CarbonCleared — The Building Performance Compliance & Filing Engine

02Final decision: Blueprint

This run produces a go decision. The candidate cleared the evidence threshold on all twelve required elements: identified buyer, specific painful problem, verified pain statistics, verified existing spend, active demand evidence, competitor/budget validation, credible win logic, narrow MVP wedge, practical first-sale path, service-first fulfillment, no unresolved fatal blocker, and a credible 50%+ gross-margin path.

03Executive summary

Cities have quietly built a nationwide web of building-energy filing laws with brutal, calendar-driven penalties, and in 2025 the first real fines started landing. New York's Local Law 97 alone now exposes every covered building to $0.50 per square foot per month for a missed report — before a single ton of excess carbon is counted. Boards and small owners are publicly confused and angry (Crain's, Habitat, a co-op coalition begging DOB to waive late fees), DOB guidance is vague, and the compliance work — collecting twelve months of utility bills, mapping meters, entering ENERGY STAR Portfolio Manager data, computing emissions against caps, and getting a licensed PE/RA to certify the filing — is exactly the kind of document-and-data drudgery frontier AI now does at near-zero marginal cost.

CarbonCleared sells one outcome: your building is filed, certified, and penalty-clear in every jurisdiction, every year — you never touch Portfolio Manager, DOB NOW, or a utility portal. The wedge is a free Penalty Exposure Scan built from public covered-buildings lists and disclosure data, converting to a $349 RDP-certified benchmarking filing, a $1,450–$2,950 LL97 annual report package, and a $175–$450/building/month multi-law compliance desk. AI performs bill extraction, meter mapping, data QA, deviation detection, emissions math, and report drafting; deterministic validators check caps, factors, and completeness; a licensed RDP reviews and certifies every filing. Expert humans are concentrated at exactly one judgment chokepoint the law itself created — which is why this scales revenue faster than headcount and gets stronger every time models improve.

04Thesis

When a government converts an engineering discipline into an annual filing obligation with per-day and per-square-foot penalties, it creates a permanent, recurring, deadline-driven back-office market. Building energy compliance crossed that line in 2024–2026: LL97's first reports came due May 2025, BERDO's emissions caps went live in 2025, DC BEPS Cycle 1 is evaluated on 2026 data, and 50+ cities now run benchmarking calendars. The incumbents are either boutique engineering firms billing hourly (expensive, slow, retrofit-focused) or self-serve SaaS that makes the owner do the work. The underserved middle — co-op/condo boards and owners of one to fifteen buildings — wants neither a $25,000 consulting study nor another software login. They want the fine to never arrive. An AI-native desk that industrializes utility-data extraction and report drafting, keeps a licensed RDP at the certification chokepoint, and prices flat per building can serve this segment at 50–70% gross margin and compound as models improve, because the dominant cost line is precisely the work models are eating: reading bills, reconciling meters, synthesizing fragmented city rules, and drafting compliant reports.

05Discovery rationale

This run began with a duplicate sweep of all 200 prior manifest entries. Heavily-mined terrains (healthcare RCM, customs/trade, SEC/financial filings, insurance licensing, HR compliance, freight permitting, legal ops) were deprioritized in favor of untouched terrain: the manifest contains no real-estate-energy, benchmarking, or building-performance entry — the nearest neighbors (California climate disclosure, EPCRA/TRI, NERC CIP) differ entirely in buyer, statute, data, and artifact. Fourteen targeted searches were then run across five candidate spaces: building performance compliance, DEA pharmacy compliance, multi-state contractor licensing, CCDF child-care subsidy billing, and NYC DHCR rent registration. Building performance compliance won on the combination of (a) fresh, verifiable 2025–2026 penalty enforcement, (b) organized public buyer distress, (c) an RDP certification requirement that functions as a licensing moat and human chokepoint, (d) a public non-filer list enabling surgical outbound, and (e) genuine multi-city expansion with one repeatable production system (all ordinances converge on ENERGY STAR Portfolio Manager).

06Candidate comparison

Candidates B–E fail on at least one preferred-pattern element that A satisfies; none triggered a fatal disqualifier outright, but only A clears the evidence threshold with margin.

07CODE validation

C — Consumer / buyer trend

Enforcement switched on. First LL97 reports came due May 1, 2025 with retroactive monthly fines; DOB moved LL84 penalty payment into DOB NOW in Dec 2025; BERDO caps went live for large buildings in 2025 with an Aug 15, 2026 reporting deadline; DC assesses Cycle 1 BEPS penalties in 2027 on 2026 data; and since 2025 NYC requires a Registered Design Professional to certify both LL84 and LL97 filings — converting a data chore into a professional-service purchase. Meanwhile more cities keep passing ordinances (Evanston 2025). VERIFIED

O — Opportunity

The failing incumbents are (1) engineering firms selling $10,000–$50,000 studies and hourly retrofit consulting to a board that first just needs to be filed and penalty-clear, and (2) self-serve SaaS (VertPro-style) that still makes the owner gather bills, chase tenant data, and understand the ordinance. Nobody owns the done-for-you, flat-fee, multi-law, multi-city filing desk for the 1–15-building owner. DOB itself is described by owners as "vague, if not unresponsive." VERIFIED

D — Demand

Buyers are visibly spending and complaining: Crain's reports LL97 compliance is "complicated, pricey" for co-ops/condos; a coalition of co-op/condo representatives formally petitioned to waive LL97 late fees (organized distress); Habitat, Brick Underground, and QNS run continuous coverage; ~1,400 owners were still unfiled in Oct 2025; an entire vendor ecosystem (Bright Power, EnPG, Cotocon, Energo, VertPro, ENGIE Impact) already bills for benchmarking — proof of budget, not saturation of the underserved segment. VERIFIED

E — Economic sizing

NYC: ~50,000 LL84-covered buildings and (by borough CBL lists) an LL97 population in the tens of thousands; at $300–$3,000/building/yr the NYC filing-services pool alone is $25M–$120M/yr. Adding LL87 audit coordination (avg ~$9,500/building on a 10-year cycle) and 50+ other cities (Boston, DC, Chicago, Philadelphia, LA, Seattle, Denver, Cambridge, Montgomery County…) plausibly triples the serviceable pool. A 1,500-building desk at a blended $1,400/building/yr is a $2.1M revenue business run by a handful of people. Range is wide; the floor is high enough. INFERRED

08Rubric scorecard

Composite: 4.3 / 5. No fatal disqualifier triggered.

