Title

BiasGate Clear is a done-for-you Multi-State AI-Hiring Bias Audit & Compliance Desk for mid-market U.S. employers (roughly 500–5,000 employees, multi-location or multi-state) that use one or more third-party AI-enabled hiring tools — resume screeners, automated video-interview scoring, chatbot pre-screens, algorithmic candidate ranking — and must now navigate a fast-diverging patchwork of state and city AI-employment laws with no federal floor to fall back on, no in-house AI-governance counsel, and rising litigation exposure now that a federal court has confirmed algorithmic hiring decisions are not shielded from discrimination liability.

Final Decision

FINAL DECISION: BLUEPRINT

This candidate clears the evidence threshold and all six gates, with one gate (Sam Altman test) scored a qualified 4/5 rather than 5/5 for reasons made explicit in the Rubric Scorecard and Anti-Commoditization Analysis below. A full build follows.

Executive Summary

Since early 2023, U.S. employers who use AI in hiring have gone from operating under a single (if thin) federal EEOC guidance framework to operating under no federal guidance at all and a rapidly multiplying set of independent state and city rules that do not track each other. New York City's Local Law 144 has required an annual independent bias audit and public posting of results since July 2023, with violations fined $500–$1,500 each and every day of continued non-compliance treated as a separate violation. California's Civil Rights Council automated-decision-system employment regulations took effect October 1, 2025. Illinois amended its Human Rights Act (HB 3773) to prohibit discriminatory AI use in employment decisions and require employee notice, effective January 1, 2026. Colorado's AI Act was signed, then amended, then had its effective date pushed to June 2026 with narrowed obligations — itself a compliance-tracking burden, since employers preparing for the original date had to re-scope their programs mid-build. Meanwhile, in early 2025 the EEOC quietly removed its own AI-in-hiring technical guidance under the new administration, and multiple national employment-law firms (K&L Gates, Cooley, Husch Blackwell) confirmed the federal floor is gone even though the underlying discrimination statutes (Title VII, the ADEA) still apply in full — a void several states are now filling with their own, non-uniform rules.

The litigation signal sharpened materially in 2026. On March 6, 2026, a federal judge in the Northern District of California denied Workday's motion to dismiss in Mobley v. Workday, ruling that the 1967 Age Discrimination in Employment Act applies to algorithmic hiring systems and rejecting the argument that only a human employer, not its AI vendor's tool, can be held to account. Workday's own filings acknowledged its software processed applications "numbering in the billions" during the relevant period, and roughly 14,000 workers opted into the resulting nationwide collective action by the March 7, 2026 deadline. The clear takeaway employment counsel is now giving clients: using a vendor's AI tool does not transfer legal responsibility for its discriminatory effects away from the employer that deployed it.

Demand for AI hiring tools keeps climbing even as the compliance floor gets more fragmented: compiled industry survey data puts AI-hiring-tool adoption at 87% of companies overall and 99% of Fortune 500 firms, with 93% of recruiters planning to increase AI use in 2026 (treated here as Inferred rather than Verified, given its secondary-compilation sourcing; see Claim Table). A recognizable specialist market has already formed around NYC LL144 compliance specifically — named vendors include Warden AI, VerityAI, BiasAuditUSA, AEDTAudits.com, VerifyWise, FairNow, Classet, and DeepInspect, several of which gate pricing behind a sales demo and most of which scope their service to the single, oldest law (NYC) rather than the full, growing multi-state matrix. That existing, named competitive set is treated in this run as budget-validation evidence (buyers are already paying for exactly this category of service) and as the basis for this business's differentiation: a flat-fee, per-tool, multi-state Compliance Desk built for the mid-market employer who is not large enough to negotiate a custom enterprise platform contract and not small enough to ignore four-state exposure, priced and packaged so a single NYC-only audit purchase naturally upgrades into ongoing multi-state tracking as new state laws (Colorado in 2026, and others expected to follow) come online.

Thesis

AI-hiring compliance is not one law to satisfy once; it is a moving multi-jurisdiction matrix (NYC since 2023, California since October 2025, Illinois since January 2026, Colorado from mid-2026, with more states drafting bills) that a mid-market, multi-state employer's own HR team has neither the legal training nor the bandwidth to track continuously, and that federal deregulation in 2025 made worse, not better, by removing the one guidance document that offered a single interpretive anchor. The underlying audit work — inventorying which HR-tech tools meet each jurisdiction's statutory definition of a covered tool, calculating adverse-impact/four-fifths ratios from vendor-supplied scoring data, drafting the specific notices and public postings each law requires, and tracking the annual/ongoing renewal cadence — is exactly the kind of document-and-data-heavy, deadline-driven, judgment-at-the-chokepoints work this factory targets: independence from the employer is a statutory requirement (not a nice-to-have), which structurally forces outsourcing rather than in-house self-audit; the market is real and already spending money on it, evidenced by a dozen-plus named competitors; and litigation risk, freshly sharpened by the March 2026 Mobley ruling, gives buyers an urgent, board-visible reason to act now rather than defer. The candidate does not score a perfect six-gate rubric — the "Sam Altman test" is qualified because the core statistical methodology (adverse-impact-ratio calculation under the four-fifths rule) is fixed math that frontier models do not materially improve, even though everything around that calculation (legal-text monitoring across a growing number of states, notice drafting, evidence-packet assembly) gets faster and cheaper as models improve. That qualification is treated honestly rather than smoothed over, consistent with the evidence-threshold requirement that a business clear the bar strongly rather than just barely.

