Building Performance Standards (BPS) Emissions-Compliance Engine

01 Thesis

A wave of state and city Building Performance Standards (BPS) now forces owners of large existing buildings to measure, report, and hit emissions/energy targets — or pay. New York City's Local Law 97 caps carbon for ~50,000 properties over 25,000 sq ft and fines $268 per metric ton of CO₂e over the cap, every year; the annual emissions report is due May 1 and must be certified by a Registered Design Professional (a licensed PE or RA) filed through the DOB BEAM portal (Verified). Washington State's Clean Buildings Performance Standard made Tier 1 buildings over 220,000 sq ft due June 1, 2026, with penalties up to $5,000 + $1.00/sq ft per year (Verified). There are now 16 active BPS laws and ~60 jurisdictions requiring benchmarking of large buildings, and the list is growing as states fill a federal vacuum (Verified).

We sell the outcome, not a sustainability dashboard. An internal AI engine ingests a building's utility bills, ENERGY STAR Portfolio Manager data, meter/account inventory, and gross-floor-area records; reconciles whole-building energy across meters and fuels; computes the building's emissions/EUI against the jurisdiction's cap tables; models penalty exposure for the current and 2030 periods; drafts the exact report each jurisdiction requires; and — for LL97 — a Registered Design Professional on our bench reviews and certifies the filing. When a covered-buildings-list dispute or notice of violation arrives, we prepare the response package. The customer experiences a managed result: "every building is filed, on time, at the lowest defensible number, and nobody is chasing a city deadline."

This is not a co-pilot. A REIT asset manager with 60 buildings across NYC, Boston, DC, Denver, and Washington State does not want another platform her team must operate against dozens of portals and deadlines; she wants the filings done and the penalty exposure managed, by a provider who can stand behind a licensed certification. The work is document-and-data synthesis against fixed statutory rules and city forms, so it decomposes into automatable steps with judgment concentrated at a few licensed chokepoints — the shape that lets revenue scale faster than headcount toward software-like margins.

02 Discovery rationale

This run scanned opportunity zones for regulated, document-heavy, deadline-driven work that is commonly outsourced and decomposes cleanly: energy/environmental compliance, insurance operations, financial information-reporting, and consumer-finance compliance. Six candidates were generated and three deep-validated (see §3). BPS compliance won on the combination of (a) verified, statutory, recurring, penalty-backed demand — a mandatory annual self-report in a fast-growing patchwork of city/state laws; (b) a hard, dated catalyst landing right now — Washington's Tier 1 deadline was June 1, 2026 and NYC's annual report is due May 1 with penalties already live and 2030 caps tightening sharply; (c) a real licensing chokepoint — NYC LL97 legally requires a PE/RA to certify the filing — which both forms a moat and bounds the model; and (d) a public, address-level prospect database (NYC's Covered Buildings List and LL84 disclosure data) that reveals exactly which buildings are over their cap, turning outbound into a data problem.

The decisive evidence was the asymmetry between how mechanical the work is (utility bills → meter reconciliation → cap-table lookup → certified report) and how painful and fragmenting it is at scale (dozens of jurisdictions, incompatible portals, whole-building data spread across tenants and utilities, penalties that recur annually). Incumbents sell owners software they must operate (Measurabl, nZero) or one-off engineering consulting; nobody offers a mid-market owner a single done-for-you, RDP-certified, multi-jurisdiction managed filing. That whitespace is where an internal AI engine plus a thin licensed-review layer produces a done-for-you outcome at a margin a boutique MEP firm doing it by hand cannot.

03 Candidate comparison

Six candidates generated this run; the top three deep-validated. Scores are the author's 1–5 rollup of the 15-factor screen.

04 Hard disqualifier check

No disqualifier fails. Item 11 is the honest tension of the category and is addressed head-on in §17 and §20.