09Target buyer

Wedge ICP: New York City co-op/condo boards and owners of small-to-mid multifamily and mixed-use buildings — 25,000–200,000 ft², one to fifteen buildings, no in-house energy manager — currently served (badly) by a managing agent who forwards DOB notices and a consultant who quotes studies. The economic buyer is the board president / owner-principal, who personally absorbs fines through the operating budget; the champion is the property manager or managing agent who is contractually blamed when a filing is missed.

Expansion ICPs: regional property-management firms (portfolio white-label), owners in Boston (BERDO), DC (BEPS), Chicago, Philadelphia, LA (EBEWE), Seattle, Denver; HUD-assisted multifamily owners (benchmarking already normalized by HUD toolkit); and net-lease commercial landlords with covered buildings scattered across ordinance cities.

10Jobs-to-be-Done

• "Make the DOB notice go away and make sure another one never comes."
• "File everything my building owes this year — benchmarking, emissions report, attestations — without me learning the acronyms."
• "Tell me my fine exposure now and at the 2030 cap tightening, in dollars, before my board meeting."
• "Collect utility data from Con Ed, National Grid, and my tenants without me chasing anyone."
• "Get a licensed professional to certify the filing so I'm not personally attesting to numbers I don't understand."
• "If a violation lands anyway, cure it and handle the penalty paperwork."
• "When I buy or manage a building in another city, tell me what it owes there and just handle it."

11Painful problem

A covered NYC building must now, every single year: verify itself on four covered-buildings lists (LL84/87/88/97); obtain 12 months of whole-building energy and water data across owner and tenant meters; enter and QA it in ENERGY STAR Portfolio Manager by May 1; compute greenhouse-gas emissions against a building-specific cap using DOB factors; assemble and file an LL97 report in DOB NOW with a $210 filing fee; and have both filings certified by a Registered Design Professional. Miss the benchmarking filing: $500/quarter. Miss the LL97 report: $0.50 per ft² per month — $12,500/month on a modest 25,000 ft² building, uncapped until filed, with fines the DOB has said it will not waive. Exceed the cap: $268/ton. Misstate: up to $500,000. The same owner may simultaneously owe BERDO reporting in Boston (daily $150–$300 fines), BEPS pathway compliance in DC (up to $10/ft²), and June benchmarking in Chicago or Philadelphia ($300/day). Boards report that DOB guidance is vague, managing agents give wrong answers about applicability, and consultants quote five-figure studies for what is, in year one, a data-and-filing problem. VERIFIED

12The outcome we sell

We deliberately do not sell energy savings, retrofits, or design work — we sell the compliance outcome, and we referee the retrofit vendors (with referral partnerships) when the exposure math says upgrades beat fines.

13First one-feature MVP wedge

• ICP: NYC co-op/condo board or owner, 25,000–100,000 ft², one covered building, no energy staff.
• Trigger event: DOB violation or penalty notice (LL84/LL97), a quarterly $500 hit, media-driven board panic before the May 1 cycle, or a lender/purchaser asking for compliance status.
• Pain: Accruing, uncapped monthly fines plus personal embarrassment of a board that "missed a law it never heard of."
• One-feature MVP: the May 1 Filing Package — done-for-you LL84 benchmarking filing, RDP-certified, for one building, one cycle.
• Input: signed authorization letter + utility account list (5-minute intake); we pull aggregated whole-building data from the utility and bills via owner forwarding.
• Output: accepted ESPM submission + certification + one-page exposure statement (including the building's LL97 position).
• Human chokepoint: analyst review of meter mapping/GFA; subcontracted or staff RDP (PE/RA) certifies.
• Success metric: filing accepted ≥ 10 business days before deadline; ≤ 45 min human time/building by day 90; zero defect escapes.
• What users ask for next if the wedge works: "Do my LL97 report too" → "Handle everything, every year" (desk subscription) → "Do my Boston/DC buildings" → "Coordinate my LL87 audit and cure my old violations."

14Evidence summary

• Penalty reality (Verified): LL97 $0.50/ft²/mo failure-to-file + $268/ton + $500k false statement; LL84 $500/quarter; LL88 $1,500/yr; BERDO $150–$1,000/day + $234/ton ACP; DC BEPS up to $10/ft² capped at $7.5M; Philadelphia $300/day; Seattle $5,000 + $1/ft²/day; Chicago $100 + $25/day; Denver $0.30/kBtu.
• Demand (Verified): ~94% first-cycle LL97 engagement still left ~1,400 non-filers (public list); coalition petitioned DOB for late-fee waivers; sustained trade-press coverage of board confusion and cost.
• Budget (Verified): benchmarking services from ~$299/building; PE studies $10k–$50k; national vendors (ENGIE Impact, Bright Power, VertPro, Partner Energy, EnPG, Cotocon, Energo) monetize this today.
• Chokepoint (Verified): RDP certification mandatory for LL84 and LL97 since 2025; LL88 requires professional attestation.
• Structural growth (Verified): 50+ cities enforce ordinances, ~9 strict BPS, and new cities keep passing them; DC and Boston penalty cycles land 2026–2027.
• Sizing (Inferred): NYC filing-services pool $25M–$120M/yr; national pool a multiple.

