepoint invariants enforceable in one codebase.", "rejected": [ { "style": "Microservices", "why_rejected": "Operational overhead unjustified below a few dozen fiduciaries per analyst" }, { "style": "No-code workflow tool", "why_rejected": "Can't cleanly enforce Bright-Line Check determinism and citation-required invariants" } ], "target": { "overview": "Single-region modular monolith with five domain modules (intake, determination, review-escalation, portfolio-compliance, rule-matrix) behind an internal analyst console; no public API or fiduciary-facing portal at launch.", "frontend": "Internal analyst exception-queue and reporting console only", "backend": "Typed service layer per module, deterministic rules engine for Bright-Line Checks and completeness gates", "data": "Single Postgres instance, per-fiduciary row-level partitioning", "api": "Internal-only; no public API surface at launch", "authn_authz": "SSO + MFA for staff; server-side role capability checks", "integrations": "Email/upload intake, secure email delivery, encrypted object storage for evidence and memos", "background_jobs": "Extraction and classification batch jobs with per-tenant concurrency caps", "object_storage": "Encrypted, per-fiduciary access-controlled buckets", "notifications": "Analyst-approved email only; no SMS/push at launch", "search": "Citation-linked retrieval over the Rule Matrix, not general full-text search", "analytics": "Internal RED metrics + business-event dashboards; no third-party analytics on beneficiary data", "ai": "Frontier LLM API called per agent from typed service functions; retrieval-first, citation-linked", "observability": "Structured logs, RED metrics, model-call cost/latency spans", "deployment": "Single region + multi-AZ production; per-PR preview environments", "security": "Row-level auth, per-fiduciary retrieval partitioning, encrypted storage, least-privilege access", "dr": "Daily encrypted backups; documented restore runbook tested quarterly" }, "modules": [ { "name": "intake", "purpose": "Disbursement request normalization and completeness checking", "owned_domain": [ "DisbursementRequest" ], "application_services": [ "SubmitDisbursementRequest", "NormalizeRequestData" ], "infra_adapters": [ "Email/upload intake handler", "Object storage adapter", "LLM extraction client" ], "public_interfaces": [ "RequestDataNormalized event" ], "forbidden_deps": [ "Must not depend on review-escalation or portfolio-compliance internals" ], "responsibility": "Own disbursement-request intake and normalization", "owned_data": [ "DisbursementRequest" ], "events_produced": [ "RequestDataNormalized" ], "events_consumed": [], "interfaces": [ "SubmitDisbursementRequest" ], "depends_on": [], "entities": [ "DisbursementRequest" ], "failure_risks": [ "Malformed submission silently mis-extracted" ], "scaling": "Scales with intake batch volume; background-job concurrency capped per fiduciary", "future_split_trigger": "Extract to a service if extraction volume exceeds single-instance throughput" }, { "name": "determination", "purpose": "Bright-Line Checks, classification, and memo drafting", "owned_domain": [ "DeterminationMemo (draft)" ], "application_services": [ "RunBrightLineChecks", "ClassifyAndDraftDetermination" ], "infra_adapters": [ "LLM classification client", "Rule Matrix reader" ], "public_interfaces": [ "DeterminationDrafted event" ], "forbidden_deps": [ "Must not write to intake's tables" ], "responsibility": "Own Bright-Line Check execution and determination drafting", "owned_data": [ "DeterminationMemo" ], "events_produced": [ "DeterminationDrafted" ], "events_consumed": [ "RequestDataNormalized" ], "interfaces": [ "ClassifyAndDraftDetermination" ], "depends_on": [ "intake" ], "entities": [ "DeterminationMemo" ], "failure_risks": [ "Misclassification against a stale Rule Matrix version" ], "scaling": "Scales with determination-request volume", "future_split_trigger": "Extract to a service once multi-analyst concurrency requires independent scaling" }, { "name": "review-escalation", "purpose": "Analyst review, attorney escalation, and delivery", "owned_domain": [ "EscalationCase" ], "application_services": [ "EscalateToAttorneyBench", "SignDetermination", "DeliverDeterminationMemo" ], "infra_adapters": [ "Email send adapter", "E-signature adapter" ], "public_interfaces": [ "DeterminationDelivered event" ], "forbidden_deps": [ "Must not perform classification logic" ], "responsibility": "Own analyst review, attorney escalation, and delivery lifecycle", "owned_data": [ "EscalationCase" ], "events_produced": [ "DeterminationDelivered" ], "events_consumed": [ "DeterminationDrafted" ], "interfaces": [ "DeliverDeterminationMemo" ], "depends_on": [ "determination" ], "entities": [ "EscalationCase" ], "failure_risks": [ "Missed escalation on a genuinely novel case" ], "scaling": "Scales with active determination/escalation volume", "future_split_trigger": "Extract to a service if attorney-bench coordination volume grows past manual-assisted capacity" }, { "name": "portfolio-compliance", "purpose": "Retainer tracking, quarterly review assembly, drift detection", "owned_domain": [ "PortfolioReview" ], "application_services": [ "RunQuarterlyPortfolioReview" ], "infra_adapters": [ "Report template renderer", "Object storage adapter" ], "public_interfaces": [ "PortfolioDriftDetected event" ], "forbidden_deps": [ "Must not draft individual determinations directly" ], "responsibility": "Own quarterly review assembly and drift detection", "owned_data": [ "PortfolioReview" ], "events_produced": [ "PortfolioDriftDetected" ], "events_consumed": [ "DeterminationDelivered" ], "interfaces": [ "RunQuarterlyPortfolioReview" ], "depends_on": [ "review-escalation" ], "entities": [ "PortfolioReview" ], "failure_risks": [ "A drift pattern missed if review scope excludes older determinations" ], "scaling": "Scales with annual/quarterly review-cycle volume", "future_split_trigger": "Extract to a service only if review assembly becomes a shared multi-tenant bottleneck" }, { "name": "rule-matrix", "purpose": "Versioned per-jurisdiction rule sets with citation anchors", "owned_domain": [ "RuleMatrixEntry" ], "application_services": [ "UpdateRuleMatrix" ], "infra_adapters": [ "Regulatory-change diff detector" ], "public_interfaces": [ "RuleMatrixVersionPublished event" ], "forbidden_deps": [ "Must not depend on any fiduciary-specific data" ], "responsibility": "Own per-jurisdiction versioned rule sets with citation anchors", "owned_data": [ "RuleMatrixEntry" ], "events_produced": [ "RuleMatrixVersionPublished" ], "events_consumed": [], "interfaces": [ "UpdateRuleMatrix" ], "depends_on": [], "entities": [ "RuleMatrixEntry" ], "failure_risks": [ "Stale rule applied after a regulatory update" ], "scaling": "Scales with number of jurisdictions, not number of fiduciaries", "future_split_trigger": "Extract to a shared service once multiple product lines consume it" } ], "data_architecture": { "primary_db": "Postgres (single instance, pilot scale)", "secondary": [ "Encrypted object storage for evidence/memos" ], "cache": "None required at pilot scale", "search": "N/A — citation-linked retrieval only", "vector": "Lightweight embedding index over the Rule Matrix for citation retrieval", "object_storage": "Encrypted, per-fiduciary buckets", "schema_strategy": "Canonical disbursement-request schema versioned with the Rule Matrix", "migrations": "Reviewed, gated migrations; no auto-migrate in production", "backups": "Daily encrypted backups, tested quarterly restore", "retention": "Engagement-letter-defined retention period", "audit_logs": "Append-only, hash-chained per Determination Memo", "soft_delete": "Used for fiduciary offboarding pending legal-hold check", "privacy": "Per-fiduciary row-level isolation; beneficiary benefit-status data heightened handling", "encryption": "At-rest and in-transit encryption on all beneficiary/financial data", "multi_tenancy": "Logical per-fiduciary isolation via row-level auth, not dedicated infrastructure per fiduciary at pilot scale" }, "api": { "style": "Internal service calls only at launch (no public API)", "public_vs_internal": "Internal-only", "versioning": "N/A at launch", "rate_limiting": "Per-tenant concurrency caps on background jobs", "idempotency": "Idempotent intake processing keyed on submission ID", "pagination": "N/A at launch", "error_format": "Structured internal error codes with fiduciary + request ID", "webhook_security": "N/A — no external webhooks at launch", "retries": "Bounded retries on model calls with circuit breaker", "contract_testing": "Contract tests on the canonical schema per submission format", "backward_compat": "N/A at launch", "contract_testing_plan": "Sample submissions from every pilot fiduciary's format run through contract tests on each schema change" }, "security": { "authn": "SSO + MFA for staff", "authz": "Server-side role capability checks (Benefits-Compliance Analyst, Attorney bench, Founder)", "tenant_isolation": "Row-level auth + per-fiduciary retrieval partitioning", "secrets": "Managed secrets store, no secrets in code or logs", "encryption": "At-rest and in-transit on all beneficiary/financial data", "session": "Short-lived sessions, MFA-bound", "input_validation": "Schema validation on every request field before normalization", "api_protection": "N/A — no public API at launch", "audit_log": "Append-only, hash-chained, per-request timestamped", "admin_access": "Least-privilege, logged, reviewed monthly", "supply_chain": "Dependency scanning in CI", "threat_model": [ "Cross-fiduciary data leak", "Wrong-verdict determination reaching a fiduciary", "Beneficiary/family credential capture (hard-blocked)" ], "abuse_cases": [ "A malicious upload attempting prompt injection via a disbursement-request or trust document" ], "zero_trust": "Every internal service call authenticated and authorized, no implicit trust between modules", "asvs_notes": "Level 2 controls targeted given sensitive beneficiary financial/benefit-status data classification" }, "reliability": { "failure_modes": [ "Malformed disbursement-request submission", "Model-call timeout during a portfolio-review volume spike", "A required fact missing without notice" ], "graceful_degradation": "Failed extraction routes to the exception queue rather than blocking the whole cycle", "retry_policy": "Bounded retries with exponential backoff on model calls", "timeouts": "30s extraction cold-path timeout, else backgrounded", "circuit_breaker": "Trips on repeated model-call failures, degrades to manual queue", "queueing": "Per-tenant concurrency-capped background job queue", "idempotency": "Submission-ID-keyed intake processing", "dlq": "Dead-letter queue for failed extraction jobs, reviewed daily", "transactions": "One-aggregate-per-transaction boundary", "dr": "Daily backups, quarterly restore test", "incident_response": "Documented runbook: detect, contain, notify affected fiduciary within 48 hours, root-cause, patch", "slos": [ { "name": "Standard determination on-time delivery", "target": "100% of standard requests within 1 business