05 Rubric scorecard

Composite 4.2/5. Gate 1 (5): owners already outsource energy/benchmarking work to consultants and PEs; the buyer wants the finished, certified filing. Gate 2 (4): the workflow is highly decomposable — extraction, meter reconciliation, cap-table math, report drafting — with judgment at pathway selection and certification. Gate 3 (3, honest): the emissions arithmetic is deterministic and free estimators exist; the real intelligence is multi-jurisdiction rule synthesis, messy whole-building data reconciliation, and defensibility — so we win on operations + certification, not proprietary math. Gate 4 (4): the PE/RA certification requirement and up-to-$500,000 false-filing penalty deter casual entrants. Gate 5 (5): entirely digital. Gate 6 (4): better models sharpen extraction/retrieval/penalty-modeling while the RDP stays the trust interface.

06 Opportunity

Sizing (Inferred): NYC alone lists ~50,000 covered properties; nationally ~60 jurisdictions benchmark billions of square feet. If a managed per-building fee runs $2,500–$5,000/yr and a specialist captures even a low-single-digit share of BPS-penalty-jurisdiction buildings, that is a mid-eight-figure revenue band before adjacent benchmarking-only work and penalty-response packages. Ranges are wide and must be firmed bottom-up in pilots.

07 Evidence & source-claim matrix

08 Why now

Verified regulatory catalysts

• Washington CBPS Tier 1 landed June 1, 2026 (>220k sq ft); Tier 1 >90k sq ft follows June 1, 2027, and >50k sq ft June 1, 2028 — a rolling three-year wave of first-time filers with penalties attached (Verified).
• NYC LL97 penalties are live and the annual RDP-certified report is due each May 1 (2026 grace to June 30; extension to Aug 29 for $60). The 2030 caps tighten sharply, pulling a large cohort out of "on track" (Verified).
• The map keeps growing: 16 active BPS laws, ~60 benchmarking jurisdictions; Evanston and West Hollywood adopted BPS in 2025, with Santa Monica, Lakewood (CO), and a New York statewide bill flagged as likely 2026 actions (Verified).
• Federal vacuum shifts action to states/cities: IMT's 2026 outlook frames local/state leadership as the driver as federal building-energy policy recedes — the opposite of a rule at risk of repeal (Verified).

Inferred capability shift

• Frontier multimodal models can now parse heterogeneous utility bills, reconcile meter/account/BIN mappings, and retrieve/apply jurisdiction rule sets at a quality and cost that makes a low-touch, high-accuracy managed service viable (Inferred).

Unverified

• That owners will switch from status-quo consultants/software to a specialist managed service at the pricing modeled — a hypothesis to test in the pilot, not a settled fact (Unverified).

09 Customer & PMF

10 The outcome we sell

• Acceptance criteria: report accepted by the authority having jurisdiction by the statutory deadline (or approved extension); benchmarking marked compliant; no rejection or resubmission attributable to our data handling.
• Customer promise: "Every covered building filed on time and certified, at the lowest defensible number, with the exposure quantified and a record you can hand a lender or auditor."
• Exclusions: physical retrofits, capital projects, energy-code design, legal advice, and any guarantee of a specific penalty amount. We report and certify; we do not do construction.
• Rework / make-good policy: if a rejection or late filing is caused by our error, we remediate at no charge and cover the resulting late-filing penalty up to the fee paid — a defined, bounded warranty, not an open-ended contingency.
• Success metric: 100% on-time certified filings; measured reduction in avoidable penalty exposure versus the owner's baseline.

11 Internal AI engine architecture

An internal operating engine — the owner never touches it. The RDP is the interface.