15Claim table

16Source-claim matrix

17Market and demand evidence

New York (wedge): LL84 covers buildings over 25,000 ft² (threshold lowered by LL133), a population of roughly 50,000 buildings; LL97 covers most buildings over 25,000 ft² with borough-by-borough covered-buildings lists republished each January. First-cycle LL97 engagement was ~94% — meaning the city simultaneously proved (a) owners take this seriously and (b) ~1,400 addressable laggards exist on a public list, plus thousands more who filed late, filed wrong, or paid consultants they resent. DOB has issued LL84 violations continuously and moved penalty collection into DOB NOW in Dec 2025 — administrative hardening, not softening.

Expansion cities (verified calendars): Boston BERDO (reporting Aug 15, 2026; caps live; ACP $234/t), DC BEPS (Cycle 1 judged on 2026 data; up to $10/ft²), Chicago (June 1; $100 + $25/day), Philadelphia (June 30; $300/day), LA EBEWE (June 1 + audit/retuning cycles), Seattle (Apr 1 benchmarking + tune-ups; $5,000 + $1/ft²/day), Denver (May 1; $0.30/kBtu over target). One production engine — Portfolio Manager ingestion + city-portal filing + certification — serves all of them; only the rules corpus varies. IMT counts 50+ enforcing cities and a continuing pipeline of new ordinances.

Structural tailwinds: caps tighten sharply in 2030 (NYC) and per-cycle elsewhere, so exposure grows even for compliant buildings; RDP certification requirements professionalize the filing; and insurers/lenders increasingly ask for compliance status at refinance — a second trigger event we exploit in GTM. INFERRED for the lender-trigger frequency; the rest VERIFIED.

18Active buyer conversations

• Crain's: LL97 compliance is "complicated, pricey" for co-op and condo owners — boards hiring consultants they can't evaluate. V
• Habitat Magazine (Feb 2025): co-op/condo coalition formally petitions DOB to waive LL97 late filing fees — organized, named buyer distress. V
• QNS op-ed (Apr 2025): owners call consultant measurement requirements "a make-work program for high-priced companies"; $9,100/household assessments cited. V
• Colbert Law / Brick Underground / Mann Report: recurring board Q&A columns on LL97 applicability, financing, and deadline mechanics — the questions repeat, unanswered, cycle after cycle. V
• Managing agents misinforming boards that LL97 "doesn't apply to co-ops" (documented by attorneys) — a trust vacuum an expert desk fills. V
• Vendor webinars (RAND, FirstService, Cotocon, ASHRAE NY DOB series) drawing owner audiences every spring — proof buyers show up to learn this. V

19Competitive landscape

20Competitor and budget validation

Existing budget source: owners already pay $299–$1,500/building/yr for benchmarking help, $210 LL97 filing fees, $10k–$50k for engineering studies, ~$9,500 average for LL87 cycles, and — involuntarily — quarterly LL84 fines and accruing LL97 penalties. The line item exists in every covered building's operating budget; we redirect it, we do not create it. Why alternatives are insufficient: engineering firms optimize for retrofit engagements, not filing certainty; SaaS shifts labor back to the buyer; managing agents lack competence; and nobody in the small/mid segment offers a flat-fee, guaranteed, multi-jurisdiction desk with a licensed certifier in the loop. Why we win: AI collapses the labor cost that forces incumbents to either charge five figures or make the customer do the work — we deliver engineer-certified filings at SaaS-adjacent prices with a penalty guarantee no software can offer. Why this is not a clone: we sell the outcome (penalty-clear, certified, every law) rather than software access or consulting hours, across every ordinance the portfolio touches.

21Pricing evidence and proposed pricing

Evidence: ~$299/building self-supplied-bills benchmarking (UtiliSave-class); VertPro-style flat filing packages and compliance subscriptions; LL97 report preparation quoted by engineering firms in the $1,500–$7,500 range informally (labeled Inferred — firm quotes are unpublished); LL87 $0.10–$0.30/ft²; DOB's own $210 filing fee; and penalty schedules that make any of these prices trivially rational (one missed LL97 month on 25,000 ft² = $12,500).

Founding cohort: 40% off first cycle, capped at 8 buildings-owners. Never hourly. No contingency or savings-share pricing — flat fees keep us clear of performance-fee disputes and align with public-agency-facing norms.

22Regulatory and compliance considerations

• Professional certification: LL84/LL97 filings require RDP (PE/RA) certification; LL88 requires professional attestation. We satisfy this via a staff or subcontracted licensed PE/RA who actually reviews what they sign (see §23). New York limits practice of engineering/architecture to licensees and lawful business entities — our entity structure and contracts are built so the RDP renders the professional service and we provide administrative/data services around it.
• No legal advice: penalty challenges, hardship applications framed as legal claims, and violation defenses are referred to counsel; we prepare data and documentation packages.
• No engineering design: retrofit design, LL87 audit fieldwork, and submetering installs are subcontracted to licensed firms or referred; we coordinate and file.
• Data handling: utility data authorizations executed per utility (e.g., Con Edison aggregated whole-building data requests); tenant privacy respected via aggregation thresholds; SOC 2-style controls on client credentials for ESPM/DOB NOW portal delegation.
• Accuracy liability: false-statement penalties (up to $500k under LL97) mean our QA engine and RDP review are existential, not optional; E&O insurance covering professional services from day one.
• Pricing legality: flat per-unit fees raise no contingency/fee-split issues; we avoid savings-share pricing that could implicate performance-contracting rules.