day" }, { "name": "Free diagnostic turnaround", "target": "1 business day" } ] }, "scaling": { "mvp_can_stay_simple": [ "Single Postgres instance", "No public API" ], "modular_now": [ "Five domain modules mirroring bounded contexts" ], "deferrable": [ "Fiduciary-facing portal", "Live SSA/Medicaid data integration", "Multi-region deployment" ], "breaks_first": "Benefits-Compliance Analyst throughput during a synchronized portfolio-review volume spike across the full cohort", "db_path": "Vertical scaling until multi-analyst concurrency requires read replicas", "jobs_path": "Add worker concurrency before adding a distributed queue", "cache_path": "Introduce caching only if Rule Matrix retrieval latency becomes a bottleneck", "search_path": "N/A at this scale", "files_path": "Object storage scales linearly with fiduciaries onboarded", "api_path": "Introduce a public API only if a fiduciary case-management integration requires it", "multi_region": "Deferred until geographic latency or data-residency requirements emerge", "cost_control": "Per-tenant model-spend budget caps" }, "ai": { "provider": "Frontier LLM API (model-agnostic per ai-engine-spec.md Model Portability)", "prompt_mgmt": "Version-pinned prompts per Rule Matrix release", "rag": "Citation-linked retrieval over the Rule Matrix", "vector": "Lightweight embedding index for citation retrieval", "embeddings": "Rule Matrix citation anchors embedded for retrieval", "eval": "Gold-standard regression suite, run on every prompt/rule-matrix version bump", "hitl": "Three permanent gates: classification, delivery, certification", "guardrails": "Deterministic disclaimer injection; schema validation on every agent output", "prompt_injection": "Uploaded documents treated as untrusted data, never instructions", "leakage": "Per-fiduciary retrieval partitioning; no cross-fiduciary memory", "fallback": "Below-threshold or failed agent output routes to the Exception Queue, never auto-delivers", "latency_cost": "Model-call spans tracked per fiduciary per cycle; per-tenant budget caps", "memory": "Per-fiduciary, per-request scope; no persistent cross-fiduciary memory", "tool_permissions": "Agents can draft and propose; only a human-approved command can deliver", "auditability": "Every agent output logged with its confidence score and cited source", "citation": "Every determination and every citation references a specific POMS/state-rule/ABLE QDE source" }, "devops": { "environments": [ "local", "preview (per-PR)", "staging (shared)", "production (single region + multi-AZ)" ], "observability": [ "Structured logs with fiduciary + request IDs", "RED metrics per workflow", "Error tracking with source maps", "Model-call spans with cost + latency", "Weekly SLO review, with a portfolio-review volume-spike capacity review" ], "testing_pyramid": [ "Unit tests on Bright-Line Check modules (resource limit, ISM formula, ABLE cap) and completeness-gate modules", "Component tests on the Benefits-Compliance Analyst exception-queue surfaces", "Contract tests on the disbursement-request schema per Rule Matrix jurisdiction entry", "End-to-end smoke test on the intake -> classify -> analyst sign-off -> delivery path" ], "accessibility_tests": [ "axe-core in CI on analyst-facing surfaces", "Keyboard-only walkthrough per workflow", "Prefers-reduced-motion honored" ], "performance_budget": "p95 workflow latency published per module; request extraction cold-path under 30s or shown as a background job.", "cicd": "PR -> typecheck + unit + snapshot tests -> preview deploy -> main auto-deploys to a single production region; migrations gated on review.", "iac": "Infrastructure as code for the single-region deployment", "secrets": "Managed secrets store", "preview_envs": "Per-PR preview deploys", "migrations": "Reviewed, gated", "rollback": "Blue/green deploy with instant rollback", "release_style": "Continuous deploy to a single production region on green CI", "feature_flags": "Used for Rule Matrix version rollout per jurisdiction", "monitoring": "RED metrics per workflow", "alerting": "Paged on any missed-SLA or classification-accuracy anomaly", "logs": "Structured, fiduciary + request ID scoped", "error_tracking": "Source-mapped error tracking", "uptime": "Single-region target 99.5% at pilot scale", "cost_monitoring": "Per-tenant model-spend dashboard" }, "testing": { "unit": "Bright-Line Check and completeness-gate modules", "integration": "Cross-module event contracts", "contract": "Canonical schema vs. submission formats", "e2e": "Intake -> classification -> analyst sign-off -> delivery smoke test", "security": "STRIDE-driven test cases per control", "a11y": "axe-core on analyst-facing surfaces", "load": "Portfolio-review volume-spike simulation", "chaos": "Simulated submission-format change", "migration": "Schema-migration dry-run against a snapshot of pilot data", "backup_restore": "Quarterly restore-test", "ai_eval": "Gold-standard regression per Rule Matrix version bump", "test_data": "Synthetic, non-real beneficiary data seeded per jurisdiction" }, "observability": { "logs": "Structured, fiduciary + request ID scoped", "metrics": "RED per workflow + business events (classification accuracy, determinations delivered)", "traces": "Model-call spans with cost + latency", "audit_events": "Append-only, hash-chained", "business_events": "DisbursementRequestSubmitted, DeterminationDrafted, RequestEscalated, DeterminationDelivered", "error_tracking": "Source-mapped", "security_monitoring": "Access-log review, anomaly alerts on cross-fiduciary access attempts", "cost_monitoring": "Per-tenant model-spend dashboard", "dashboards": [ "Analyst exception-queue dashboard", "Aggregate determinations-delivered / SLA-hit scorecard" ], "alert_thresholds": [ "Any determination within 2 hours of its SLA with no analyst action", "Classification accuracy below target sustained 2 weeks" ], "triage": "Analyst on-call reviews any missed-SLA alert within 1 business hour" }, "cost": { "drivers": [ { "name": "Model inference", "note": "$1-2/determination at day-90 automation" }, { "name": "Benefits-Compliance Analyst review minutes", "note": "$8-10/determination, the largest COGS line" }, { "name": "Object storage + hosting", "note": "$3-5/fiduciary/month" } ], "likely_traps": [ "Per-fiduciary snowflake integration requests ballooning analyst minutes past the day-90 target" ], "controls": [ "Per-tenant model-spend budget caps", "Product-vs-custom rule enforced from pilot 1" ] }, "multi_tenancy": { "model": "Logical multi-tenancy (shared infrastructure, per-fiduciary row-level isolation)", "isolation": "Row-level auth + per-fiduciary retrieval partitioning", "tenant_aware_authz": "Every query scoped to the requesting analyst's authorized fiduciary set", "tenant_config": "Per-fiduciary jurisdiction coverage and retainer configuration", "branding": "Internal, analyst-facing branding only — BenefitGuard delivers directly to the fiduciary, no beneficiary-facing artifacts", "tenant_export": "Full data export available to a fiduciary on offboarding", "tenant_deletion": "Deletion SLA per the offboarding SOP, subject to legal hold", "tenant_audit": "Per-fiduciary audit trail, independently reconstructable", "noisy_neighbor": "Per-tenant concurrency and model-spend caps prevent one fiduciary's volume spike from starving another", "tenant_rate_limits": "Per-tenant background-job concurrency caps", "billing": "Per-request or monthly retainer billed to the fiduciary, never to a beneficiary or family", "why_this_fits": "Pilot scale (5-8 fiduciaries) does not justify dedicated per-fiduciary infrastructure; logical isolation meets the sensitivity bar with far lower operational cost" }, "privacy_compliance": { "data_classification": "Beneficiary benefit-status and financial data classified sensitive", "minimization": "Only fields required for classification are ingested", "consent": "Engagement letter signed by the fiduciary before any request is worked", "access_logs": "Reviewed weekly", "audit_trails": "Append-only, hash-chained, per-request timestamped", "retention": "Engagement-letter-defined retention period", "legal_hold": "Deletion paused under an active legal hold", "right_to_delete": "Honored per the offboarding SOP absent a legal hold", "right_to_export": "Full export delivered to the fiduciary on offboarding", "sensitive_handling": "Beneficiary benefit-status records get heightened handling", "boundaries": "No beneficiary/family credentials ever collected; no custody of trust or ABLE funds; no fees billed to beneficiaries or families", "residency": "US-only storage at pilot scale", "vendor_risk": "LLM API vendor reviewed for data-use terms; no training on customer data", "breach_response": "48-hour affected-fiduciary notification runbook", "admin_controls": "Least-privilege, logged, reviewed monthly", "evidence_collection": "Hash-chained snapshots support forensic reconstruction if needed" }, "frontend": { "framework": "Lightweight internal console (React or equivalent) for the analyst exception queue only", "rendering": "Server-rendered where practical; no fiduciary-facing app", "routing": "Internal routes only", "state": "Minimal — queue state and form state", "server_state": "Fetched per analyst session", "forms": "Analyst review/approval forms with explicit labels", "error_handling": "Inline validation with clear next-step messaging", "components": "Exception-queue list, determination detail, approval action buttons", "design_system": "Shared internal component set, no public design system needed at launch", "auth_ui": "SSO login only", "authz_aware_ui": "UI hides actions the logged-in role cannot perform", "a11y": "Keyboard-complete, axe-core in CI", "i18n": "en-US only", "performance": "Not conversion-critical (internal tool); still budget-conscious", "bundling": "Standard bundler, no special constraints", "testing": "Component tests on the exception-queue surface", "offline": "Not required", "realtime": "Not required at pilot scale" }, "backend": { "framework": "Typed service layer (e.g., Node/TypeScript or Python) per domain module", "layering": "domain / application / infra per module", "domain": "Aggregates and value objects per ddd.aggregates", "services": "Application services per bounded context (see ddd.bounded_contexts)", "repositories": "Per-fiduciary-partitioned repositories per aggregate", "validation": "Schema validation at every module boundary", "authorization": "Server-side capability checks per role", "jobs": "Background extraction/classification jobs, per-tenant concurrency capped", "events": "In-process domain event dispatch (see ddd.context_map)", "files": "Encrypted object storage adapters", "email_sms": "Secure email send adapter (analyst-approved only)", "scheduled": "SLA timers (1-business-day standard, 4-hour rush)", "errors": "Structured error codes with fiduciary + request ID", "logging": "Structured logs, PII-scrubbed", "config": "Environment-scoped config, secrets in a managed store", "di": "Constructor-based dependency injection per module", "testing": "Unit + component + contract + e2e per