Discovery Rationale

This run began by cloning ainbis_repo fresh, confirming manifest.json was healthy (517 prior runs, no truncation), and reading the project's own manifest-truncation-incident record, which documents that a manifest-truncating process external to this repository fires hourly near :00 and that a GitHub Actions auto-heal workflow plus a schedule shift for this factory's own trigger has mitigated (not eliminated) the risk — the standing instruction to re-verify row count immediately before the final push was followed. A keyword sweep of all 517 prior candidate names and market descriptions found the manifest extremely dense (200+ entries) in regulatory-filing/"completeness desk" businesses concentrated in tax/accounting, legal-compliance-filing, healthcare, real estate/HOA, and trade-compliance/customs verticals; several fresh ideas generated independently this run — self-storage lien-sale compliance, U.S. import HTS classification, duty drawback recovery, TTB/craft-beverage compliance, I-9/E-Verify audit services — were each confirmed via manifest search to already exist as prior entries (U.S. import HTS classification & duty-exposure services, U.S. employer I-9 audit, remediation, and worksite-enforcement compliance services, two separate craft-beverage TTB blueprints, an ocean-freight demurrage/detention recovery desk, and a Foreign-Trade Zone compliance blueprint among them) and were rejected on that basis. Self-storage lien-sale compliance was independently researched and separately rejected: a venture-funded, PMS-integrated software incumbent (Ai Lean, which raised $1.9M specifically for this niche and is a marketplace partner of SiteLink and Storable) already owns this space with a self-serve software product, making a new AI-native service entrant a weak differentiation case. Veterinary DEA controlled-substance compliance was researched and found real but already served by a mix of software (VetSnap, Cubex) and boutique DEA-compliance consultants (Titan Group DEA, EAS Consulting), with a smaller and less time-pressured evidence base than the eventual winner. Multi-state non-compete/restrictive-covenant compliance was researched and found to have genuine whitespace (no dedicated AI-native vendor found; the field is served by traditional billed-hourly employment law firms) but was set aside because the core deliverable is legal drafting and enforceability judgment on binding contracts across divergent state law — a materially higher unauthorized-practice-of-law exposure than this factory's other document-completeness and audit-production patterns, and not clearly separable into a safe non-attorney chokepoint without a licensed-attorney-of-record structure this run could not evidence as operationally ready. The AI-hiring bias-audit and multi-state compliance desk candidate was selected after confirming, via targeted keyword search, that no prior manifest entry addresses AEDT/algorithmic-hiring bias audits, NYC Local Law 144, or the state AI-employment-law patchwork, and after building an evidence base strong enough, and current enough (a March 2026 federal ruling), to clear the six gates.

Candidate Comparison

CODE Validation

Consumer/buyer trend: Employer adoption of AI hiring tools has become the default, not the exception (87% of companies per compiled survey data, 99% of Fortune 500), while the regulatory environment governing that use inverted in the opposite direction in 2025–2026 — federal guidance disappeared and state/city law multiplied instead.

Opportunity: The specific underserved problem is translation and continuity: no single employer-facing playbook currently tells a multi-state, mid-market HR team which of its specific AI-enabled tools trigger which specific jurisdiction's audit/notice/posting/retention obligations, on what cadence, and existing vendors are split between enterprise-only platforms with demo-gated pricing and single-jurisdiction (NYC-only) boutique auditors — leaving the multi-state mid-market employer with no clean off-the-shelf option.

Demand: Direct, named evidence of buyers already spending on this exact problem: at least eight identifiable vendors (Warden AI, VerityAI, BiasAuditUSA, AEDTAudits.com, VerifyWise, FairNow, Classet, DeepInspect) are actively selling NYC LL144 bias-audit or related compliance services today, several with case studies, a G2 review category ("AI Bias Audit Services") exists, SHRM has published buyer-facing guidance ("AI Bias Audits Are Coming. Are You Ready?"), and law firms are actively fielding client questions (Bloomberg Law: "AI Hiring Compliance Is a Patchwork and Leaves Big Employer Gaps"; DarrowEverett's 2026 legal-analysis update). The NY State Comptroller's own December 2025 audit of LL144 enforcement found weak agency follow-through to date, which is itself an evidentiary signal of under-compliance risk sitting on employers' books right now, likely to surface as the city sharpens enforcement.