23Licensing boundary

24AI-native advantage

The unit of work is: gather 12–40 utility documents, reconcile them to a meter tree, normalize to Portfolio Manager schema, apply a city's rule set, and produce a certified filing plus a plain-English exposure narrative. Incumbents solve this with analyst hours (and price accordingly) or push it back on the owner. Frontier models change the economics of every step: vision models read any bill format; LLMs reconcile meter/tenant ambiguities and draft the narrative; grounded retrieval keeps 50+ city rule sets current; deterministic code does the arithmetic. What cannot be automated — professional certification, GFA judgment calls, exception decisions on odd fuel sources or mixed-use allocations — is a thin, high-value layer the law itself demands stay human. AI tasks: extraction, classification, mapping, drafting, monitoring, QA pre-checks. Human tasks: exception adjudication, RDP review/signature, client counsel-referral decisions. Automation tasks: portal submissions, deadline orchestration, data requests. Deterministic tasks: emissions factors, cap math, fee math, completeness gates. Never fully automated: certification, exposure-statement final review, anything filed under professional seal.

25Internal AI engine architecture

1. Intake layer: authorization letter e-sign; utility account census; portal delegation capture; building profile (BBL/BIN, GFA docs, use types); completeness gate refuses ambiguous intakes.
2. Normalization layer: bill/statement OCR to canonical meter-month records; meter-tree construction; unit conversions; gap detection.
3. Retrieval & knowledge layer: versioned corpus of ordinances, DOB rules/service notices, ESPM technical references, emission-factor tables, covered-building lists — every answer grounded and cited internally.
4. AI workbench layer: extraction, meter-mapping suggestions, anomaly explanations, report and exposure-statement drafting, cure-letter drafting.
5. Deterministic rules layer: emissions math, cap comparisons, penalty calculators, deadline calendars, filing-fee math, jurisdiction applicability tests — code, not model output.
6. Human chokepoint layer: analyst exception queue (confidence-routed); RDP certification bench with mandatory review checklist.
7. QA layer: dual-path recomputation (model-extracted vs deterministic re-parse), year-over-year deviation tripwires, sampling re-review, pre-filing checklist.
8. Delivery layer: filed confirmations, certification letters, annual Compliance & Exposure Statement, client portal-free PDF pack.
9. Learning loop: every RDP correction and city kickback becomes a rule, prompt, or validator; defect postmortems feed SOPs.
10. Model-portability layer: vendor-agnostic prompts + eval harness on a gold set of buildings so model upgrades are one-day swaps.

26AI-vs-human operations pipeline

27Dynasty translation layer

1. Buyer translation

Who pays: the board/owner whose operating budget absorbs fines. Urgent problem: statutory deadlines with uncapped monthly penalties and a city that will not waive them. Outcome wanted: "never see a DOB energy violation again, and know my 2030 number."

2. Service translation

Done-for-you: we collect data, compute, draft, certify (via RDP), file, confirm, and monitor. Customer receives filings + certification + exposure statement + guarantee. Automation handles extraction, math, drafting, filing; humans handle exceptions and the professional signature.

3. Workflow translation

Scan → intake & authorizations → data acquisition → normalization → computation → draft → analyst review → RDP certification → filing → confirmation → exposure debrief → monitoring → renewal (auto-proposed 120 days before each deadline).

4. Tooling translation

Day one: ENERGY STAR Portfolio Manager, DOB NOW, utility data programs (Con Ed AWBD etc.), e-sign, shared drive, spreadsheet trackers, frontier-model API for extraction/drafting, a simple CRM (HubSpot-class), penalty calculator scripts. Later: intake portal, orchestration queue, rules-corpus service, eval harness. No custom platform before revenue.

5. Sales translation

"NYC now fines $0.50 per square foot per month if your building misses its emissions filing. We file everything your building owes — certified by a licensed engineer — for a flat fee, guaranteed. Here's your free penalty-exposure scan; it takes us two days and costs nothing."

6. Delivery translation

First three customers fulfilled with spreadsheets, manual ESPM entry, and a subcontracted PE — fully manual is profitable at $349/$1,450 price points. Automate extraction first, then portal work, then monitoring.

7. Expansion translation

City playbooks become templated rule packs; the desk becomes a multi-city subscription; managing agents get a white-label tier; LL87/tune-up coordination and refinance compliance letters become attach products; eventually the rules corpus + eval harness is licensable infrastructure.

28Anti-duplication analysis

What exists: engineering consultancies (RAND, Bright Power, Cotocon), benchmarking SaaS/services (VertPro, Touchstone IQ, ReDocs, UtiliSave), enterprise sustainability platforms (ENGIE Impact), and compliance trackers (SiteCompli). Why we are not a copy: none combines (a) done-for-you fulfillment (vs SaaS self-serve), (b) flat per-building pricing with a filing-performance guarantee (vs hourly/study pricing), (c) multi-law, multi-city scope on one engine (vs single-ordinance add-ons), (d) owned RDP certification as the product's spine (vs a signature scrounged per filing), and (e) an AI-native cost structure that makes the small/mid segment profitably serviceable at all. Under-served segment: 1–15-building owners and boards — too small for Bright Power, too busy for VertPro. Unsolved pain: tenant/whole-building data chasing, multi-law calendar ownership, violation cures, plain-dollar exposure narratives. Vs prior manifest entries: zero overlap — first building-energy entry in 200 runs; nearest neighbors (CA climate disclosure, EPCRA/TRI, NERC CIP, HOA estoppel) differ in statute, buyer, data, and artifact. Honest note: this reuses the manifest's proven free-diagnostic → per-unit filing → desk-subscription operating pattern in a genuinely new domain.