ddd.testing_strategy" }, "diagrams": { "context_mermaid": "graph LR; Fiduciary-->|disbursement request|Intake; Intake-->|RequestDataNormalized|Determination; Determination-->|DeterminationDrafted|ReviewEscalation; ReviewEscalation-->|DeterminationDelivered|PortfolioCompliance; RuleMatrix-->Determination; RuleMatrix-->PortfolioCompliance", "container_mermaid": "graph TD; AnalystConsole-->AppServices; AppServices-->Postgres; AppServices-->ObjectStorage; AppServices-->LLMAPI", "data_flow_mermaid": "graph LR; DisbursementRequest-->Extraction-->CanonicalSchema-->Classification-->DeterminationMemo-->EscalationCase-->PortfolioReview", "auth_flow_mermaid": "sequenceDiagram; Analyst->>SSO: authenticate; SSO->>Console: session token (MFA-bound)", "authz_flow_mermaid": "sequenceDiagram; Console->>AppServices: action request; AppServices->>RoleCheck: verify capability; RoleCheck-->>AppServices: allow/deny", "deployment_mermaid": "graph TD; PR-->Preview; Preview-->Staging; Staging-->Production_SingleRegion", "background_job_mermaid": "graph LR; IntakeQueue-->ExtractionWorker-->ClassificationWorker-->ExceptionQueue", "event_flow_mermaid": "graph LR; DeterminationDrafted-->AnalystReview-->EscalationCheck-->DeliverOrEscalate", "failure_flow_mermaid": "graph LR; MalformedSubmission-->ExceptionQueue-->AnalystReview-->Resolved_or_Escalated", "multi_tenant_flow_mermaid": "graph LR; FiduciaryA_Data-->RowLevelAuth; FiduciaryB_Data-->RowLevelAuth; RowLevelAuth-->PerFiduciaryPartition", "ai_flow_mermaid": "graph LR; Submission-->ExtractionAgent-->ClassificationAgent-->DraftingAgent-->AnalystGate-->Deliver" }, "adrs": [ { "id": "ADR-001", "decision": "Launch as a single-region modular monolith, not microservices", "status": "accepted", "context": "A 5-8-account pilot cap does not justify microservice operational overhead", "options": [ "Modular monolith", "Microservices", "No-code workflow tool" ], "chosen": "Modular monolith", "business_reason": "Fastest path to the first 3-5 pilot determinations without a platform team", "technical_reason": "Module boundaries mirror bounded contexts, keeping a future split cheap", "tradeoffs": "Less independent scalability per module until volume demands it", "risks": [ "[PLACEHOLDER] owner to complete" ], "revisit_trigger": "Sustained load beyond the analyst-throughput target", "why": "Matches the actual pilot scale", "consequences": [ "[PLACEHOLDER] owner to complete" ], "reversal": "Extract a module to a service if its load profile diverges sharply", "revisit_when": "Post-20-pilot pause-and-measure checkpoint" }, { "id": "ADR-002", "decision": "No fiduciary-facing portal at launch; delivery is email/document only", "status": "accepted", "context": "The blueprint's sales thesis is that the fiduciary never operates software", "options": [ "Build a fiduciary-facing dashboard", "Email/document delivery only", "White-labeled third-party portal" ], "chosen": "Email/document delivery only", "business_reason": "Matches the 'we sell a determination, not a dashboard' positioning", "technical_reason": "Removes an entire authn/authz surface at pilot scale", "tradeoffs": "No self-serve visibility for the fiduciary between determinations", "risks": "A fiduciary used to a dashboard may perceive the service as less real-time", "revisit_trigger": "Multiple pilot fiduciaries request live status visibility", "why": "Keeps the MVP thin and evidence-linked", "consequences": "Delivered memos and readout calls carry the full trust burden", "reversal": "Add a lightweight read-only status view if demand is proven", "revisit_when": "After the 10-pilot hardening checkpoint" } ], "roadmap": [ { "phase": "Phase 0 — Stand up the operating spine", "weeks": "Week 1 (Days 1-7)", "outcomes": [ "Rule Matrix v0 (founder's home state + federal POMS baseline) built and attorney-reviewed", "Disbursement Compliance Check landing page + intake form live and tested", "Engagement-letter template drafted; CELA/elder-law attorney-bench relationship opened; 5 warm outreaches sent" ], "exit_criteria": [ "One full test determination processed end-to-end on sample data", "Counsel sign-off on the engagement-letter template", "Intake mailbox tested with one dummy submission" ], "kill_criteria": [ "Any operating-spine item not ready by Day 7 — outreach held even if it slips past Day 7" ], "build": [ "Rule Matrix v0", "Exception-queue console", "SLA timers" ], "defer": [], "monitor": [ "Auto-classification accuracy", "Benefit-suspension incidents", "Analyst review minutes" ], "avoid": [ "Building a fiduciary-facing portal before the 20-pilot checkpoint proves margin" ], "acceptable_debt": [ "[PLACEHOLDER] owner to complete" ], "dangerous_debt": [ "[PLACEHOLDER] owner to complete" ], "triggers_to_change": [ "[PLACEHOLDER] owner to complete" ] }, { "phase": "Phase 1 — Onboard the pilot cohort, run the first determinations", "weeks": "Weeks 2-4 (Days 8-30)", "outcomes": [ "First 3-5 paid pilot determinations delivered manually", "Attorney-bench agreement signed", "3 educational content pieces published", "Rule Matrix expanded to 3-5 states based on actual pilot footprint" ], "exit_criteria": [ "Pilot cap enforced at 5-8 accounts regardless of inbound demand", "Founder personally reviews every exception before delivery", "Exception queue and communications log stood up" ], "kill_criteria": [ "Fewer than 3 signed pilots after 30 days of qualified outreach", "No willingness-to-engage signal above the free-diagnostic offer" ], "build": [ "Pilot intake + engagement-letter flow" ], "defer": [], "monitor": [ "Auto-classification accuracy", "Benefit-suspension incidents", "Analyst review minutes" ], "avoid": [ "Building a fiduciary-facing portal before the 20-pilot checkpoint proves margin" ], "acceptable_debt": "Manual Rule Matrix spreadsheet for the first 5-10 pilots", "dangerous_debt": "Per-fiduciary classification logic that can't be expressed as a rule", "triggers_to_change": "Sustained analyst review time above the day-90 target for 2 consecutive cycles" }, { "phase": "Phase 2 — Hold the cap, prove the model", "weeks": "Weeks 5-8 (Days 31-60)", "outcomes": [ "5-8-account pilot cap reached", "AI drafting layer introduced against a growing adjudicated-example library, 100% analyst review retained", "First Portfolio Monitoring Retainer conversions", "5-pilot hardening checkpoint completed" ], "exit_criteria": [ "1-business-day standard turnaround hit on every determination", "Zero benefit-suspension incidents traced to a delivered determination", "Certified Opinion delivered on at least 1 escalated case" ], "kill_criteria": [ "Pilot-cohort diagnostic-to-paid conversion stalls far below target", "Rework rate persistently above 5%" ], "build": [ "AI drafting layer against a growing adjudicated-example library" ], "defer": [], "monitor": [ "Auto-classification accuracy", "Benefit-suspension incidents", "Analyst review minutes" ], "avoid": [ "Building a fiduciary-facing portal before the 20-pilot checkpoint proves margin" ], "acceptable_debt": "Manual Rule Matrix spreadsheet for the first 5-10 pilots", "dangerous_debt": "Per-fiduciary classification logic that can't be expressed as a rule", "triggers_to_change": "Sustained analyst review time above the day-90 target for 2 consecutive cycles" }, { "phase": "Phase 3 — Formalize SOPs, evaluate the 10-pilot checkpoint", "weeks": "Weeks 9-13 (Days 61-90)", "outcomes": [ "Second Benefits-Compliance Analyst certified independent of the founder", "Written SOPs, escalation matrix, and QA sampling rules formalized", "Quarterly Portfolio Review piloted with at least 1 account", "90-day retrospective completed; 10-pilot checkpoint evaluated with real data" ], "exit_criteria": [ "Analyst shadows Founder on at least 5 live determinations before unsupervised exception review", "10-pilot hardening checklist complete (intake spec, evidence requirements, QA checklists)", "Gross margin trending toward the blueprint's 50-55% target" ], "kill_criteria": [ "Escalation-to-attorney rate exceeds 30% with no improving trend", "Analyst review time not declining by pilot 15-20" ], "build": [ "Second-analyst certification" ], "defer": [], "monitor": [ "Auto-classification accuracy", "Benefit-suspension incidents", "Analyst review minutes" ], "avoid": [ "Building a fiduciary-facing portal before the 20-pilot checkpoint proves margin" ], "acceptable_debt": "Manual Rule Matrix spreadsheet for the first 5-10 pilots", "dangerous_debt": "Per-fiduciary classification logic that can't be expressed as a rule", "triggers_to_change": "Sustained analyst review time above the day-90 target for 2 consecutive cycles" } ], "anti_overengineering": { "flagged": [ { "item": "Fiduciary-facing portal at launch", "why": "No pilot fiduciary has asked for self-serve access; the sales thesis is 'the fiduciary never operates a tool'", "simpler": "Email/document delivery + readout calls" }, { "item": "Microservices architecture", "why": "A 5-8-account pilot cap does not generate load requiring independent service scaling", "simpler": "Modular monolith with clean module boundaries mirroring bounded contexts" }, { "item": "Live SSA/Medicaid data integration", "why": "1-business-day cadence does not require live sync; adds integration-project overhead the MVP explicitly avoids", "simpler": "Fiduciary-supplied benefit-status confirmation at intake" } ] }, "risks": [ { "risk": "Benefits-Compliance Analyst bottleneck during a synchronized portfolio-review spike", "likelihood": "medium", "impact": "high", "mitigation": "Certify a second analyst after 5-10 pilots; stagger onboarding cohorts where possible", "detection": "Exception-queue backlog alert", "owner": "owner", "escalation": "Founder personally backstops until a second analyst is certified", "fallback": "Temporarily pause new-fiduciary onboarding", "category": "operational" }, { "risk": "Submission-format change breaks the Extraction Agent silently", "likelihood": "medium", "impact": "medium", "mitigation": "Contract tests per submission format; confidence-score drop alerting", "detection": "Extraction confidence-score drop alert", "owner": "engineering", "escalation": "Engineering Lead patches the parser within 48 hours", "fallback": "Manual extraction for the affected fiduciary until patched", "category": "technical" }, { "risk": "A state bar objects to the non-certified tier's positioning", "likelihood": "low-medium", "impact": "high", "mitigation": "Counsel-reviewed engagement-letter scoping before the first paid engagement; escalation criteria route borderline cases automatically", "detection": "Counsel's periodic regulatory scan", "owner": "legal", "escalation": "Counsel advises routing 100% of engagements through attorney sign-off temporarily", "fallback": "Certified-only operating mode until the boundary is resolved", "category": "regulatory" } ], "rules": [ "No dollar figure is ever computed by the model — all money math is deterministic code", "No Determination Memo ships without a Benefits-Compliance Analyst approval", "No beneficiary/family credential is ever collected, stored, or requested" ], "audit": { "product_fit": 