Economic Sizing: A precise, single-number TAM is not available from public data and is not asserted; instead this run builds a bounded, explicitly uncertain range (see Market and Demand Evidence). The credible planning range is 15,000–40,000 U.S. employers with 500+ employees and confirmed multi-state operations in at least two of the four regulated jurisdictions, of which a meaningful share (Inferred, not directly counted) use at least one AI-enabled hiring tool given 87% overall adoption; at a blended $12,000–$35,000/year Compliance Desk price point (see Pricing), even a low single-digit percentage penetration of that range supports a real business, and the range itself is wide enough that this run treats the resulting revenue ceiling as Inferred, not Verified.

Rubric Scorecard

Total: 26/30. A strong, not perfect, clear.

Target Buyer

Jobs-to-be-Done

The functional job is: "know, for every AI-enabled hiring tool I use and every state or city I operate in, exactly what I am legally required to do, by when, and be able to produce evidence that I did it." The emotional job is: "stop being surprised by a new state law or a plaintiff's demand letter that assumes I already had a compliance program I didn't know I needed." The social job is: "be able to tell my CEO and board, in one sentence, that AI-hiring legal exposure is handled by a named, credentialed, independent party — not something HR is guessing at."

The Painful Problem

A mid-market employer operating in New York City, California, Illinois, and preparing for Colorado does not face one compliance obligation; it faces four independently drafted, non-uniform ones, layered onto a federal backdrop that got materially less clear in 2025 when the EEOC's own AI-hiring guidance was removed. NYC requires an annual independent bias audit before continued use, a specific 10-business-day candidate notice, and public posting of the audit summary, with $500–$1,500 per-violation fines that accrue daily. California's regulations, effective October 1, 2025, add automated-decision-system-specific recordkeeping and anti-bias obligations under FEHA. Illinois, effective January 1, 2026, prohibits discriminatory AI use in employment decisions and adds an employee-notice requirement under a different statutory mechanism than New York's. Colorado's AI Act, originally slated for February 2026, was amended and its effective date pushed to June 2026 with narrowed obligations — meaning any employer that had already begun building a Colorado-specific compliance program had to re-scope it mid-stream. No single internal HR or legal generalist can reliably track four independently moving statutory clocks, translate each into tool-specific obligations, and produce audit-grade documentary evidence, and the cost of getting it wrong is no longer hypothetical: a federal judge's March 2026 refusal to dismiss Mobley v. Workday confirmed that deploying a vendor's AI tool does not insulate the employer from age-discrimination liability, and roughly 14,000 workers opted into that collective action in a single month.

The Outcome We Sell

BiasGate Clear sells a completed, evidence-ready compliance posture, not a dashboard the client must operate: a documented inventory of every AI-enabled hiring tool in use mapped against every jurisdiction the employer operates in; a signed, independent bias-audit report for each covered tool, calculated using the EEOC four-fifths adverse-impact-ratio methodology against vendor-supplied scoring and applicant demographic data; drafted, ready-to-publish or ready-to-send jurisdiction-specific notices and public postings; and a maintained renewal calendar that automatically resurfaces the client's obligations as new state laws take effect or existing ones change (as Colorado's did). If a regulator inquiry, EEOC charge, or plaintiff's demand letter ever arrives, the client receives a pre-assembled, counsel-ready evidence packet rather than starting from a blank page under time pressure.

First One-Feature MVP Wedge

Evidence Summary

The strongest evidence in this run is legal and current: four independently sourced, cross-checked confirmations that NYC (2023), California (Oct. 1, 2025), and Illinois (Jan. 1, 2026) AI-hiring rules are live, that Colorado's is delayed to June 2026, and that federal EEOC guidance was removed in early 2025 (K&L Gates, Cooley, Husch Blackwell, and NatLawReview all independently cover the rescission). The litigation catalyst — the March 6, 2026 Mobley v. Workday ruling and ~14,000-worker opt-in — is corroborated across Forbes, Akin Gump's AI law tracker, and Wiggins Childs' plaintiff-side case-update page. Competitor/budget evidence is direct: at least eight named vendors are actively selling into this exact problem today. The weakest evidence is market sizing (no single authoritative count of "mid-market multi-state AI-hiring employers" exists) and pricing (most competitors gate pricing behind a sales call), both of which are treated as Inferred/Unverified with explicit ranges rather than asserted as precise.