29Anti-commoditization analysis

If future general models let an owner "just ask AI" to benchmark: (1) the RDP certification requirement still forces a licensed human into the loop — a legal, not technical, barrier; (2) liability transfer is the product — a chatbot doesn't reimburse your late fee or carry E&O; (3) whole-building data acquisition requires utility authorizations, tenant coordination, and portal standing that owners won't self-manage annually; (4) the rules corpus across 50+ mutating city programs is a maintained asset, not a prompt; (5) our unit economics improve with the same model progress, letting us cut price ahead of DIY substitution while widening the guarantee. If cities themselves auto-benchmark (some auto-populate utility data), the desk migrates up-stack to LL97/BPS performance reporting, adjustments, cure work, and multi-city portfolio ownership — the judgment-and-liability layer, which regulation keeps expanding (2030 cap step, BEPS Cycle 2, BERDO tightening).

30Service delivery workflow

1. Scan (free): pull CBL status across LL84/87/88/97 (+ other cities), public disclosure history, violation records; compute current + projected exposure; deliver 2-page report and 20-minute debrief.
2. Engage: flat-fee order form; authorization letter; utility census; portal delegation.
3. Acquire: utility aggregated-data requests; owner bill forwarding; tenant data protocol where required.
4. Produce: extraction → meter tree → ESPM entry → deterministic emissions/cap math → drafts.
5. Review: analyst exception queue → pre-filing checklist → RDP certification.
6. File: ESPM submission / DOB NOW report (+$210 fee paid by client) / other-city portal; capture confirmations.
7. Deliver: compliance pack + exposure statement + debrief; log defects (target zero).
8. Monitor: kickbacks, rule changes, CBL updates, violation dockets; cure anything that appears.
9. Renew: auto-proposal 120 days pre-deadline; desk subscribers roll automatically.

31Operations as product

Variance is eliminated by treating the desk as a production line: SOPs per ordinance per cycle; structured intake checklists with hard completeness gates (no ambiguous GFA, no missing meter, no unsigned authorization enters production); required-evidence lists (deed/DOF records for GFA, 12 full months per meter); automated completeness checks; a single exception queue with confidence-scored routing; reviewer assignment by ordinance specialization; audit trails linking every filed figure to a source document hash; version-controlled rule packs with effective dates; gold-standard example buildings for regression-testing prompts and validators; red-team checks before each filing season (seed known errors, verify tripwires catch them); customer-ready output templates; root-cause analysis on any kickback or defect; and a postmortem loop that converts every correction into a validator, SOP line, or prompt change within one week.

32No-holes quality engine

• Dual-path verification: every filed number recomputed by an independent deterministic re-parse; mismatches block filing.
• YoY tripwires: >15% deviation in any meter-month or EUI triggers mandatory human explanation before filing.
• Cap-math lockdown: emissions factors and limits live in versioned tables with test suites — never model-generated.
• RDP checklist: certification cannot be signed until the system shows green on completeness, dual-path, deviation, and GFA-evidence checks.
• Escape audit: monthly sample re-review of filed packages by a second analyst; any escape triggers postmortem + validator.
• Season red-team: pre-May and pre-June synthetic-error drills.

33What the human expert actually does

34Minimum viable offer

"The May 1 Filing Package — $349, certified, guaranteed." We benchmark and file one building's LL84 submission, certified by a licensed design professional, delivered ≥10 business days before deadline, with a free Penalty Exposure Scan included and our late-filing guarantee behind it. Founding cohort (first 8 buildings): $209. Every deliverable in the package doubles as the diagnostic that sells the LL97 report and the annual desk.

35Fulfillment process (first 3 customers, manual)

1. Run scan from public data (2 analyst-hours with AI assist).
2. E-sign engagement + DOB/ESPM delegation + utility authorization (day 1).
3. Request Con Ed/National Grid aggregated data; owner forwards missing bills (days 2–10).
4. Manual ESPM entry from AI-extracted bill table; spreadsheet meter tree (day 11–13).
5. Subcontract PE reviews package against checklist, certifies (day 14–16).
6. Submit; capture confirmation; deliver pack + 20-min debrief; propose LL97/desk (day 17–20).

Do not automate at first: portal filing, utility requests, debriefs. Automate later: extraction QA, monitoring, renewal proposals. Never automate: certification.

36Tools and systems

• ENERGY STAR Portfolio Manager (universal reporting rail) + DOB NOW + city portals (Boston BERDO portal, DC DOEE, Chicago, Philly, LA, Seattle, Denver).
• Utility data programs: Con Edison aggregated whole-building data, National Grid, plus per-city equivalents.
• Frontier-model API (bill extraction, drafting, rule-grounded Q&A) + OCR fallback; Python validators for factors/caps/penalties.
• E-sign, shared drive with hash-logged evidence store, lightweight CRM, ticketing/exception queue (Airtable/Linear-class), calendar orchestration for 50+ deadline sets.
• Public-data harvesters: covered-building lists, LL84 disclosure dataset, DOB violation records — powering scans and outbound.
• E&O policy; engagement-letter templates separating administrative vs professional services.

37Human-in-the-loop quality control

Three human gates: (1) analyst exception adjudication on anything below confidence threshold or tripping a deviation rule; (2) pre-filing checklist sign-off by a second operator on every unit in the first 90 days, sampling thereafter; (3) RDP review and certification on 100% of filings, permanently — the law makes this gate non-negotiable, which is precisely why it anchors the business. Every human correction is logged with a reason code and feeds the weekly engine-improvement review.