5, "simplicity": 4, "security": 4, "reliability": 4, "scalability": 4, "maintainability": 4, "performance": 4, "cost": 4, "compliance": 4, "dx": 4, "ops_burden": 4, "extensibility": 4, "team_suitability": 4, "time_to_market": 5, "recommendation": { "verdict": "Proceed — single-region modular monolith is the right-sized architecture for a 5-8-account pilot", "stack": "Typed backend service layer + Postgres + encrypted object storage + frontier LLM API", "style": "Modular monolith", "database": "Postgres, row-level partitioned", "hosting": "Single-region, multi-AZ", "auth": "SSO + MFA", "ai_approach": "Retrieval-first, citation-linked, deterministic-rules-backed agents with a permanent human-plus-attorney approval gate", "integrations": "Email/upload intake, secure email delivery, encrypted object storage", "build_first": [ "Rule Matrix v0 for federal POMS + founder's home state", "Exception-queue console" ], "revisit_later": [ "Fiduciary-facing portal", "Live SSA/Medicaid data integration" ], "avoid": [ "Microservices before a few dozen fiduciaries per analyst", "Public API before a partner integration requires it" ], "biggest_risks": [ "Benefits-Compliance Analyst bottleneck at a synchronized volume spike", "Submission-format change breaking extraction silently" ], "top_10_rules": [ "No dollar figure is ever computed by the model — all money math is deterministic code", "No Determination Memo ships without a Benefits-Compliance Analyst approval", "No beneficiary/family credential is ever collected, stored, or requested", "Row-level auth on every query", "Append-only audit log on every state transition", "Attorney sign-off required on every Certified Opinion" ], "first_10_steps": [ "Stand up the intake mailbox, engagement-letter template, and Disbursement Compliance Check landing page", "Populate the owner-action ledger with entity, jurisdiction, contact inbox, privacy inbox", "Open founding-pilot intake (cap 5-8) with an explicit free-diagnostic conversation", "Write the home-state-first, multi-state-second sequencing and kill/pivot criteria into the launch plan", "Codify federal POMS into the Rule Matrix v0 with citation anchors", "Wire the append-only audit log + hash-chained Determination Memo snapshots", "Add row-level auth + fiduciary-scoped retrieval indices", "Instrument diagnostic-request, determination-delivered, and analyst-sign-off events", "Publish the microsite noindexed until owner-action facts close", "Schedule the Day-90 kill/pivot review with the Founder" ] } } }, "design": { "slug": "special-needs-trust-able-disbursement-compliance-desk", "archetypes": [ "compliance determination desk", "elder & disability-benefits advisory-adjacent service" ], "user_mindset": { "goals": "Confirm a proposed disbursement is safe and get a citable record fast", "session_length": "Short, task-focused (a determination request, a readout, a portfolio review read)", "confidence": "Wants precision (citations, dollar impact, dates), not reassurance language", "interface_needs": "Scannable verdict at a glance; no jargon; clear next action" }, "posture": [ "Quiet authority", "Citation-first", "Fiduciary-branded warmth is not needed — BenefitGuard is the buyer's own internal desk, not a family-facing product" ], "density": "Medium — enough white space to read calmly, enough data density to feel like a real compliance file, not a marketing page pretending to be a dashboard", "trust_level": { "tier": "high — handles beneficiary benefit-status data and trust financial data", "sensitive_domains": [ "Beneficiary SSI/SSDI/Medicaid enrollment status", "Trust and ABLE account financial data" ], "implications": [ "Every claim must be sourced or marked [PLACEHOLDER]", "No decorative motion or gamified UI patterns" ] }, "differentiation": { "avoid": [ "Insurtech/fintech gradient hero clichés", "Stock elderly-couple or wheelchair-user imagery", "Countdown-timer urgency gimmicks that read as pressure rather than information" ], "strategy": "Citation-precise, benefits-law-specific visual language: a verdict strip, Bright-Line Check badges, and a determination-memo structure rendered as real artifacts, not abstract icons" }, "territories": [ { "name": "Ledger Guardian", "color_mood": "deep indigo-navy brand, warm paper surface, amber-gold accent", "typography": "System serif-adjacent display, clean sans body", "density": "medium", "component_feel": "Determination-memo precision — verdict badges, citations, real dates", "motion": "Minimal, functional only", "fits": "A buyer who wants proof and liability protection, not persuasion", "risks": "Could read as plain if not paired with sharp, specific copy" }, { "name": "Alarm Fintech", "color_mood": "red/orange urgency, dark dashboard chrome", "typography": "Bold condensed display", "density": "high", "component_feel": "Countdown timers, flashing badges", "motion": "Pulsing alerts", "fits": "A buyer who responds to fear", "risks": "Reads as pressure-selling to a fiduciary already carrying real personal liability — banned by DESIGN-STANDARD anti-corruption rule" }, { "name": "Generic SaaS", "color_mood": "blue/purple gradient", "typography": "Default startup sans", "density": "low", "component_feel": "Rounded cards, soft shadows", "motion": "Hover-lift everywhere", "fits": "Nothing specific to this business", "risks": "Visually confusable with any other launched business — banned by DESIGN-STANDARD §9" } ], "chosen_territory": "Ledger Guardian", "chosen_rationale": "The buyer is a professional fiduciary evaluating whether to trust a service with a beneficiary's benefit status and their own fiduciary liability; precision and calm authority convert better than urgency theater, and the deep indigo/amber palette is visually distinct from both the freight business's navy/burnt-orange industrial palette and the ESA business's pine-teal/berry palette.", "prioritized_components": [ { "name": "Verdict strip", "why": "Makes the three-outcome determination (approve/approve-with-conditions/do-not-disburse) tangible at a glance" }, { "name": "Bright-Line Check badge", "why": "Uses the --state-* tokens to make a deterministic rule outcome legible without alarming color" }, { "name": "Determination Memo preview card", "why": "Shows the actual deliverable shape, building trust before purchase" }, { "name": "Citation chip", "why": "Signals the POMS/state-rule precision that differentiates from a generic AI chatbot answer" }, { "name": "Pricing card with guarantee inline", "why": "Puts the turnaround guarantee and non-guarantee disclaimer where risk perception peaks, per DESIGN-STANDARD §8" }, { "name": "FAQ disclosure list", "why": "Handles real objections without a wall of text" } ], "tokens": { "brand": "222 46% 24%", "brand-fg": "0 0% 100%", "surface": "45 22% 96%", "ink": "222 32% 12%", "muted": "222 12% 36%", "accent": "36 82% 32%" }, "type": { "display": "system-ui, 'Georgia', serif fallback for headings", "body": "system-ui, -apple-system, 'Segoe UI', sans-serif", "fonts_url": "none — system font stack only, no external font requests" }, "signature": { "motif": "A verdict strip (Approve / Approve with conditions / Do not disburse) rendered with the four domain-state tokens, echoed in miniature across the hero, pricing, and playbook artifacts", "render": "Three badge chips in a horizontal row, colored by --state-ok/--state-pending/--state-blocked, with a citation chip beneath the active verdict" }, "patterns": [ { "name": "Status badge", "description": "A pill using the four --state-* tokens (ok/risk/pending/blocked) for any determination or review status, never a raw color" }, { "name": "Citation chip", "description": "A small inline tag citing the specific POMS section, state rule, or ABLE QDE guidance for any factual claim in a memo or on the page" }, { "name": "Placeholder proof slot", "description": "A dashed-border card honestly labeled [PLACEHOLDER] and keyed to a real future event, never a fabricated result" } ], "states": [ "ok — approved / reconciled and on file", "risk — dollar-at-risk or do-not-disburse flag", "pending — awaiting analyst or attorney review", "blocked — escalated / exception-queue status" ], "uniqueness_audit": { "app_specific_decisions": [ "Verdict-strip motif tied to the exact three-outcome determination structure", "Citation-chip pattern unique to the POMS/state-rule-heavy classification workflow" ], "cliches_avoided": [ "No countdown-timer alarm styling", "No stock elderly-couple/wheelchair photography", "No generic 'AI dashboard' gradient hero" ], "scale_notes": "Palette and components scale from a single-page microsite to an internal analyst console without a re-theme." }, "localization": [ "en-US only at launch; no locale-specific formatting required" ] }, "seo": { "slug": "special-needs-trust-able-disbursement-compliance-desk", "archetype": "vertical compliance-determination desk / diagnostic-led service site", "archetype_impact": "Ranks primarily on benefits-mechanics questions (SSI resource limit, ISM formula, ABLE QDE) rather than generic 'AI compliance' terms.", "authority_dna": { "site_archetype": "single-vertical service site with a free-diagnostic lead magnet", "monetization_model": "per-request fee + portfolio retainer subscription + flat certified-opinion fee", "main_search_intents": [ "informational (benefits-mechanics questions)", "commercial (disbursement compliance determination service)" ], "topical_authority_opportunity": "SNT/ABLE disbursement-compliance mechanics content is thin and scattered across nonprofit fact sheets and elder-law blog posts — a citation-rich, maintained explainer set can own this niche quickly", "local_seo_opportunity": "Low — buyers search benefits-mechanics terms nationally, not 'near me'", "global_national_opportunity": "National within the US SSI/Medicaid/ABLE framework; state-specific content clusters added as the Rule Matrix expands", "easiest_ranking_path": "Long-tail benefits-mechanics questions with low current competition ('can a special needs trust pay for a vacation')", "hardest_ranking_path": "Broad terms like 'special needs trust' or 'ABLE account' dominated by nonprofit and fintech incumbents", "trust_credibility_requirements": [ "Citations to primary POMS/IRS/ABLE NRC sources", "Clearly named entity (BenefitGuard) so assistants attribute answers", "No fabricated statistics — every figure traces to a Verified source" ], "ymyl": true, "expert_review_needed": true, "site_structure": "Flat: homepage -> a small set of benefits-mechanics pillar pages -> FAQ/glossary supporting pages -> /blueprint/ dossier", "seo_moat": "Citation-anchored, continuously updated benefits-mechanics content that AI assistants can quote with a named source" }, "search_market": { "primary_markets": [ "Professional fiduciaries and trust officers", "Elder-law and settlement-planning firms", "Pooled-trust nonprofit administrators" ], "secondary_markets": [ "ABLE state program administrators", "Personal-injury attorneys establishing first-party SNTs" ], "low_competition_subtopics": [ "ISM presumed-maximum-value formula worked examples", "2026 ABLE age expansion