1. Intake: utility bills (PDF/CSV/EDI), ENERGY STAR Portfolio Manager exports/web-service feeds, rent rolls & GFA documents, prior benchmarking/LL97 filings, meter/account inventories, DEP water data, and each jurisdiction's covered-buildings-list status.
2. Normalization: OCR/parse bills; reconcile meters ↔ accounts ↔ BIN/BBL and property; deduplicate; convert units (kWh, therms, district-steam Mlb, oil gallons); verify gross floor area to DOB standards; version every input.
3. Retrieval & knowledge: a jurisdiction rule library — LL97 occupancy-group cap tables and emissions coefficients, Article 320 vs 321 pathways, WA CBPS EUI targets (WAC 194-50), Boston BERDO, DC BEPS, Denver Energize thresholds, benchmarking rules, deadlines, and fees — plus ESPM property-type mappings.
4. AI workbench: extract bill line items, classify space/property types, map meters into ESPM, compute EUI/emissions, model penalty exposure (current + 2030), draft the compliance report and narrative, and flag anomalies (weather-normalized spikes, missing months, unmapped meters).
5. Deterministic rules layer: emissions-coefficient math, cap-table lookups, penalty formulas, deadline calendars, and pathway eligibility are rule-based and unit-tested — never left to a probabilistic model.
6. Human chokepoint layer: a senior energy analyst approves data completeness and pathway; a Registered Design Professional (PE/RA) reviews and certifies the LL97 filing; a qualified energy professional signs WA CBPS documentation.
7. QA layer: completeness checks (12 full months per meter, all meters mapped, GFA reconciled), gold-standard comparisons, red-team on the emissions calc, and per-building confidence scoring gate release.
8. Delivery layer: file via BEAM / DOB NOW (NYC), submit ESPM, file to WA Commerce and other city portals; deliver the client compliance package and penalty memo.
9. Learning loop: DOB rejections, covered-buildings-list disputes, audit findings, and RDP edits feed back into rules, prompts, templates, and SOPs.
10. Model-portability layer: a provider-abstraction interface for extraction, retrieval, and drafting so models can be swapped/benchmarked as they improve.

12 AI-vs-human operations pipeline

Everything left of the RDP step is automatable and getting cheaper; the certified-professional step is the durable, license-gated chokepoint.

13 Operations as product

• SOPs per jurisdiction: versioned playbooks for LL97, LL84, WA CBPS, BERDO, BEPS, Energize Denver — updated when rules or cap tables change.
• Structured intake checklists & required-evidence lists: exactly which bills, meters, and GFA documents each building needs before work starts.
• Automated completeness checks: a filing cannot advance until 12 months of data per meter, full meter coverage, and reconciled GFA are present.
• Exception queues & reviewer assignment: master-metered multifamily, district steam, mixed-use, and data-gap cases route to specialists.
• Confidence scoring, audit trails, version control: every number is traceable to a source document and a rule version.
• Gold-standard examples & red-team checks: a library of correct filings; adversarial review of any building near a cap boundary.
• Root-cause & postmortem loop: every rejection, dispute, or missed deadline gets a written root cause that becomes a rule, check, or template.

14 No-holes quality engine

The failure we most fear is a wrong number on a certified filing — which can trigger the up-to-$500,000 false-statement penalty and expose the certifying PE. Controls:

• Deterministic calc, not model math: emissions and penalties are computed by tested code against versioned coefficient/cap tables; the model drafts prose and proposes mappings, it does not do the arithmetic.
• Completeness gates: no filing releases with missing months, unmapped meters, or unreconciled GFA.
• Two-key release: analyst sign-off + RDP certification; boundary cases (within a threshold of a cap) get mandatory second review.
• Source-linked evidence file: every input traceable; nothing enters ESPM/BEAM without a document behind it.
• Anomaly detection: weather-normalized spike detection and year-over-year deltas flag data errors before they reach a filing.
• Change management: when a jurisdiction updates cap tables or coefficients, the rule library is versioned and prior filings are re-validated.

15 Pricing, pricing legality, and unit economics

Model

• Per-building annual managed-compliance fee — benchmarking + report + RDP certification + penalty memo. Target $2,500–$5,000 (NYC LL97, certified); $800–$1,500 for benchmarking-only jurisdictions (Inferred).
• Fixed-fee penalty-exposure & compliance-pathway analysis (current + 2030) — a paid diagnostic that upgrades the free teaser (Inferred).
• Fixed-fee Notice-of-Violation / covered-buildings-list dispute response packages (Inferred).
• Portfolio retainer for multi-jurisdiction owners — volume-tiered per-building rate.