Claim Table (Verified / Inferred / Unverified)

Source-Claim Matrix

Market and Demand Evidence

No single public data source counts "mid-market, multi-state U.S. employers using AI hiring tools," so this run builds and clearly labels a bounded range rather than asserting a precise TAM. The U.S. Small Business Administration's 2025 Small Business Profile confirms roughly 36.2 million U.S. small businesses (defined as under 500 employees), implicitly placing the "large employer" (500+) population as a distinct, much smaller segment that SBA's own public profile does not enumerate precisely in the material reviewed this run; a widely-cited secondary approximation places U.S. employer firms with 500+ employees in the low tens of thousands, but this run did not locate a primary Census SUSB table confirming an exact figure and does not assert one. Applying the compiled 87% AI-hiring-tool adoption figure (Inferred, not Verified) to any plausible large-employer population, combined with a requirement of operations in 2+ of the four regulated jurisdictions, this run's own bounded estimate is 15,000–40,000 employers as a near-term addressable population — presented explicitly as Unverified. Demand-side confirmation is stronger than supply-side counting: a dozen-plus named, operating vendors (see Competitive Landscape) already sell into this exact buyer, a G2 review category exists, and SHRM (the largest U.S. HR professional association) has published buyer-facing "get ready" guidance, all of which is direct evidence that real budget is already moving toward this category, independent of this run's own TAM uncertainty.

Active Buyer Conversations

Evidence of live buyer-side conversation includes SHRM's own member-facing article "AI Bias Audits Are Coming. Are You Ready?", Bloomberg Law's practitioner-oriented piece "AI Hiring Compliance Is a Patchwork and Leaves Big Employer Gaps" (explicitly framing the multi-state gap this business is built to close), DarrowEverett's 2026 "AI Hiring & Workforce Management" legal-analysis update aimed at in-house counsel, and law-firm client alerts from Akerman, Ogletree, Seyfarth Shaw, and Littler all published in 2025–2026 specifically to answer employer questions about this exact patchwork — the volume and cadence of this coverage (new client alerts appearing through mid-2026 as Colorado's law was amended twice) is itself evidence that buyers are actively asking their outside counsel what to do, not that the problem is settled or well understood internally.

Competitive Landscape

The pattern across the named field: vendors split between enterprise platforms with custom, demo-gated pricing, and single-jurisdiction (almost always NYC-only) boutique auditors. None of the reviewed competitors was found to explicitly package an ongoing, flat-fee, multi-state (NYC + CA + IL + CO-ready) Compliance Desk purpose-built for the 500–5,000-employee segment — that gap is this business's wedge.

Competitor and Budget Validation

Existing budget for this problem is not hypothetical: employers already pay outside employment counsel by the hour to interpret each new state law (evidenced by the volume of 2025–2026 client-alert publishing from Akerman, Ogletree, Seyfarth, Littler, Jackson Lewis, and others), and a dozen-plus named specialist vendors are actively transacting NYC LL144 audits today. This is not a "no competitors, therefore blue ocean" pitch; it is the opposite — competitor density validates the spend exists, and the wedge is packaging (multi-state, flat-fee, mid-market-sized) rather than inventing demand from nothing. The service does not attempt to replace outside employment counsel; it is structured to reduce the volume of hourly-billed counsel time spent on routine tracking and document production, while explicitly routing genuine legal-interpretation questions (e.g., "is this specific tool a violation") back to the client's own counsel (see Licensing Boundary).

Pricing Evidence and Proposed Pricing

Direct competitor pricing is largely demo-gated and not publicly disclosed (labeled Unverified in the Claim Table); the closest public signal is qualitative "flat-fee" or "transparent flat-fee pricing" language from BiasAuditUSA, without dollar figures. Proposed pricing here is therefore Inferred, benchmarked against comparable B2B compliance-audit price points documented elsewhere in this factory's own manifest (compliance "completeness desk" services generally price $2,000–$15,000 per unit/engagement) and against general market comparables such as SOC 2 Type I/II audit engagements, which commonly run $15,000–$60,000. Pricing is per-unit (per tool audited, per jurisdiction covered), never hourly, per this factory's operating rule.

Regulatory and Compliance Considerations

Four live/imminent regimes govern this business's own subject matter: NYC Local Law 144 (in effect since July 2023; independent audit, 10-business-day candidate notice, public posting; enforced by the NYC Department of Consumer and Worker Protection); California's Civil Rights Council automated-decision-system employment regulations (effective October 1, 2025, under FEHA); Illinois HB 3773 amending the Illinois Human Rights Act (effective January 1, 2026; discriminatory-AI-use prohibition plus employee-notice requirement); and Colorado's AI Act (delayed to June 2026, narrowed obligations, still evolving). The federal backdrop — Title VII and the ADEA — remains fully in force regardless of the EEOC's 2025 guidance withdrawal, as confirmed by Mobley v. Workday. The business itself must track and re-scope its own service delivery every time one of these four laws changes (as Colorado's already has twice), which is precisely the ongoing-tracking value proposition it sells to clients.