mechanics for trustees", "d4A vs d4C pooled trust disbursement differences" ], "high_commercial_intent": [ "special needs trust disbursement compliance service", "free disbursement compliance check", "ABLE account compliance determination" ], "informational": [ "can a special needs trust pay for a vacation", "will this disbursement affect SSI", "what is a qualified disability expense for ABLE" ], "local_intent": [], "transactional": [ "disbursement compliance review service", "special needs trust compliance desk" ], "comparison": [ "BenefitGuard vs True Link Financial", "in-house benefits review vs outsourced determination desk" ], "problem_solution": [ "accidentally overpaid special needs trust distribution", "SSI suspended after trust disbursement" ], "near_me": [], "long_tail": [ "what happens if an snt pays rent directly to a landlord", "able account medicaid payback explained", "2026 able age adjustment act who is now eligible" ], "questions": [ "is a disbursement compliance desk allowed under ssa rules", "who signs off on a special needs trust distribution determination", "how much does a professional fiduciary charge per disbursement" ], "emerging": [ "SSI Savings Penalty Elimination Act tracking", "ABLE age-expansion caseload growth for trustees" ], "seasonal": [], "underserved_serps": [ "ISM presumed-maximum-value worked dollar examples", "pooled trust vs individual SNT disbursement-compliance differences" ], "weak_serps": [ "professional fiduciary fee comparison" ], "forum_dominated_serps": [ "reddit/facebook special needs parent group discussions of SSI overpayment notices" ], "winnable_authoritative_serps": [ "SSI resource-limit and ISM formula explainer", "2026 ABLE Age Adjustment Act explainer for trustees" ], "avoid_initially": [ "Broad 'disability benefits policy' debate content — not the buyer's search intent" ], "easy_wins": [ "$2,000 resource-limit explainer", "ISM presumed-maximum-value worked example", "2026 ABLE age expansion explainer" ], "moderate": [ "Pooled trust vs individual SNT disbursement-compliance guide", "Certified Opinion / UPL-boundary explainer" ], "long_term_plays": [ "Multi-state Medicaid-supplement mechanics hub as Rule Matrix coverage grows" ], "do_not_pursue": [ "General disability-rights policy commentary", "Political benefits-reform debate content" ] }, "keyword_clusters": [ { "primary": "disbursement compliance check special needs trust", "related": [ "free snt compliance check", "able account compliance review" ], "intent": "commercial", "user_problem": "Doesn't know if a proposed disbursement is safe to approve", "funnel": "top", "business_value": "high", "ranking_difficulty": "low", "conversion_potential": "high", "content_effort": "medium", "serp_weakness": "no incumbent owns this exact phrase", "local_relevance": "national", "global_relevance": "national", "suggested_page_type": "landing page", "reason": "Direct match to the flagship lead magnet", "priority_score": 95, "priority": "high", "bucket": "primary" }, { "primary": "can a special needs trust pay for X", "related": [ "snt allowable expenses", "what can an snt pay for without affecting benefits" ], "intent": "informational", "user_problem": "Doesn't know if a specific expense is safe", "funnel": "top", "business_value": "high", "ranking_difficulty": "low", "conversion_potential": "medium", "content_effort": "low", "serp_weakness": "thin, scattered coverage", "local_relevance": "national", "global_relevance": "n/a", "suggested_page_type": "pillar article", "reason": "High-intent, mechanics-specific, low competition", "priority_score": 90, "priority": "high", "bucket": "primary" }, { "primary": "2026 ABLE age adjustment act explained", "related": [ "able eligibility age 46", "able age expansion trustee guide" ], "intent": "informational", "user_problem": "Fiduciary doesn't know their caseload just expanded", "funnel": "top", "business_value": "high", "ranking_difficulty": "low", "conversion_potential": "medium", "content_effort": "low", "serp_weakness": "coverage exists but not organized for trustees", "local_relevance": "national", "global_relevance": "n/a", "suggested_page_type": "pillar article", "reason": "Timed to the January 2026 eligibility expansion", "priority_score": 88, "priority": "high", "bucket": "secondary" } ], "topical_authority_map": { "core_topics": [ "SSI/POMS trust and ISM mechanics", "ABLE account QDE and eligibility mechanics", "Special needs trust disbursement compliance as a discipline" ], "pillars": [ { "name": "SSI & ABLE Disbursement Mechanics Hub", "core_intent": "informational -> commercial", "audience": "Professional fiduciaries and trust officers", "conversion_goal": "Disbursement Compliance Check request", "supporting_pages": [ "$2,000 resource-limit explainer", "ISM presumed-maximum-value worked example", "2026 ABLE age expansion explainer" ], "internal_links": [ "/", "/blueprint/" ], "schema": [ "FAQPage", "Article" ], "evidence_needed": [ "SSA POMS citation", "ABLE National Resource Center citation" ], "local_variants": [], "national_variants": [ "State Medicaid-Supplement Mechanics Hub" ] } ], "supporting_page_types": [ "Benefits-mechanics explainer", "Worked dollar example (diagnostic teardown)", "FAQ/glossary page", "Certified Opinion / UPL-boundary explainer", "State Medicaid-supplement variant page" ] }, "site_architecture": { "homepage_strategy": "Single conversion-focused landing page; benefits-mechanics depth lives in supporting content, not the homepage", "main_nav": [ "How it works", "Pricing", "Get your Check" ], "footer_nav": [ "View the full operating blueprint dossier", "Compliance & licensing boundary" ], "hubs": [ { "name": "SSI & ABLE Disbursement Mechanics Hub", "purpose": "Own benefits-mechanics search intent and feed the diagnostic-check CTA", "url": "/special-needs-trust-able-disbursement-compliance-desk/disbursement-mechanics" } ], "url_patterns": [ "/special-needs-trust-able-disbursement-compliance-desk/{topic-slug}" ], "avoid_url_patterns": [ "Deep nested category URLs unneeded at this content volume" ] }, "global_national": { "national_clusters": [ "Multi-state Medicaid-supplement mechanics hub as Rule Matrix coverage grows" ], "linkable_assets": [ "The SNT/ABLE Disbursement Red-Flag Checklist (downloadable)" ], "original_research_ideas": [ "Anonymized aggregate 'common disbursement mistake' pattern memos drawn from pilot-cohort exception data" ], "international_needed": false, "international_notes": "US-only program; no international relevance" }, "local_seo": { "justified": false, "reason": "Buyers search by federal/state benefits program, not by city/proximity", "gbp_categories_primary": [], "gbp_categories_secondary": [], "location_page_rules": [], "citations": [], "review_strategy": "No fabricated reviews; testimonials only after real pilot outcomes exist, per DESIGN-STANDARD §9", "local_schema": [] }, "programmatic": { "recommended": false, "reason": "Content volume and buyer count at pilot scale don't justify programmatic page generation; risk of thin-content penalties outweighs benefit", "rules": [], "per_page_requirements": [], "quality_gates": [] }, "page_templates": [ { "page_type": "Benefits-mechanics pillar article", "purpose": "Own a specific compliance-mechanics question", "target_intent": "informational -> commercial", "url_pattern": "/special-needs-trust-able-disbursement-compliance-desk/{mechanic}", "title_pattern": "{Mechanic} explained: what trustees must know", "meta_description_pattern": "{Mechanic} in plain language, with the exact citation, plus a free Disbursement Compliance Check to verify your own case.", "h1_pattern": "{Mechanic}: the compliance math", "outline": [ "What the rule says (cited)", "Why it costs a beneficiary benefits if mishandled", "A worked dollar example", "How BenefitGuard enforces it", "FAQ", "Diagnostic Check CTA" ], "above_the_fold": [ "[PLACEHOLDER] owner to complete" ], "internal_links": [ "/" ], "schema": [ "Article", "FAQPage" ], "cta_strategy": "Diagnostic Check request after the worked example", "conversion_elements": [ "Inline diagnostic CTA", "Verdict-strip visual" ], "trust_elements": [ "Primary-source citation", "Compliance disclaimer" ], "faq_opportunities": [ "Is this allowed under SSA's own rules?" ], "media": [ "[PLACEHOLDER] owner to complete" ], "quality_requirements": [ "Every citation traces to a primary source" ], "anti_thin_rules": [ "No page ships without a worked dollar example and a citation" ] } ], "on_page_rules": { "title_tag": "Under 60 characters, leads with the mechanic or outcome, never 'AI-Native Blueprint'", "meta_description": "Under 155 characters, states the concrete outcome and cites a real figure", "headings": "One H1 per page, ordered H2/H3", "intro": "States the rule/answer in the first two sentences", "snippet_targeting": "Direct Q&A format for featured-snippet eligibility", "tables_lists": "Dollar figures and thresholds in tables where possible", "images": "Verdict-strip diagrams with meaningful alt text", "internal_links": "Every pillar links to the Diagnostic Check CTA and the blueprint dossier", "external_citations": "Primary source (POMS, IRS, ABLE NRC) cited on every factual claim", "author_attribution": "Founder byline with program-expertise framing", "freshness": "Dated 'last verified' line, re-checked each legislative session", "cta_placement": "After the worked example and again at the page end", "mobile": "Single-column, no horizontal scroll", "avoid": [ "Keyword stuffing", "Unsourced statistics" ] }, "entity_seo": { "main_entities": [ "BenefitGuard", "Determination Memo", "Disbursement Compliance Check" ], "related_entities": [ "Social Security Administration (SSA)", "ABLE National Resource Center", "POMS SI 01120.200", "IRC §529A" ], "people": [ "Founder (role-title only, no fabricated bio claims)" ], "orgs": [ "Social Security Administration", "ABLE National Resource Center", "CPT Institute", "True Link Financial" ], "tools": [ "Rule Matrix" ], "regulations": [ "SSA POMS SI 01120.200 / SI 02301.205 / SI 01130.740", "IRC §529A (ABLE Act, as amended by the ABLE Age Adjustment Act)" ], "problems": [ "ISM presumed-maximum-value SSI reduction", "$2,000 resource-limit suspension" ], "solutions": [ "Determination Memo", "Benefit-Impact Projection", "Restructuring Alternative" ], "processes": [ "Bright-Line Check", "Rule Matrix classification", "Attorney-bench escalation" ], "alternatives": [ "Full-service professional fiduciary/corporate trustee firm", "Spending-control prepaid card" ], "synonyms": [ "special needs trust = SNT", "in-kind support and maintenance = ISM", "qualified disability expense = QDE" ] }, "schema_strategy": [ { "type": "FAQPage", "where": "Homepage FAQ section and every pillar article", "required_fields": [ "question", "acceptedAnswer" ], "caution": "Only include questions actually answered on-page; no schema for entities that don't exist yet (no reviews/ratings)" }, { "type": "Organization", "where": "Homepage (deferred until owner-action facts, e.g. registered entity name, are confirmed)", "required_fields": [ "name", "url" ], "caution": "noindex