Why not hourly: hourly caps margin, punishes the automation we're building, and misaligns with an outcome buyer. Why not contingency (penalty-avoided): the counterfactual is unmeasurable, it could perversely pressure the calc toward under-reporting emissions (a false filing risking up to $500,000 and the PE's license), and it compromises the certifying professional's independence. Per-unit fixed fees are the clean, defensible structure.

Unit economics (per NYC certified building/year — Inferred)

The RDP minute is the dominant cost, so the whole build is about compressing certification time via gold standards, deterministic calc, and pre-certified templates — moving margin from ~40% at launch toward 60–70% at volume. All figures Inferred; validate in pilots.

16 Nonlinear scaling plan

• Revenue decouples from headcount because the RDP touches a pre-validated, deterministic package — minutes, not hours — and one analyst supervises a growing batch as gold standards and rules accumulate.
• Automation path: ~55% at launch (heavy human data-cleanup), ~75% by day 90 (extraction + meter-mapping SOPs), ~88% by year 1 (rule library + templates mature).
• Cycle time: weeks at launch → days at steady state; the spring deadline crunch is smoothed by year-round intake and off-cycle jurisdictions (WA June, Boston, DC).
• Targets: rework <8%, quality-failure <2%, escalation <10%; margin expands as certification minutes fall and portfolio retainers concentrate volume.

17 Moat & Sam Altman test

Why model improvement compounds us: better multimodal models make utility-bill extraction, whole-building meter reconciliation, multi-jurisdiction rule retrieval, and penalty modeling cheaper and more accurate — dropping our COGS while the PE/RA certification remains a legally required human interface. As we file more, the rule library, gold standards, and covered-buildings intelligence deepen — assets a general model does not have.

Strongest commoditization threat (honest): the emissions arithmetic is deterministic and free estimators already exist (NYC Accelerator, RegWatch, Boiler Room); a future general model could let an owner self-serve the calculation. Anti-commoditization answer: we do not sell arithmetic or a dashboard — we sell the certified, filed, defended, multi-jurisdiction outcome. A general model cannot sign a PE certification, cannot assume the professional liability, cannot manage 60 buildings across a dozen portals and deadlines, and cannot produce the audit-ready record a lender or city demands. The moat is regulation (certification + liability) + operations (portfolio management, defensibility) + accumulated data — not a clever prompt.

18 Buyer-specific go-to-market

The buyer is a professional CRE decision-maker reached by trust-building and precise, data-led outreach — not a viral loop.

• Why this GTM fits: owners are named in public data and clustered in known associations; the sale is consultative and deadline-driven.
• First 50 prospects: owners of NYC buildings shown over their 2024 cap in the public LL84/LL97 data, ranked by exposure; plus WA Tier 1 (>220k sq ft) owners who just passed June 1, 2026.
• Trigger events: pending May 1 / June 1 deadlines, covered-buildings-list notices, over-cap public data, lender/ESG diligence, prior late-filing penalties.
• Channels: data-driven outbound off public lists; property-management & owner associations (BOMA, IREM, REBNY, and multifamily owner groups); managing agents (one relationship = many buildings); C-PACE lenders and brokers as referral partners; compliance webinars; and answer-engine optimization so "LL97 filing / WA Clean Buildings deadline" queries surface us.
• Credibility asset: a named RDP of record + a free portfolio penalty-exposure report generated from public data.
• Conversion path: free exposure report → paid pathway analysis → annual managed-compliance contract → portfolio retainer.
• Metrics: report-to-call rate, call-to-pilot rate, pilot-to-annual conversion, buildings per logo; sales cycle ~30–75 days into the deadline, shorter under penalty pressure.