38Nonlinear scaling and unit economics

Automation share of total task minutes: ~55% at launch → ~75% day 90 → ~85% year one (certification minutes are the floor). Throughput: 6–8 filings/analyst/day at launch → 18–25 by day 90. Cycle time: 20 days → 8 days (data acquisition latency dominates). Targets: rework ≤3%, quality-failure escapes ≤0.5%, escalation ≤8%. Margin expands with volume because rule-corpus and validator costs are fixed while RDP unit review shrinks as packages arrive pre-verified. CAC payback ≤ 60 days at $349 wedge (CAC target ≤$250 via outbound-on-public-lists). Funnel assumptions (all Unverified kill tests): scan→paid ≥25%; filing→desk attach ≥60%; desk renewal ≥90%; wedge→LL97 attach ≥50%. Human labor and model costs are stated above — nothing hidden in "operations."

39Distribution proof table

40Sales and outreach plan

Founder-led, diagnosis-first. Every conversation starts from the building's own numbers (scan), not a demo. Sequence: scan delivered → 20-minute debrief (walk the board through current accrual, 2030 cliff, and cure options) → single-filing close (procurement-friendly $349–$2,950 flat) → desk attach at delivery ("never think about this again for $295/month") → portfolio/white-label conversation once trust is earned. Objection playbook: "our engineer handles it" (we file cheaper and guarantee it; your engineer keeps the retrofit work), "we'll do it ourselves" (here's the checklist; call us in April), "price" (one month of LL97 late fines = 3–8 years of our desk).

41Founder-led content plan

Teach the pain in dollars. Pillars: (1) penalty math made visceral (per-building examples from public data); (2) deadline mechanics per city; (3) misconception demolition ("co-ops are exempt" — no; "benchmarking is optional" — no); (4) cost-of-doing-nothing ledgers; (5) rule-change alerts (DOB service notices, BERDO extensions, BEPS cycles) explained within 48 hours; (6) anonymized teardown of real exposure scans; (7) buyer FAQ columns mirroring Habitat/Brick Underground question patterns. Cadence: 2 LinkedIn posts + 1 long-form piece weekly; every piece ends in the scan CTA. High-performing organic pieces become paid-retargeting creative in month 3.

42First 30 days of content

10 educational posts

1. "$0.50/ft²/month: what LL97's late fee actually does to a 40,000 ft² co-op's budget."
2. "The four NYC lists your building might be on (LL84/87/88/97) — and how to check in 5 minutes."
3. "Your managing agent said LL97 doesn't apply to co-ops. The DOB disagrees."
4. "Why 2025 changed everything: the RDP certification rule nobody told boards about."
5. "LL84's quiet $2,000/year: how quarterly benchmarking fines snowball."
6. "2030 is the real cliff: your building's cap is about to drop — here's the math."
7. "Boston owners: BERDO's deadline moved to Aug 15, 2026 — what that does and doesn't buy you."
8. "DC BEPS penalties start landing in 2027. Your 2026 data decides them — this year."
9. "Whole-building data: why Con Ed's aggregated request is the step everyone botches."
10. "Fines vs fixes: when paying $268/ton is (briefly) rational, and when it's board malpractice."

3 diagnostic teardown formats

1. Anonymized Penalty Exposure Scan walk-through: a 60k ft² Queens co-op, line by line.
2. "We re-checked 10 public LL84 disclosures: 4 had errors that change the LL97 number."
3. Violation autopsy: how one missed authorization letter became $6,000 in fines.

2 lead-magnet angles

1. Free Penalty Exposure Scan (flagship — from public data, no owner effort).
2. "The 2027 Deadline Grid": one-page PDF of every date your building owes in NYC/Boston/DC/Chicago/Philly/LA/Seattle/Denver.

1 webinar

"Your building's LL97 number, live": we run real-time scans on attendee buildings submitted in advance (with consent), co-hosted with a managing agent.

1 outbound diagnosis template

"Subject: [Address] — accruing ~$[X]/month since [date]. We ran your building against DOB's covered-buildings list and violation records. Attached is a 2-page exposure scan: current accrual, what filing cures it, and your 2030 position. If useful, 20 minutes this week and we'll walk your board through it. No charge for the scan — filing is a flat $[Y] if you want it handled, certified by a licensed PE."

43Lead magnet and waitlist plan

Buyer receives before paying: the Penalty Exposure Scan — building-specific fine accrual, deadline grid, 2030 projection, and cure path, built from public data in 48 hours. Why it creates trust: it demonstrates we already understand their building better than their agent does, cites DOB's own records, and prices the problem in dollars. Pain signal captured: requesting a scan self-identifies deadline anxiety; the scan intake captures portfolio size, cities, and violation history. Follow-up: 20-minute debrief within 5 days, then deadline-keyed drip. Sales-ready qualification: covered building + upcoming deadline ≤120 days or active violation + decision-maker on the debrief. Waitlist mechanics: founding cohort capped at 8 buildings at 40% off; overflow joins a dated waitlist keyed to their next statutory deadline — scarcity is real because RDP review capacity is finite. Waitlist signups are explicitly not treated as PMF; only paid filings, desk attach, and renewal are.

44Warm GTM plan

Work the scan pipeline weekly; re-engage every scan recipient at T-90/T-45/T-21 before their deadline with their own accrual math updated. Offer consultative reviews to managing agents' portfolios (one free portfolio-wide scan for a signed referral agreement). Mine webinar attendees and calculator users with deadline-triggered follow-ups. Existing-contact pass: every prior professional contact in NYC real estate gets the one-page deadline grid with a personal note. Every warm touch leads with the recipient's own building data, never a generic pitch.