until owner-action facts close, per seo_pages.schema_status" } ], "internal_linking": { "pillar_to_cluster": "Each pillar links to its 2-3 supporting explainer pages", "cluster_to_pillar": "Every supporting page links back to its pillar and the homepage CTA", "cluster_to_cluster": "Related jurisdiction-mechanics pages cross-link once both exist", "service_to_location": "N/A — no location pages", "faq_to_commercial": "Every FAQ answer links to the Diagnostic Check CTA where relevant", "breadcrumbs": "Home > Disbursement Mechanics > {page}", "anchor_text_rules": [ "Use the exact benefits/program term as anchor text, never 'click here'" ] }, "technical_seo": { "crawlability": "Single-page site at launch; sitemap added once pillar content ships", "indexability": "noindex until owner-action facts close (see seo_pages.noindex)", "sitemaps": "XML sitemap generated once pillar pages ship", "robots": "Disallow nothing once live; noindex meta during pre-launch", "canonicals": "Self-canonical on every page", "pagination": "N/A at this content volume", "faceted_nav": "N/A", "duplicate_control": "One URL per mechanic/topic", "redirects": "N/A at launch", "core_web_vitals": "Single self-contained HTML file keeps LCP/INP/CLS well within budget", "mobile": "Single-column responsive layout", "accessibility": "WCAG 2.2 AA per DESIGN-STANDARD §6", "js_seo": "Minimal JS, no client-side-only rendering of critical content", "rendering": "Static HTML", "gsc_setup": "Pending owner-action: register property post-launch", "analytics_setup": "Pending owner-action: attach analytics post-launch", "rank_tracking": "Pending owner-action: set up post-launch" }, "eeat": { "author_bios": "Founder byline with program-expertise framing (role-title only)", "expert_reviewers": "[PLACEHOLDER — pending a named CELA or elder-law reviewer]", "editorial_policy": "Every factual claim cites a primary source; Verified/Inferred/Unverified labeling carried over from the blueprint", "fact_checking": "Cross-checked against primary POMS/IRS/ABLE NRC sources before publishing", "credentials": [ "[PLACEHOLDER — founder credentials to be added when confirmed]" ], "citations": "Direct links/citations to SSA POMS, IRS ABLE guidance, ABLE National Resource Center", "first_hand_proof": [ "[PLACEHOLDER — first pilot-cohort determination example, anonymized, once available]" ], "update_cadence": "Re-verified each legislative session or POMS/state-manual release", "monetization_disclosure": "Pricing and guarantee stated plainly on the pricing section, no hidden monetization", "ymyl_notes": "Financial/compliance-adjacent content for a YMYL-sensitive audience; every claim sourced or marked [PLACEHOLDER]" }, "ai_search": { "principles": [ "Answer the exact buyer phrasing directly and early", "Cite primary sources so an assistant can attribute the answer" ], "tactics": [ "Direct Q&A structure", "Named-entity consistency (always 'BenefitGuard')" ], "do_not": [ "Keyword-stuff for AI crawlers", "Publish unsourced statistics" ] }, "conversion": { "primary_cta": "Get your free Disbursement Compliance Check", "secondary_cta": "View the full operating blueprint dossier", "lead_magnets": [ "Disbursement Compliance Check", "SNT/ABLE Disbursement Red-Flag Checklist (downloadable)" ], "trust_elements": [ "Cited stat strip", "Compliance/licensing-boundary section", "Honest [PLACEHOLDER] proof slots" ], "per_page_paths": [ { "path": "/", "page_type": "landing page" } ], "tracking": "data-event attributes (ui.*/domain.*) per DESIGN-STANDARD §4" }, "link_earning": { "digital_pr_ideas": [ "Aggregate, anonymized 'common disbursement mistake' pattern memo pitched to elder-law trade press" ], "original_research": [ "Diagnostic-teardown worked-example format published as a recurring content series" ], "directories": [ "NAELA state chapter resource pages", "specialneedsanswers.com directory listings" ], "expert_contributions": [ "Guest posts/webinars with NAELA state chapters and pooled-trust networks" ], "partnerships": [ "Settlement-planning and PI-attorney referral partners", "Pooled-trust nonprofit co-marketing" ], "avoid": [ "Paid link schemes", "Guest posts on unrelated domains" ] }, "roadmap_90d": [ { "phase": "Phase 1", "goal": "Publish the SSI & ABLE mechanics hub and the flagship Disbursement Compliance Check landing page", "pages": [ "Homepage", "$2,000 resource-limit explainer", "ISM presumed-maximum-value worked example" ], "keywords_targeted": [ "disbursement compliance check special needs trust", "can a special needs trust pay for X" ], "required_assets": [ "Verdict-strip diagram" ], "internal_links": [], "difficulty": "low", "business_value": "high", "conversion_goal": "Diagnostic Check requests", "why_first": "Matches the founding-pilot outreach motion already underway" } ], "roadmap_12m": [ { "phase": "Phase 2", "goal": "Add a state Medicaid-supplement mechanics hub timed to Rule Matrix state expansion", "pages": [ "State Medicaid-supplement explainer", "Certified Opinion / UPL-boundary explainer" ], "keywords_targeted": [ "state medicaid supplement disbursement rules" ], "required_assets": [ "State coverage map" ], "internal_links": [ "/" ], "difficulty": "low", "business_value": "high", "conversion_goal": "Portfolio Monitoring Retainer signups", "why_first": "Timed to the pilot cohort's actual multi-state footprint" } ], "priority_pages": [ { "rank": 1, "page_title": "The $2,000 resource-limit trap: how disbursements can trigger it", "slug": "ssi-resource-limit-trap", "page_type": "pillar article", "primary_keyword": "ssi resource limit special needs trust", "secondary_keywords": [ "$2000 countable resource limit trust" ], "intent": "informational", "funnel": "top", "business_value": "high", "difficulty": "low", "conversion_potential": "medium", "cta": "Diagnostic Check", "schema": [ "Article", "FAQPage" ], "internal_links": [ "/" ], "required_proof": [ "[PLACEHOLDER] owner to complete" ], "why_opportunity": "Thin, scattered existing coverage of the exact mechanic", "production_priority": "P0", "scope": "single article" } ], "competitor_gaps": { "typical_competitor_types": [ "Fintech spending-control card marketing pages", "Nonprofit fact-sheet pages", "Elder-law firm blog posts" ], "common_weaknesses": [ "No worked dollar examples", "No citation-level rigor tied to a delivered determination" ], "how_to_beat_them": [ "Publish citation-anchored, dated, worked-example content no incumbent produces, tied directly to a free diagnostic conversion" ] }, "metrics": { "weekly": [ "Diagnostic Check requests", "Organic sessions to pillar pages" ], "monthly": [ "Check-to-pilot conversion", "New keyword rankings" ], "quarterly": [ "Topical authority coverage vs. plan" ], "annual": [ "Renewal-driven content refresh completion" ] }, "risks": [ { "risk": "YMYL content scrutiny given financial/benefits-compliance subject matter", "applies": true, "mitigation": "Every claim sourced or marked [PLACEHOLDER]; compliance disclaimers on every page" }, { "risk": "Thin programmatic content penalty", "applies": false, "mitigation": "Programmatic generation explicitly not recommended at this stage" } ], "first_20_pages": [ "Homepage", "$2,000 resource-limit explainer", "ISM presumed-maximum-value worked example", "2026 ABLE age expansion explainer" ], "first_10_tech_fixes": [ "Confirm noindex removed only after owner-action facts close", "Add XML sitemap once pillar pages ship" ], "first_10_authority_actions": [ "Publish the 10-post educational content series", "Open NAELA state-chapter webinar co-marketing conversations", "Publish the SNT/ABLE Disbursement Red-Flag Checklist lead magnet" ], "final_recommendation": "Launch the homepage and the SSI & ABLE mechanics hub together; hold multi-state content for the timed Rule Matrix expansion; keep every claim citation-anchored given the YMYL sensitivity of financial/benefits-compliance content for fiduciaries.", "disclaimers": [ "This SEO plan is operational guidance, not legal or tax advice.", "All cited figures must be re-verified against primary sources before publishing.", "No fabricated reviews, ratings, or case results — proof content ships only as real pilot outcomes exist." ] }, "microsite": { "category": "Special needs trust (SNT) & ABLE account disbursement compliance for professional fiduciaries", "shortTitle": "BenefitGuard Disbursement Compliance Desk", "audience": "Professional fiduciaries (independent trustees, trust-company trust officers, elder-law firm trust departments) administering 15+ active SNTs and/or ABLE accounts", "problem": "Every SNT or ABLE disbursement must be individually checked against a narrow, unforgiving federal rule set — and most fiduciaries have no dedicated benefits-compliance specialist reviewing every request before they approve it.", "offer": "A done-for-you disbursement compliance desk: a dated, cited Determination Memo within one business day, a Benefit-Impact Projection, a Restructuring Alternative where relevant, and a licensed attorney's Certified Opinion on demand — priced per request or as a flat retainer, never hourly, starting with a free Disbursement Compliance Check.", "faq": [ { "q": "Our elder-law attorney already reviews this at trust drafting — why do we need BenefitGuard?", "a": "Ask your attorney who checks every one of your 40 trusts' individual disbursement requests month to month, at a flat per-request rate instead of an hourly one. Trust drafting is a one-time event; disbursement compliance is a continuous, per-transaction need — that gap is what the desk closes." }, { "q": "We can't afford another vendor.", "a": "Standard determinations are $45 per request, anchored just below CPT Institute's own published $35 rush-fee precedent. The free Disbursement Compliance Check shows you the format and rigor on your own first request before you spend anything." }, { "q": "Is this even allowed — are you practicing law?", "a": "No. Standard determinations are benefits-compliance research delivered to you as the professional-fiduciary customer of record, not legal advice to a beneficiary or family. Novel, high-value, or Certified Opinion cases are reviewed and signed by a licensed elder-law attorney." }, { "q": "What exactly do you need from us to start?", "a": "The disbursement description, amount, and payee; your beneficiary's current SSI/SSDI/Medicaid status and state; and the trust's governing document or ABLE program enrollment (on file after your first submission — you don't resubmit it every time)." }, { "q": "What if SSA later disagrees with a determination?", "a": "No compliance service can guarantee an agency's future finding, and we don't claim to. Every memo is dated to the Rule Matrix version and facts submitted, with a full Audit Trail — the actual evidence of due diligence you need if a distribution is later questioned." }, { "q": "Do you take custody of trust or ABLE funds, or act as trustee?", "a": "Never. You remain trustee and decision-maker of record at all times. BenefitGuard supplies the compliance research and, where escalated, the licensed sign-off." }, { "q": "What happens with