Licensing Boundary

No single professional license governs "conducting a bias audit" under NYC LL144 as written — the statute requires independence, not a specific credential — but industry practice, and this business's own risk posture, requires every final adverse-impact-ratio report to be signed by a qualified, credentialed, independent reviewer: an Industrial-Organizational (I-O) psychologist (ideally SIOP-affiliated) or a professional statistician, engaged as a contractor whose compensation is structurally independent of any individual audit's outcome, consistent with the statutory independence requirement. AI may draft the tool inventory, jurisdiction mapping, notice language, and the underlying adverse-impact-ratio data tables; trained (non-licensed) operators may collect client data, populate validated calculation templates, and assemble client-facing deliverables; the credentialed independent reviewer must review and sign every audit report before delivery. The business must never render a legal opinion on whether a specific tool or practice violates Title VII, the ADEA, or a state discrimination statute — that determination is reserved to the client's own retained employment counsel, and every deliverable carries an explicit disclaimer to that effect plus a standing referral relationship to outside counsel for escalations. Required audit-trail elements: dated engagement letter establishing auditor independence, documented data sources and calculation methodology, named signing reviewer with disclosed credentials, and a version-controlled record of every notice/posting delivered to the client. Given the subject matter, the business must also model and disclose its own AI-tool usage transparently, since a compliance vendor discovered to be non-transparent about its own AI use would face acute reputational risk.

AI-Native Advantage

The advantage is not "uses ChatGPT to write reports." It is: (1) continuous multi-jurisdiction legal-text monitoring that flags a statutory change (such as Colorado's 2026 amendment) within days rather than at the next quarterly counsel check-in; (2) automated ingestion and normalization of vendor-supplied scoring/demographic data across heterogeneous HR-tech export formats, which previously required a consultant to manually reformat spreadsheets; (3) instant drafting of jurisdiction-specific notice and posting language from a maintained template library that updates the moment a law changes; and (4) evidence-packet assembly on demand, collapsing what would be days of paralegal/consultant time into a same-day turnaround when a client receives a demand letter. None of this replaces the credentialed independent reviewer's sign-off, which remains the trust-bearing chokepoint.

Internal AI Engine Architecture

AI-vs-Human Operations Pipeline

Dynasty Translation Layer

Buyer translation: A CHRO/VP Talent Acquisition who needs to tell their board, in one sentence, that AI-hiring legal exposure is handled. Service translation: a done-for-you Compliance Desk delivering signed audit reports, notices, and a renewal calendar — not a dashboard the client operates. Workflow translation: intake questionnaire → AI normalization & drafting → deterministic rule checks → independent human sign-off → QA → delivery → renewal-calendar follow-up. Tooling translation: a maintained legal-text knowledge base, a data-normalization pipeline, a four-fifths-rule calculation template, a client portal, and a CRM-driven renewal calendar — favoring available tooling (spreadsheet-based calculation templates, standard document generation, a lightweight portal) over custom software at launch. Sales translation: "You're using AI to hire. Four states now regulate that, and a federal court just ruled you can't blame the vendor. We'll tell you exactly what you're exposed to, free, in one scan." Delivery translation: manual-first for the first cohort (founder + contracted reviewer), templated calculation and drafting tooling built incrementally as patterns repeat. Expansion translation: from single-audit engagements to the Compliance Desk subscription, to a template/playbook library sold to HR-tech vendors themselves who want to help their own customers stay compliant, to eventual software-assisted self-serve tiers for the smallest qualifying employers once the manual playbook is fully hardened.

Anti-Duplication Analysis

Similar services/tools that exist: at least eight named vendors (see Competitive Landscape) sell NYC LL144 bias audits or adjacent AI-hiring-governance software today. This is not a claim of "no competitors" — it is explicitly the opposite. The differentiation is narrow and specific: (1) scope — a genuinely multi-state (not NYC-only) compliance desk that tracks California, Illinois, and Colorado alongside New York as a single bundled service, re-scoping automatically as laws change; (2) buyer segment — purpose-built and priced for the 500–5,000-employee mid-market buyer, between the enterprise-only demo-gated platforms and the smallest solo-practitioner NYC auditors; (3) delivery model — done-for-you, not a self-serve dashboard the client's own HR team must learn to operate; and (4) continuity — a maintained renewal calendar and litigation-readiness evidence packet as standing features, not one-off deliverables. What existing tools leave unsolved: the multi-state tracking burden itself, and the gap between "enterprise platform with a six-figure contract" and "solo NYC auditor with no multi-state capability" that leaves the mid-market buyer underserved.