Lead magnet & content

The wedge lead magnet is a free Portfolio Penalty-Exposure Report: paste addresses/BBLs, we pull public LL84 data, apply the LL97 cap tables, and return a ranked list of buildings, current vs 2030 exposure, and filing status. Founder/expert content teaches "how LL97 penalties actually accrue," "the meter-data mistakes that inflate your number," and "the multi-jurisdiction deadline map" — the same organic pieces later become paid-ad creative.

19 Pilot design & early-demand-trap mitigation

• Pilot cap: 10 owners / ~150 buildings across the 2026 NYC May-1 cycle and the WA June-1 Tier 1 cohort.
• Pilot profile: mid-market owners with a mix of clean and messy (master-metered, mixed-use, district-steam) buildings to stress the engine.
• Success criteria: 100% on-time certified filings; measured penalty-exposure reduction; <8% rework; RDP certification minutes trending down.
• Manual-workaround tracking: log every hand-fix (odd meter configs, DEP water pulls, GFA disputes); anything recurring becomes an SOP, rule, or automation — not a permanent human patch.
• Process-hardening gates: after 5 buildings harden intake & completeness checks; after 10 harden SOPs, exception queues, and reviewer checklists; after 20 pause new pilots until COGS, rework, escalation, and cycle time are measured.
• Kill signal: if RDP certification time cannot fall below the level that supports 50%+ margin, or if data-access (whole-building meter data) proves structurally unavailable at scale, stop and re-scope.

20 Competitive landscape

Whitespace: a done-for-you, RDP-certified, multi-jurisdiction managed filing service for mid-market owners — nobody in the scan occupies it cleanly.

21 Regulation, compliance & licensing boundary

• What we must not claim: that we can certify without an RDP, guarantee a specific penalty outcome, perform engineering design, or give legal advice.
• Controls: engagement letters defining the RDP's role and responsibility; owner consent to pull utility/DEP data; encrypted storage; access controls; complete audit logs; professional-liability (E&O) coverage for the certifying professional.
• Pricing legality: per-unit fixed fees avoid the independence and false-filing hazards of contingency pricing (see §15). No fee is tied to a certified number.
• Privacy: utility and tenant energy data is sensitive; a data-processing agreement, minimization, and retention limits are required.

22 Compact founding team & expert map

23 Exhaustive risk register

24 Tech stack & build plan

• Intake & storage: document store + object storage for bills/PDFs; structured DB (Postgres) for meters, buildings, filings, deadlines; per-building record with version history.
• Extraction: multimodal LLM + OCR for utility bills and GFA docs; a review UI for meter↔account↔BIN mapping with human confirm.
• Rules engine: versioned, unit-tested modules for emissions coefficients, cap tables (LL97 by occupancy group), WA EUI targets, penalty formulas, deadlines, fees.
• Knowledge/retrieval: jurisdiction rule library with citations; ESPM property-type mappings; change-log when rules update.
• Integrations: ENERGY STAR Portfolio Manager web services; portal-assist for BEAM / DOB NOW / WA Commerce; utility data-request automation.
• Ops layer: exception queues, reviewer assignment, confidence scoring, audit-trail, client status reporting.
• Build sequence: (1) NYC LL97 + LL84 end-to-end for the 2026 cycle; (2) WA CBPS Tier 1; (3) Boston BERDO / DC BEPS / Denver; (4) portfolio dashboard & retainer tooling; (5) model-portability abstraction and automation hardening.

25 Metrics & KPIs

Also tracked: certification minutes per building, buildings per operator per day, cycle time, customer acceptance rate, penalty-exposure reduction, exposure-report → contract conversion, and pilot-to-annual conversion.