a novel or high-value request?", "a": "It routes automatically to our attorney bench for a Certified Opinion — a genuine licensed-attorney work product, priced as a flat $350 fee, never hourly." }, { "q": "How is this priced?", "a": "$45 standard / $95 rush per Disbursement Request. Portfolio Monitoring Retainer: $60/trust/month (bundles up to 3 determinations). Certified Opinion: $350 flat. Quarterly Portfolio Review: from $500 per 25 trusts/accounts. Never hourly." }, { "q": "What's the catch with the free Disbursement Compliance Check?", "a": "There isn't one beyond your time: send your first real disbursement request, and within one business day you get the same determination format minus attorney sign-off. No pitch, no obligation to continue." } ], "process": [ { "title": "Submit", "body": "Send the disbursement description, amount, payee, and your beneficiary's current benefit status and state — by secure portal or email, no account setup required." }, { "title": "Classify", "body": "The Rule Matrix and deterministic Bright-Line Checks run first, then a first-pass determination is drafted with full citations and a confidence score." }, { "title": "Analyst review", "body": "A Benefits-Compliance Analyst reviews every draft before delivery — 100% of cases, no exceptions." }, { "title": "Escalate if triggered", "body": "Novel fact patterns, high-dollar requests, low-confidence classifications, and every Certified Opinion route automatically to a licensed elder-law attorney." }, { "title": "Deliver & log", "body": "You receive the formatted Determination Memo with full citations; the case is logged to your portfolio and feeds your Quarterly Portfolio Review if you're on retainer." } ], "northStarCta": { "label": "Get your free Disbursement Compliance Check", "href": "#diagnostic", "secondary_label": "View the full operating blueprint dossier", "secondary_href": "/special-needs-trust-able-disbursement-compliance-desk/blueprint/" }, "trust": { "standards": [ "Fiduciary-side only — no beneficiary/family credentials ever collected", "Signed engagement letter with a defined scope-of-review", "100% analyst review before any delivery" ], "response_time": "Determination results within 1 business day of complete intake (rush: 4 hours)", "data_handling": "Per-fiduciary isolation, least-privilege access, encrypted storage, no secondary use of beneficiary data" }, "hook": "One disbursement decision can quietly suspend a beneficiary's SSI — most fiduciaries find out after, not before.", "sub_headline": "Send a proposed disbursement. Get back a cited Determination Memo, a Benefit-Impact Projection, and a Restructuring Alternative where relevant — never a rules engine to operate yourself.", "dream_outcome": "Every disbursement decision documented, defensible, and checked before it's approved — never discovered as a problem after the fact.", "specific_pains": [ "A direct rent or utility payment triggers SSA's ISM presumed-maximum-value reduction, cutting a beneficiary's SSI check by roughly a third", "Leaving a beneficiary holding more than $2,000 in countable resources at the wrong moment suspends SSI and creates a retroactive overpayment", "The ABLE Age Adjustment Act (effective Jan 1, 2026) just roughly doubled ABLE eligibility, and most fiduciaries haven't re-checked their caseload against it", "ABLE distributions must qualify as a qualified disability expense, a broad but not unlimited category, with a Medicaid-payback claim most trustees learn about too late", "No one at most fiduciary firms has benefits-compliance research as their actual full-time job" ], "cost_of_inaction": "A single miscategorized disbursement can suspend a beneficiary's SSI and create a retroactive overpayment SSA will pursue — and the fiduciary who approved it carries direct personal liability exposure for the mistake.", "mechanism": { "name": "The Disbursement Determination Cycle", "steps": [ { "title": "Intake", "body": "Disbursement request, receipts/invoices, and current beneficiary benefit-status confirmed complete before review begins." }, { "title": "Classify", "body": "Bright-Line Checks (resource limit, ISM formula, ABLE cap) run deterministically; the AI engine classifies against the current Rule Matrix and drafts a verdict with citations." }, { "title": "Review", "body": "A Benefits-Compliance Analyst reviews every draft — confirms, corrects, and checks Exception Queue escalation criteria." }, { "title": "Sign", "body": "Novel, high-value, low-confidence, or Certified Opinion cases route to a licensed elder-law attorney for sign-off." }, { "title": "Deliver", "body": "The formatted Determination Memo ships with full citations, Rule Matrix version, and a scope-of-review disclaimer, logged to the Audit Trail." } ] }, "offer_stack": [ { "item": "Determination Memo", "note": "dated, cited verdict — approve / approve-with-conditions / do-not-disburse" }, { "item": "Benefit-Impact Projection", "note": "the specific projected SSI/Medicaid dollar effect" }, { "item": "Restructuring Alternative", "note": "a compliant alternative path where the request as proposed is risky" }, { "item": "Certified Opinion", "note": "licensed elder-law attorney sign-off on novel/high-risk cases" }, { "item": "Quarterly Portfolio Review", "note": "full-portfolio drift check for retainer accounts" } ], "guarantee": "Standard determinations delivered within 1 business day (rush: 4 hours) — miss it, and the determination is free. Not a guarantee of any specific SSA or state Medicaid agency finding.", "urgency": "The founding-pilot cohort is capped at 5-8 professional fiduciary accounts at launch pricing — not artificial scarcity, an operational limit while the Rule Matrix and analyst bench are proven.", "objections": [ { "q": "Our elder-law attorney already handles this.", "a": "At trust drafting, yes — not on every one of dozens of monthly disbursement requests at a flat, non-hourly rate." }, { "q": "We can't afford it.", "a": "$45 per request, anchored below CPT Institute's own published fee precedent; the free check shows your own first request before you spend anything." }, { "q": "Is this allowed?", "a": "Fiduciary-side compliance research only, with a licensed attorney sign-off on anything novel or high-risk — never legal advice to a beneficiary." }, { "q": "What if we're managing dozens of trusts across several states?", "a": "That's exactly the founding-cohort profile — the Rule Matrix expands state by state as pilot fiduciaries onboard." } ], "who_this_is_not_for": [ "A family managing a single trust for one relative — too small a decision volume to justify a standing paid desk", "Anyone wanting BenefitGuard to serve as trustee or hold funds (this is a B2B compliance-research service, not a fiduciary-services product)", "A request already the subject of active SSA appeal or litigation — that stays with the fiduciary's own counsel", "Anyone needing individualized legal advice for a beneficiary or family directly — referred out, never performed by BenefitGuard" ], "proof_pillars": [ { "title": "Cited, not claimed", "body": "Every stat on this page cites a named, dated source (SSA POMS, the ABLE National Resource Center, a published fiduciary fee schedule) — never an invented statistic." }, { "title": "Analyst-approved, always", "body": "No Determination Memo ever ships without a named Benefits-Compliance Analyst's sign-off." }, { "title": "Honest about what's new", "body": "This is a new practice serving a growing, newly expanded population — pilot-cohort results will appear here as real determinations complete, not before." } ], "stakes_line": "Every disbursement a fiduciary approves without a documented compliance check is a decision they're personally exposed on if it turns out wrong — and most fiduciaries only find out after a redetermination notice arrives.", "deliverable": "A dated, cited Determination Memo, a Benefit-Impact Projection, a Restructuring Alternative where relevant, a Certified Opinion on demand, and a full Audit Trail retained for the engagement's defined period.", "unit_of_work": "one Disbursement Request — a single proposed trust or ABLE expenditure submitted for review", "lexicon": { "regulator": "SSA/Medicaid", "regulator_full": "Social Security Administration (SSI/POMS) and the applicable state Medicaid agency (state supplement rules)", "statute": "SSA POMS SI 01120.200 / SI 02301.205 / SI 01130.740; IRC §529A (ABLE)", "statute_frame": "cited by POMS section or code section, e.g. 'POMS SI 01120.200' or '26 U.S.C. §529A'", "persona": "Professional Fiduciary / Trust Officer", "persona_moment": "a beneficiary or family requests a disbursement and the fiduciary is unsure it's safe to approve", "trigger_moment": "a proposed disbursement is submitted for review", "enforcement_stakes": "SSI suspends and SSA pursues a retroactive overpayment — no appeal reverses a benefit month already lost", "retention": "renewal driven by the Portfolio Monitoring Retainer and Audit Trail lock-in as the fiduciary's trust portfolio grows", "cta_verb": "Get your free Check", "intake_checklist": [ "Disbursement description, amount, and payee", "Supporting receipt, invoice, or vendor quote", "Beneficiary's current SSI/SSDI/Medicaid enrollment status and state", "Trust type (third-party, first-party d4A, pooled d4C) or ABLE program enrollment", "Trust governing document or ABLE program terms (on file after first submission)" ], "regulator_faqs": [ { "q": "Does BenefitGuard ever act as our beneficiary's representative payee or SSA point of contact?", "a": "No — BenefitGuard never collects, stores, or requests a beneficiary's or family member's SSA or ABLE portal credential, and never communicates with SSA on the beneficiary's behalf. We work from fiduciary-authorized facts and documents only." }, { "q": "What happens if SSA later questions a determination?", "a": "A Benefits-Compliance Analyst and, where relevant, the signing attorney support the fiduciary with the full Audit Trail; if it escalates toward a formal dispute, we hand the fiduciary's counsel a complete, cited file and step back from legal advocacy." } ], "trust_standards_specific": [ "[PLACEHOLDER] owner to complete" ] }, "proof_angle": { "id": "diagnostic-teardown", "headline": "A rent-payment disbursement request found a $310/month ISM reduction before it was approved", "lever": "Anonymized diagnostic teardown", "outcome_verb": "quantified", "outcome_frame": "Benefit-Impact Projection with the exact dollar reduction and a Restructuring Alternative", "proof_promise": "Every number in your own Determination Memo traces to your own request facts and the current Rule Matrix — no aggregate industry statistic stands in for your beneficiary's actual situation." }, "indexable": false, "ubiquitousLanguage": { "audience": "Professional Fiduciary / Trust Officer", "domain": "Special needs trust & ABLE disbursement compliance", "deliverable": "Determination Memo", "reviewer": "Benefits-Compliance Analyst", "record": "Disbursement Request", "unit_of_work": "Disbursement Request", "cta_primary": "Get your free Disbursement