Anti-Commoditization Analysis

If future general-purpose AI models become capable enough that any HR team could self-serve a compliant bias audit end to end, the parts of this business that survive are: the statutorily required independence of the human signer (a model cannot satisfy the "independent auditor" requirement itself, regardless of capability), the maintained, continuously updated multi-state legal-text knowledge base (a moving target that requires ongoing human legal review even if AI drafts the first pass), and the accumulated template/playbook library built from real client edge cases across four jurisdictions. The core commoditization risk is the adverse-impact-ratio calculation itself, which is simple, well-defined statistics that a capable model (or even a spreadsheet) already handles adequately today — meaning this business's defensibility rests more on regulatory-independence requirements and multi-state tracking continuity than on any proprietary AI capability, which is exactly why the Sam Altman gate was scored 4/5 rather than 5/5 rather than overstating the moat.

Service Delivery Workflow

Intake (HR-tech stack, states, hiring volume, demographic data availability) → jurisdiction and tool-coverage mapping → vendor data collection (scoring outputs, applicant demographics where available) → AI-assisted data normalization → four-fifths-rule adverse-impact-ratio calculation → draft audit report, notices, and postings → independent credentialed reviewer sign-off → QA completeness check → client delivery via secure portal → renewal-calendar entry created for next required action.

Operations as Product

SOPs: a documented intake checklist per client covering every HR-tech tool and jurisdiction; a required-evidence list for each jurisdiction's specific audit/notice/posting obligations; an automated completeness check before any deliverable reaches the reviewer; an exception queue for edge-case tool classifications; confidence scoring on data-quality inputs (flagging small-sample or malformed vendor exports before calculation); a full audit trail from raw vendor data through final signed report; version-controlled template library for notices/postings per jurisdiction; gold-standard example reports maintained per jurisdiction; a red-team review pass on new-jurisdiction templates before first client use; and a root-cause postmortem on any missed deadline or reviewer-flagged error, feeding back into the SOP and template library.

No-Holes Quality Engine

Every deliverable passes three gates before client delivery: (1) an automated completeness check confirming every required jurisdiction-specific element is present (e.g., LL144's specific notice-content requirements, or Illinois' specific employee-notice elements); (2) the independent credentialed reviewer's substantive sign-off on the statistical calculation and classification judgment calls; (3) a second-pass QA spot-check, weighted toward new-jurisdiction or first-time-client deliverables where template maturity is lowest. Any client-reported discrepancy triggers a documented root-cause review before the same error type can recur across other clients.

What the Human Expert Actually Does

Minimum Viable Offer

The MVO is the free/low-cost "AI-Hiring Exposure & Audit-Readiness Scan" (see MVP Wedge) converting into a paid Single-Tool NYC Audit, with the Multi-State Compliance Desk offered as the natural next-step upsell once the first jurisdiction's audit is delivered and trust is established.

Fulfillment Process

The first 3 customers are fulfilled largely manually: founder runs intake calls personally, a contracted independent I-O psychologist or statistician performs and signs each calculation, and notice/posting drafts are produced from a small initial template library built specifically for those first engagements. Automatable from day one: data normalization formatting and first-draft notice language. Not automated at first: the classification judgment call on ambiguous tools, and all reviewer sign-offs. The first paid offer is the Single-Tool NYC Audit; it evolves into the Multi-State Compliance Desk as the template library and rule engine mature across the first 10–20 engagements.

Tools and Systems

Launch stack favors available tooling over custom builds: a structured intake form (Typeform/Tally-class tool), a spreadsheet-based four-fifths-rule calculation template validated against published methodology, a document-generation workflow for notices/postings, a lightweight client portal for secure delivery, and a CRM (e.g., HubSpot-class) driving the renewal calendar. Custom software (a purpose-built compliance-tracking platform) is deferred until manual-process patterns are proven across 20+ engagements, consistent with the factory's service-first requirement.

Human-in-the-Loop Quality Control

Every audit report requires the independent credentialed reviewer's signature before delivery; no report leaves the business without it. A second-pass QA reviewer spot-checks a sampling of deliverables, weighted toward new template types. Client-reported errors trigger mandatory root-cause review before the underlying template or rule is reused.

Nonlinear Scaling and Unit Economics

COGS breakdown: model inference (data normalization, drafting) ~8%; independent reviewer contractor fees ~25%; QA/second-review ~7%; hosting/software ~4%; support ~5%; rework/escalation reserve ~4%; sales follow-up ~6%. Automation percentage climbs as the template library and rule engine absorb more of the drafting and classification workload, while the reviewer sign-off line remains a largely fixed per-unit human cost by regulatory design.

Distribution Proof Table

Sales and Outreach Plan

Lead with the free Exposure & Audit-Readiness Scan as the primary top-of-funnel motion; qualify leads on confirmed multi-state footprint and named AI hiring tool usage before founder-led consultative calls; close the Single-Tool NYC Audit as the first transaction, with the Multi-State Compliance Desk positioned explicitly as the next step at delivery of the first audit rather than sold cold.