26 What could kill this

• RDP economics don't compress: if certification time/liability can't fall enough, margin stalls below 50% — the single biggest threat.
• Data access is structurally blocked: if whole-building utility data can't be obtained reliably at scale, the engine can't run.
• An incumbent bundles certified fulfillment: a Measurabl-scale player adds PE certification and mid-market pricing before we win share.
• Enforcement collapses: broad rollback/non-enforcement of BPS penalties guts willingness to pay (partially hedged by benchmarking mandates and multi-jurisdiction spread).
• Commoditization outruns the moat: owners self-serve the calc and cities relax certification requirements.

27 90-day validation & launch plan

• Weeks 1–2: build the free Portfolio Penalty-Exposure Report off public LL84/LL97 data; secure a fractional NYC RDP; draft engagement letters and the RDP-role boundary.
• Weeks 3–4: stand up NYC LL97+LL84 intake → extraction → deterministic calc → report draft; unit-test cap tables and coefficients against known filings.
• Weeks 5–6: outbound to 50 over-cap NYC owners + WA Tier 1 owners; convert 10 pilot logos (~150 buildings); run paid pathway analyses.
• Weeks 7–10: file the 2026 cycle end-to-end with RDP certification; instrument certification minutes, rework, escalation, and data-access blockers; log every manual workaround.
• Weeks 11–13: harden SOPs/exception queues at the 5/10/20 gates; measure COGS and margin; run a small pricing test; decide expansion order (WA → Boston/DC/Denver).
• Kill criteria: certification minutes not on a path to 50%+ margin; data access unresolved for >25% of pilot buildings; exposure-report → contract conversion below a viability threshold.

28 Sources

• NYC Accelerator — Local Law 97 overview ($268/ton, Article 320/321, RDP certification, May 1 filing)
• NYC DOB — LL97 GHG emissions violations & penalties ($0.50/sq ft·mo; Article 321 $10,000)
• NYC DOB — LL84 Benchmarking Law & Covered Buildings List 2026
• NYC DOB — LL97 Covered Buildings List FAQ (BIN-level compliance pathways)
• NYC DOB — Sustainability Laws & Covered Buildings List 2026 (Excel)
• NYC Open Data — Building Energy & Water Data Disclosure (LL84), public prospect dataset
• ENPG — LL97 filing via RDP/PE through BEAM; DOB fees $210 / $615 / $60
• Washington State Dept of Commerce — Clean Buildings Performance Standard (CBPS)
• WA Commerce — CBPS Tier 1 compliance schedule (June 1, 2026 / 2027 / 2028)
• WAC 194-50-150 — CBPS reporting & penalties ($5,000 + $1/sq ft·yr)
• National BPS Coalition — 49 governments (7 states, 42 localities)
• IMT — 2026 Building Policies Outlook (16 BPS laws; ~60 benchmarking jurisdictions; 92%/43% NYC data)
• IMT — Building energy benchmarking & BPS maps (>35 jurisdictions, billions of sq ft)
• Green Check Solutions — BPS deadline tracker (Boston BERDO, DC BEPS, Denver Energize, Oregon)
• Measurabl Comply — customer-operated BPS/BEPS compliance software (competitor)
• nZero — AI energy-optimization platform for commercial real estate (competitor)
• Skybriz — LL97 $268/ton penalties explained (annual accumulation)
• RegWatch — free LL97 baseline & consultant referral (demand + lead-gen validation)
• ENERGY STAR Portfolio Manager — the benchmarking platform used across jurisdictions
• Surplus-lines compliance by state (candidate comparison — $115B premium 2025; $25k/violation)
• IRS — Instructions for Form 1099-DA (candidate comparison — crypto broker reporting timeline)

Penalty amounts, deadlines, the RDP/PE certification requirement, jurisdiction and covered-building counts, the public prospect datasets, and the incumbent landscape are Verified via the sources above (retrieved 2026-07-01). Per-building fees, COGS, gross margin, revenue/FTE, automation rates, certification-minute compression, sizing, and switching willingness are author estimates labeled Inferred or Unverified in §7/§15/§16 and must be validated in pilots before decisive use. This is a hard-to-fool blueprint, not a guarantee.