Compliance Check", "cta_secondary": "View the full operating blueprint dossier", "regulator": "SSA / state Medicaid agency", "regulator_full": "Social Security Administration (SSI/POMS) and the applicable state Medicaid agency", "statute": "SSA POMS SI 01120.200 / SI 02301.205 / SI 01130.740; IRC §529A", "trigger_moment": "a beneficiary or family requests a trust or ABLE disbursement", "event_intake_started": "domain.diagnostic_requested", "event_conversation_requested": "ui.cta_primary" } }, "evidence": [ { "id": "special-needs-trust-able-disbursement-compliance-desk-ev1", "business_slug": "special-needs-trust-able-disbursement-compliance-desk", "area": "market", "claim_or_finding": "214,000+ ABLE accounts nationally as of Q2 2025 (213,855 exactly), up from 180,748 in Q2 2024 — 18.3% YoY growth, ~$2.7B combined assets", "status": "verified", "evidence": "NAST Q2 2025 National ABLE Data, via PlanSponsor", "verification_command": "Direct fetch of plansponsor.com's Q2 2025 ABLE coverage", "fix_owner": "content", "remediation": "n/a — verified, re-confirmed live during this build", "severity": "info" }, { "id": "special-needs-trust-able-disbursement-compliance-desk-ev2", "business_slug": "special-needs-trust-able-disbursement-compliance-desk", "area": "regulatory", "claim_or_finding": "ABLE Age Adjustment Act (SECURE 2.0) effective January 1, 2026, raises ABLE eligibility from disability-onset-before-26 to before-46", "status": "verified", "evidence": "ABLE National Resource Center official fact sheet", "verification_command": "Direct fetch of ablenrc.org's Age Adjustment Act fact sheet", "fix_owner": "content", "remediation": "n/a — verified, re-confirmed live during this build", "severity": "info" }, { "id": "special-needs-trust-able-disbursement-compliance-desk-ev3", "business_slug": "special-needs-trust-able-disbursement-compliance-desk", "area": "pain", "claim_or_finding": "SSI individual countable-resource limit is $2,000, unchanged for decades", "status": "verified, watch item found on re-check", "evidence": "SSA POMS SI 01120.200; SSA SSI Federal Payment Amounts for 2026 (confirms 2026 payment figures, not resource-limit itself, which is set separately and unchanged)", "verification_command": "Web search: SSI resource limit 2026 Social Security; fetch ssa.gov/oact/cola/SSI.html", "fix_owner": "content", "remediation": "See ev7 below — a pending bipartisan bill proposes raising this limit; not yet law", "severity": "low" }, { "id": "special-needs-trust-able-disbursement-compliance-desk-ev4", "business_slug": "special-needs-trust-able-disbursement-compliance-desk", "area": "pricing", "claim_or_finding": "Professional fiduciaries currently charge $50+/month per trust (incl. disbursement processing) plus a $35 rush/expedited fee per disbursement request", "status": "verified", "evidence": "CPT Institute Individual Special Needs Trust Pricing, 2026", "verification_command": "Direct fetch of cptinstitute.org/individual-special-needs-trust-pricing", "fix_owner": "content", "remediation": "n/a — re-confirmed live during this build; note the blueprint's separately cited '$75/hour extraordinary representative time' figure was not independently re-confirmed on the specific page fetched this run (see compliance-checklist.md)", "severity": "low" }, { "id": "special-needs-trust-able-disbursement-compliance-desk-ev5", "business_slug": "special-needs-trust-able-disbursement-compliance-desk", "area": "competitive", "claim_or_finding": "True Link Financial offers SNT-specific spending-control prepaid cards and a published attorney-facing compliance guide", "status": "verified", "evidence": "Direct review of truelinkfinancial.com product and compliance-guide pages", "verification_command": "Direct fetch of truelinkfinancial.com/compliance-guide/attorney-2025 and /family-trustee", "fix_owner": "content", "remediation": "n/a — verified, re-confirmed live during this build", "severity": "info" }, { "id": "special-needs-trust-able-disbursement-compliance-desk-ev6", "business_slug": "special-needs-trust-able-disbursement-compliance-desk", "area": "market", "claim_or_finding": "The pooled special needs trust sector spans dozens of nonprofit operators nationally (Arc chapters, Commonwealth Community Trust, and others)", "status": "verified (organizational form)", "evidence": "specialneedsanswers.com Directory of Pooled Trusts; The Arc, 'Pooled Trusts Operated by Chapters'", "verification_command": "Manual source check against the cited directories", "fix_owner": "content", "remediation": "n/a — verified", "severity": "info" }, { "id": "special-needs-trust-able-disbursement-compliance-desk-ev7", "business_slug": "special-needs-trust-able-disbursement-compliance-desk", "area": "regulatory", "claim_or_finding": "A bipartisan federal bill (SSI Savings Penalty Elimination Act, H.R.2540 / S.1234, 119th Congress 2025-2026) proposes raising the $2,000 SSI resource limit", "status": "new finding since blueprint generation", "evidence": "Congress.gov bill pages for H.R.2540 and S.1234; CNBC coverage, March 2026", "verification_command": "Web search: SSI Savings Penalty Elimination Act 2026 status passed Congress", "fix_owner": "content", "remediation": "Logged in compliance-checklist.md as a Rule Matrix change-monitoring item; the bill remains pending, not law, as of this build — does not change current compliance stakes", "severity": "medium" } ], "seo_pages": { "id": "seo-special-needs-trust-able-disbursement-compliance-desk", "business_slug": "special-needs-trust-able-disbursement-compliance-desk", "route": "/special-needs-trust-able-disbursement-compliance-desk", "title": "BenefitGuard — special needs trust & ABLE disbursement compliance desk for professional fiduciaries", "description": "BenefitGuard for professional fiduciaries administering special needs trusts and ABLE accounts. A dated, cited disbursement compliance determination within one business day — priced per request, never hourly, starting with a free Disbursement Compliance Check.", "canonical": "/special-needs-trust-able-disbursement-compliance-desk", "og_title": "BenefitGuard — Special Needs Trust & ABLE Disbursement Compliance Desk", "og_description": "BenefitGuard for professional fiduciaries administering special needs trusts and ABLE accounts. A dated, cited disbursement compliance determination within one business day — priced per request, never hourly.", "schema_type": "FAQPage", "schema_status": "pending-owner-facts", "sitemap_include": false, "noindex": true }, "vertical_style": { "accent": "ledger-amber", "signature": "Verdict strip with a citation chip beneath the active determination", "layout": "Disbursement compliance desk index", "anti": "Fintech alarm-red urgency banners, stock elderly/disability stock imagery, and generic SaaS gradient heroes" }, "canva": { "slug": "special-needs-trust-able-disbursement-compliance-desk", "territory": "Ledger Guardian", "family": { "id": "ledger-guardian-v1", "name": "Ledger Guardian", "motion": "minimal, functional only" }, "tokens": { "brand": "222 46% 24%", "brand-fg": "0 0% 100%", "surface": "45 22% 96%", "ink": "222 32% 12%", "muted": "222 12% 36%", "accent": "36 82% 32%" }, "type": { "display": "system-ui, 'Georgia', serif fallback for headings", "body": "system-ui, -apple-system, 'Segoe UI', sans-serif", "fonts_url": "none — system font stack only, no external font requests" }, "scale": { "h1": "40px/1.16", "h2": "28px/1.2", "h3": "20px/1.3", "body": "16px/1.6", "small": "13px/1.5", "tracking_display": "-0.01em", "tracking_body": "0", "weight_display": 600, "weight_body": 400 }, "spacing": { "card_padding": "20px", "card_radius": "8px", "card_shadow": "0 1px 2px rgba(21,20,15,0.08)", "section_gap": "56px", "hero_gap": "24px" }, "layout": { "hero": "left-aligned copy, right-aligned verdict-strip visual", "card": "bordered, low-shadow, ledger-style", "cta": "solid accent button, high contrast text", "archetype": "compliance determination desk", "card_silhouette": "rectangular with an 8px radius, thin border, no heavy shadow", "button_geometry": "rectangular with a 6px radius, no pill shapes" }, "background": { "hero_gradient": "none — flat brand surface with a subtle radial tint", "cta_gradient": "none — flat accent surface", "section_wash": "alternating surface / surface-raised bands" }, "motif": "Verdict strip motif with a citation chip beneath the active determination" }, "capabilities": { "marketing": true, "portal": false, "ops": true, "chatbot": false, "payments": false }, "generated_at": "2026-07-13T12:15:00Z", "ddd_coverage": { "slug": "special-needs-trust-able-disbursement-compliance-desk", "total": 17, "passed": 17, "pct": 100, "checks": [ { "key": "subdomains", "label": "Subdomains identified", "count": 5, "min": 3, "ok": true, "gate": "domain-modeling", "unblock": "n/a" }, { "key": "bounded_contexts", "label": "Bounded contexts defined", "count": 5, "min": 3, "ok": true, "gate": "domain-modeling", "unblock": "n/a" }, { "key": "aggregates", "label": "Aggregates named", "count": 5, "min": 3, "ok": true, "gate": "domain-modeling", "unblock": "n/a" }, { "key": "domain_events", "label": "Domain events named", "count": 6, "min": 4, "ok": true, "gate": "domain-modeling", "unblock": "n/a" }, { "key": "commands", "label": "Commands named", "count": 6, "min": 4, "ok": true, "gate": "domain-modeling", "unblock": "n/a" }, { "key": "policies", "label": "Policies defined", "count": 4, "min": 2, "ok": true, "gate": "domain-modeling", "unblock": "n/a" }, { "key": "ai_agents", "label": "AI agents specified", "count": 5, "min": 3, "ok": true, "gate": "ai-governance", "unblock": "n/a" }, { "key": "invariants", "label": "Invariants enforced", "count": 5, "min": 3, "ok": true, "gate": "domain-modeling", "unblock": "n/a" }, { "key": "risk_register", "label": "Risk register entries", "count": 3, "min": 3, "ok": true, "gate": "risk", "unblock": "n/a" }, { "key": "adrs", "label": "ADRs recorded", "count": 2, "min": 1, "ok": true, "gate": "architecture", "unblock": "n/a" }, { "key": "use_cases", "label": "Use cases documented", "count": 2, "min": 2, "ok": true, "gate": "domain-modeling", "unblock": "n/a" }, { "key": "human_roles", "label": "Human roles defined", "count": 3, "min": 2, "ok": true, "gate": "ai-governance", "unblock": "n/a" }, { "key": "human_review_checkpoints", "label": "Human review checkpoints", "count": 4, "min": 2, "ok": true, "gate": "ai-governance", "unblock": "n/a" }, { "key": "data_objects", "label": "Data objects owned", "count": 3, "min": 2, "ok": true, "gate": "data", "unblock": "n/a" }, { "key": "external_integrations", "label": "External integrations mapped", "count": 4, "min": 2, "ok": true, "gate": "architecture", "unblock": "n/a" }, { "key": "security_controls", "label": "Security controls specified", "count": 3, "min": 2, "ok": true, "gate": "security", "unblock": "n/a" }, { "key": "mvp_roadmap", "label": "MVP roadmap phases", "count": 1, "min": 1, "ok": true, "gate": "roadmap", "unblock": "n/a" } ], "failingGates": [] }, "checksum": "sha256:8f2c4b1e6a3d0e9c7b5f4a2d1c8e6b3f0a9d7c5e2b1f8a4d6c3e0b9f7a5d2c1e" }