Founder-Led Content Plan

Founder-authored content teaches the exact pain: what an AEDT is (and isn't) under each of the four laws, the real cost of a missed LL144 deadline, what the Mobley ruling actually changes for employers (not just vendors), and a running "state law tracker" post updated every time one of the four regimes changes — positioning the founder as the person already doing the multi-state tracking work publicly, which is the same work being sold privately.

First 30 Days of Content

10 educational posts: (1) What counts as an AEDT under NYC LL144; (2) California's Oct. 2025 ADS regs in plain English; (3) Illinois HB 3773: what changed Jan. 1, 2026; (4) Colorado's AI Act: why it's delayed and what that means for you; (5) What Mobley v. Workday actually ruled, for non-lawyers; (6) The four-fifths rule, explained without the statistics jargon; (7) Why the EEOC's 2025 guidance removal doesn't mean AI hiring is unregulated; (8) A checklist: does your ATS/screening tool trigger any of these four laws?; (9) What happens if you miss your LL144 renewal; (10) How other states are drafting similar bills right now.
3 diagnostic teardown formats: "We ran the Exposure Scan on a real (anonymized) mid-market employer — here's what it found" x3, one per industry vertical (retail, healthcare staffing, hospitality).
2 lead-magnet angles: "The 4-State AI Hiring Compliance Checklist" (PDF); "Is your AI hiring tool an AEDT? 5-minute self-assessment."
1 webinar: "Mobley v. Workday and the New AI-Hiring Compliance Reality: What Mid-Market Employers Need to Do in the Next 90 Days."
1 outbound diagnosis template: Personalized note referencing the prospect's specific state footprint and named HR-tech tool, offering the free scan with a specific, plausible exposure finding pre-filled.

Lead Magnet and Waitlist Plan

Primary lead magnet: the free AI-Hiring Exposure & Audit-Readiness Scan itself. Secondary: the 4-State Compliance Checklist PDF. Both feed a qualification sequence (confirmed multi-state footprint + named tool usage) before a founder-led consultative call; waitlist signups alone are not treated as validated demand (see Pilot Design).

Warm GTM Plan

Founder's existing HR-tech and employment-law network is the first outreach list; existing contacts at mid-market HR-tech vendors (who want their own customers compliant, protecting the vendor's own reputational exposure) are a natural warm-referral channel, alongside any existing relationships with employment-law firms fielding client questions they cannot operationally fulfill at scale.

Targeted Outbound Plan

Build a list of 500-5,000-employee multi-state employers with public job postings referencing named AI-enabled ATS/screening tools (LinkedIn Sales Navigator + job-posting scraping for tool-name mentions), cross-reference against confirmed operations in 2+ of the four regulated states, and lead outbound with a specific, personalized exposure diagnosis rather than a generic compliance pitch.

Answer-Engine/Search Visibility Plan

Publish structured, citation-worthy explainer content on each of the four statutes and the Mobley ruling, formatted for AI-answer-engine extraction (clear headers, direct Q&A framing, dated updates), positioning the business's own tracker page as the canonical, continuously updated multi-state reference — the same content asset both ranks organically and gets cited by AI answer engines responding to "is my AI hiring tool compliant" queries.

Pilot Design and Early-Demand-Trap Mitigation

First pilot cohort: 5 employers, capped, prioritizing those with confirmed NYC + at least one other regulated-state footprint. Early-access incentive: discounted first-year Compliance Desk pricing in exchange for a case-study reference. Feedback mechanism: structured post-delivery interview distinguishing genuine product feedback (e.g., "the scan didn't catch our chatbot screener") from one-off custom requests (e.g., a client-specific integration ask). Waitlist and scan-signup volume are explicitly not treated as product-market-fit signals; only paid conversion and 12-month renewal are.

Early-Access Feedback Flywheel

Every reviewer correction, every client-reported gap, and every new-jurisdiction template becomes a versioned addition to the SOP and rule-engine library; corrections that recur across 2+ clients are hardened into a permanent rule rather than handled ad hoc each time.

Build-Before-Scale Checkpoints

After 5 pilots: harden intake questionnaire and required-evidence checklist. After 10 pilots: harden SOPs, exception-queue routing, and reviewer-assignment logic. After 20 pilots: pause new pilot intake until COGS, rework rate, escalation rate, and cycle time are actually measured against target, not assumed; manual workarounds acceptable temporarily include founder-run intake calls and ad hoc template drafting, but a rising reviewer-bandwidth bottleneck or a rework rate persistently above 5% signals the model is not yet scalable and pilots should pause regardless of pipeline pressure.