45Targeted outbound plan

Precision lists from public data: (1) LL97 non-filers (~1,400, shrinking but refreshed each cycle); (2) LL84 violation recipients (quarterly fines accruing — provable pain); (3) buildings that filed late last cycle (chaos signal); (4) buildings near the 2030 cap cliff per public disclosure EUIs (exposure signal); (5) Boston/DC covered lists as those deadlines approach. Each target gets a personalized diagnosis memo (template §42), not a demo ask. Volume: 40 researched sends/week; every send includes the actual scan or a one-number teaser from it. Compliance: public-record data only; CAN-SPAM-clean; opt-out honored globally.

46Answer-engine / search visibility plan

Owners now ask ChatGPT/Perplexity/Google "what is my LL97 fine" and "when is Chicago benchmarking due." We publish the canonical machine-quotable answers: per-ordinance pages with exact deadlines, penalty formulas, worked dollar examples, effective-date change logs, and published flat pricing + guarantee (answer engines favor concrete, dated, attributed facts). Schema-marked FAQ and calculator pages per city; a maintained "US building-energy deadline grid" dataset page that earns citations; rule-change posts within 48 hours of DOB/BERDO/DOEE notices to win freshness. Measure: assistant-referral traffic, calculator completions, scan requests attributed to organic/AEO.

47Pilot design and early-demand trap mitigation

Founding pilot: 8 buildings max, ≤3 per owner, NYC only, LL84 + LL97 scope only (no LL87 projects, no other cities, no violation litigation support). Incentive: 40% off first cycle + locked founding desk rate. Learning objectives: minutes-per-unit truth, utility-data latency map, RDP review friction, kickback taxonomy, debrief-to-attach conversion. Feedback: weekly 15-minute owner check-in + defect log. Product feedback = anything about data acquisition, clarity of exposure statements, filing mechanics. Custom work = retrofit advice, financing analysis, litigation support — referred out, logged, not built. The early-demand trap (drowning in bespoke consulting because buyers are desperate) is mitigated by the fixed scope list, the referral bench (PE firms, counsel, energy auditors), and the rule that nothing ships to one client that isn't templated for all.

48Early-access feedback flywheel

Every RDP correction, analyst exception, city kickback, and client confusion is logged with a reason code. Weekly triage converts each into exactly one of: validator rule, prompt/corpus change, SOP line, template edit, or referral-bench addition — shipped within 7 days, regression-tested against the gold set. Fix-before-expansion list: any defect class that occurred twice must have a shipped countermeasure before the next pilot slot opens. The certification bench reviews the correction digest monthly so professional judgment patterns migrate into pre-review checks.

49Build-before-scale checkpoints

• After 5 buildings: harden intake (authorization + utility census + GFA evidence gates), completeness checks, and the exception taxonomy.
• After 10 buildings: harden SOPs per ordinance, reviewer checklists, delivery templates, kickback cure playbooks; sign second RDP.
• After 20 buildings: pause new intake until measured: COGS/unit, rework %, escalation %, cycle time, scan→paid, attach %, RDP minutes. Expand only if §38 day-90 targets are on trajectory.
• Acceptable temporary manual work: hand ESPM entry, manual portal filing, spreadsheet trackers.
• Red-flag manual work (signals non-scalability): per-client bespoke exposure models, unpayable utility-data chases exceeding 30 days/building, RDPs re-deriving instead of reviewing, recurring GFA disputes without a documented resolution path.

507-day launch plan

1. Day 1–2: entity + E&O quote; engagement-letter templates (admin vs professional services); subcontract RDP signed (one PE or RA with NYC filing experience).
2. Day 2–3: build scan generator v0 from public CBL/disclosure/violation data; produce 10 sample scans.
3. Day 3–4: one-page site with calculator + scan CTA + published pricing; deadline-grid PDF.
4. Day 4–5: first 40 outbound diagnosis memos to violation-list buildings; 3 managing-agent meetings booked.
5. Day 5–7: deliver first debriefs; close first founding filing; open defect log and correction flywheel from unit #1.

5130-day launch plan

1. Weeks 1–2: 5 founding buildings signed; utility data requests in flight; first LL84 filings produced and certified.
2. Weeks 2–3: first LL97 report package sold to a founding client; webinar #1 with a managing agent; 10 content pieces live.
3. Weeks 3–4: measure minutes-per-unit and data latency; ship top-5 validators from corrections; 2 desk subscriptions attached; pilot cap reached or waitlist forming; checkpoint review per §49.

5290-day launch plan

1. Month 2: full pilot cohort (8) served; violation-rescue offer live; AEO pages for all NYC laws ranked-tracked; second RDP onboarded; kill-test dashboard live.
2. Month 3: 20-building checkpoint; desk attach ≥60% validated or pricing/packaging revised; Boston rule pack built against Aug 15, 2026 BERDO deadline; DC scan pilot for BEPS Cycle 1 endgame; white-label agreement signed with one managing agent; decision gate: expand to city #2 vs deepen NYC.

53Metrics and KPIs

• Acquisition: scans delivered/week; scan→debrief %; scan→paid % (kill: ≥25%); CAC (≤$250 wedge); AEO/organic scan share.
• Delivery: minutes/unit (analyst + RDP separately); cycle time; utility-data latency; on-time filing rate (100% target); kickback rate; defect escapes (≤0.5%).
• Economics: GM per offer; automation %; revenue/FTE; rework %; RDP cost/unit.
• Expansion: filing→desk attach (≥60%); wedge→LL97 attach (≥50%); desk renewal (≥90%); buildings/customer; city #2 CAC parity.
• Trust: guarantee payouts ($0 target); NPS of board debriefs; agent referrals/quarter.