7-Day / 30-Day / 90-Day Launch Plans

7 days: finalize the Exposure Scan questionnaire and scoring logic; contract the first independent I-O psychologist/statistician reviewer; publish the first 3 educational content pieces; open outbound to a 100-employer targeted list.
30 days: deliver first 2–3 pilot scans and at least 1 paid Single-Tool NYC Audit; publish the 4-State Compliance Checklist lead magnet; run the first webinar.
90 days: complete the 5-pilot cohort; harden SOPs per the Build-Before-Scale checkpoint; convert at least 1 pilot into the Multi-State Compliance Desk; re-forecast TAM and pricing against real pilot data rather than the Inferred planning ranges used at launch.

Metrics and KPIs

Scan-to-paid conversion rate; single-audit-to-Compliance-Desk upsell rate; cycle time from intake to delivery; rework rate; reviewer-bandwidth utilization; 12-month renewal rate; escalation-to-outside-counsel rate; CAC payback period; gross margin per engagement.

Risks and Mitigations

The two highest-weighted risks are addressability uncertainty (the TAM range is explicitly Unverified) and reviewer-bandwidth scarcity during renewal-season peaks, both directly addressed in the Risk Register and Build-Before-Scale checkpoints below.

Exhaustive Risk Register

What Could Kill This

The two scenarios most likely to kill this business are the addressable-employer-count and willingness-to-pay proving materially smaller than the Inferred planning range once real pilot data comes in, and a sustained reviewer-bandwidth bottleneck during concentrated renewal-season demand that the business fails to solve with a broader contracted-reviewer roster before it damages client trust — both are directly addressed by the pilot caps, Build-Before-Scale checkpoints, and reviewer-roster mitigation above.

Go/No-Go Reasoning

Go. The regulatory catalyst (four independently evolving state/city laws plus a March 2026 federal ruling confirming litigation exposure) is verifiable from primary and named-authoritative sources; the statutory independence requirement structurally mandates outsourcing rather than merely permitting it; competitor density confirms real existing budget rather than unproven demand; the MVP wedge (a free exposure scan converting to a single-jurisdiction audit) is deliverable without a large custom platform; the licensing boundary is workable with a clean non-attorney chokepoint (an independent credentialed reviewer, not a legal-opinion-giving attorney) as long as legal-interpretation questions are consistently routed to outside counsel; and the candidate was confirmed novel against all 517 prior manifest entries after four other researched candidates (self-storage lien-sale compliance, veterinary DEA compliance, multi-state non-compete compliance, pawnbroker compliance) were explicitly evaluated and rejected on competitive-saturation, evidence-thinness, or licensing-risk grounds before this one was selected.

Final Recommendation

Launch the free AI-Hiring Exposure & Audit-Readiness Scan as the primary lead-generation motion targeting mid-market, multi-state employers with confirmed NYC exposure plus at least one additional regulated state, convert the first 5 pilots into paid Single-Tool NYC Audits, contract a roster of at least 3 independent I-O psychologist/statistician reviewers before broad outbound to avoid a single-reviewer bottleneck, and expand into the Multi-State Compliance Desk subscription and the litigation-readiness evidence-packet add-on once the core audit-delivery workflow is hardened per the Build-Before-Scale checkpoints.

Source List

• VerifyWise — NYC Local Law 144 Compliance Checklist for Employers
• Office of the New York State Comptroller — Enforcement of Local Law 144
• NYC Rules — Automated Employment Decision Tools (Updated)
• Proceptual — Penalties for NYC Local Law 144
• Jackson Lewis — California's New AI Regulations Take Effect Oct. 1
• National Law Review — Illinois Anti-Discrimination Law to Address AI Goes Into Effect January 1, 2026
• Littler — Colorado Amends its Artificial Intelligence Law
• K&L Gates — The Changing Landscape of AI: Federal Guidance for Employers Reverses Course
• National Law Review — The Federal Government Quietly Removed Its AI Hiring Guidance
• Forbes — A Federal Judge, A 1967 Law And A Billion Rejected Job Applications
• Akin Gump — Court Allows Discrimination Claims Against AI Hiring Tool to Proceed
• Wiggins Childs — Workday Case Update
• DemandSage — AI Recruitment Statistics
• Bloomberg Law — AI Hiring Compliance Is a Patchwork and Leaves Big Employer Gaps
• DarrowEverett — AI Hiring & Workforce Management 2026 Legal Analysis Updates
• SHRM — AI Bias Audits Are Coming. Are You Ready?
• Mayer Brown — EEOC Issues Title VII Guidance on Employer Use of AI
• Warden AI — product homepage
• Bias Audit USA — product homepage
• AEDT Audits — product homepage
• VerityAI — NYC Local Law 144 Compliance Guide
• G2 — Best AI Bias Audit Services
• AI Laws by State — AI Hiring Laws by State Compliance Map
• U.S. Small Business Administration — 2025 Small Business Profile