54Risks and mitigations (top line)

The three existential risks are instrumented from unit #1: (1) a defective certified filing (false-statement exposure, guarantee payouts) — mitigated by dual-path recomputation, deviation tripwires, 100% RDP review, escape audits, and E&O; (2) political softening of the laws (NYC has already granted extensions; BERDO moved a deadline; a future administration could gut penalties) — mitigated by multi-city spread, the fact that filing obligations survive even when penalties flex, and up-stack migration to performance-cycle work (BEPS/BERDO caps) that tightens on schedule; (3) utility/city auto-population commoditizing benchmarking — mitigated by owning the LL97/BPS judgment layer, certification, cures, and multi-law calendar rather than data entry alone. Full register below.

55Exhaustive risk register

56What could kill this

1. A certified-filing defect scandal in year one — quality engine failure is unrecoverable in a trust business (R1).
2. Wholesale repeal/gutting of LL97-class penalty regimes across multiple cities simultaneously (R2) — judged unlikely given 50+ jurisdictions and tightening cycles, but watched quarterly.
3. Utility-to-city full automation of both benchmarking and performance reporting, eliminating the filing task itself (R3) — no jurisdiction reviewed is close to this for LL97/BPS-class reports.
4. Failure of the day-90 kill tests: scan→paid <15%, attach <40%, or human minutes stuck >2× target — would force repricing or a no-scale verdict at the 20-building checkpoint.

57Go/no-go reasoning

Go. Every evidence-threshold element is satisfied with primary-source or strong secondary verification: a named buyer with organized public distress; penalties that are specific, large, recurring, and already being assessed; existing category spend from $299 filings to $50k studies; an RDP requirement that simultaneously creates the human chokepoint, the moat, and the trust interface; a public target list for outbound; a $349 one-feature wedge deliverable in under three weeks; service-first fulfillment needing zero custom software; a 50–70% gross-margin path resting on automation of exactly the tasks frontier models are best at; and honest kill tests scheduled at the 20-building checkpoint. The weakest rubric line — competitive whitespace (3/5) — is priced in: the category's existence is what proves budget, and the specific segment-offer combination (flat-fee, guaranteed, multi-law, RDP-certified desk for 1–15-building owners) is demonstrably unoccupied.

58Final recommendation

Launch the NYC wedge immediately against the 2027-cycle calendar (and the live backlog of accruing violations), with the founding cohort capped at 8 buildings. Sign one subcontract RDP this week; deliver ten Penalty Exposure Scans in the first fourteen days; close the first certified May 1 Filing Package within three weeks; attach the desk at every delivery debrief; run the 20-building checkpoint honestly against the §38 kill tests before touching city #2 (Boston, against the Aug 15, 2026 BERDO deadline, then DC ahead of 2027 BEPS penalty assessment). Do not build software beyond the scan generator and validators until the pilot cohort proves the minutes-per-unit curve.

59Source list

1. Cozen O'Connor — NYC Local Law 97 Deadline: May 1, 2025 (client alert)
2. RAND Engineering & Architecture — 2025 NYC Local Law Compliance Deadlines
3. NYC DOB — LL84 Benchmarking Violations (penalties, DOB NOW payment)
4. NYC DOB — LL84 Benchmarking Law (covered buildings)
5. Energo — NYC Local Law 84: 2026 Deadline, Penalties, RDP certification requirement
6. NYC DOB — LL97 Compliance Report Submission Process (RDP attestation, $210 Art. 321 fee)
7. Habitat Magazine — 94% of NYC buildings meet LL97 compliance; ~1,400 non-filers (Oct 2025)
8. NYC DOB — LL97 compliance extension service notice (Dec 31, 2025)
9. Crain's New York Business — LL97 compliance is complicated, pricey for co-op/condo owners
10. Habitat Magazine — Co-op/condo coalition urges waiving LL97 late fees (Feb 2025)
11. QNS — Op-ed: LL97 set to cost co-op and condo owners millions in fees and fines (Apr 2025)
12. Colbert Law — Unique challenges for condo and co-op boards facing LL97
13. UtiliSave — LL84 benchmarking service (pricing floor evidence)
14. Boston.gov — BERDO (thresholds, ACP $234/ton)
15. Ruberto Israel & Weiner — BERDO deadlines client alert (Aug 15, 2026 extension)
16. DC DOEE — Building Energy Performance Standards
17. Steven Winter Associates — DC BEPS: the first compliance year
18. IMT — 2025 Building Policies Outlook (BPS proliferation)
19. Facilities Dive — The 2026 map of building performance standards
20. Rimkus — Energy Benchmarking 2026 (multi-city deadlines & penalties)
21. City of Chicago — Chicago Energy Benchmarking
22. VertPro — Energy benchmark compliance (self-serve incumbent evidence)
23. ENGIE Impact — Benchmarking compliance (enterprise incumbent evidence)
24. Bright Power — Local Law 97 strategy guide/FAQs
25. Partner Energy — Utility data collection & benchmarking services
26. Energo — LL87 explained (audit/RCx costs, 10-year cycle)
27. NYC DOB — LL87 Energy Audits & Retro-commissioning
28. NYC DOB — LL88 Lighting Upgrades & Sub-metering
29. The Falcon Group — LL88 requirements and penalties
30. HUD Exchange — Multifamily Utility Benchmarking Toolkit (data-collection bottleneck)
31. Drug Topics — Independent pharmacies must prepare for DEA inspections (candidate B)
32. Harbor Compliance — Construction licensing solutions (candidate C incumbent)
33. New America — Changes in federally supported child care payments (candidate D)
34. MRE Consulting — DHCR compliance & rent registration services (candidate E incumbent)
35. BBG LLP — NYC rent overcharge claims and legal risk (candidate E)