GateTRU — AI-Native CARB TRU Applicable Facility Quarterly Reporting Engine

01 Thesis

California’s TRU ATCM (13 CCR 2477) forces ~6,100 “applicable facilities” — refrigerated warehouses ≥20k sq ft, grocery stores ≥15k sq ft, seaports, and intermodal railyards — to either (1) report every diesel TRU that enters the fence line each quarter to CARB, or (2) swear under penalty of perjury that only compliant TRUs operate on site. Option 2 is commercially impractical for perishable receivers. Option 1 is a recurring, document-heavy data-assembly job that facilities already outsource or scramble to do. GateTRU sells the filed quarterly Applicable Facility TRU Activity Report pack (CARB template completed, QA’d, emailed to arber@arb.ca.gov, with receipt proof) — not a customer-operated dashboard. AI extracts CARB IDNs, carrier names, and USDOT/CA numbers from gate logs, BOLs, TMS exports, and label photos; a human specialist reviews exceptions; the facility authorized signer remains the customer-facing trust interface.

02 Discovery rationale

Research terrain this hour spanned OSHA ITA, TSCA 12(b), Form SD, EU CBAM, FFATA, MoCRA, Form ADV, MSGP, AB 2013, IRIS/FIRE 1099, CPO-PQR, PCI SAQ, and CARB TRU. Most were duplicates of prior blueprints (MoCRA, CPSC eFiling, Form 990, RIA ADV program, EPCRA/TRI, 1099-DA) or failed the evidence bar (AB 2013: no public DFY unit price; GHGRP: rescission risk; CPO-PQR: no public ASP). GateTRU cleared because: (1) primary regulation + FAQ deadlines are explicit; (2) CARB App F sizes ~6,119 facilities; (3) Trucdocs publishes $980/facility/year DFY pricing; (4) Benesch documents heightened 2026 enforcement and why quarterly reporting is the pragmatic path; (5) workflow is digital/document-based with no required site visit for the reporting wedge; (6) anti-dup vs AIM/LeakClear (refrigerant leak) and EPCRA/TRI is statute-crisp.

03 Candidate comparison

04 Hard disqualifier check

05 Rubric scorecard

Gates average: 4.33 / 5. Extended scores (urgency 5, budget proof 4, MVP clarity 5, distribution 4, licensing 5) support proceed.

06 Opportunity

07 Evidence quality & source-claim matrix

08 Why now

• Verified regulatory: 2022 TRU ATCM amendments created applicable-facility duties effective 12/31/2023; quarterly reporting live since Apr 15, 2024.
• Verified enforcement: 2026 trade counsel alerts describe heightened CARB enforcement and fee invoices supporting enforcement capacity.
• Inferred AI capability: Vision + OCR + entity resolution now make gate-photo and BOL extraction accurate enough for specialist QA rather than full manual typing.
• Unverified: Exact share of facilities currently late/non-reporting (pilot diagnostic will measure).

09 Customer & PMF

10 The outcome we sell

11 Internal AI engine architecture

1. Intake: Secure upload of gate logs, TMS exports, BOLs, camera-capture label photos, carrier master lists.
2. Normalization: Parse timestamps, dock doors, tractor/trailer IDs; dedupe same TRU same day; timezone normalize.
3. Retrieval/knowledge: Index 13 CCR 2477.17/20(m) field definitions, CARB FAQ rules, prior quarter gold packs, ARBER public compliance search results when available.
4. AI workbench: OCR CARB IDN / labels; NER for carrier legal name; USDOT regex; fuzzy match to carrier roster; flag missing fields.
5. Deterministic rules: Required columns non-null; date within quarter; IDN format checks; duplicate suppression; Option-1 vs Option-2 eligibility gates.
6. Human chokepoint: Specialist reviews low-confidence (<0.85) rows; customer signer attests.
7. QA: Completeness score, sampling red-team, cross-check vs prior quarter carrier patterns.
8. Delivery: Populate CARB Excel/CSV template; email submit; archive proof PDF.
9. Learning loop: Corrections → prompt rules, carrier alias graph, OCR templates.
10. Model portability: Swap vision/LLM providers behind schema contracts; keep rules + gold packs owned.

12 AI-vs-human operations pipeline

13 Operations as product

• SOP: Quarterly Close Runbook (T-14 / T-7 / T-2 / T+0).
• Structured intake checklist + required evidence list (gate log schema, photo standards).
• Automated completeness checks before human touch.
• Exception queues: missing IDN, unknown carrier, OCR conflict, out-of-quarter timestamp.
• Reviewer assignment by facility vertical (grocery vs WHDC vs seaport).
• Confidence scoring per row; audit trail of every edit; versioned gold packs.
• Postmortem on CARB bounce/follow-up or customer dispute within 48h.

14 No-holes quality engine

Prevent bad filings via dual barriers: (1) deterministic schema blockers that refuse submit if required fields empty; (2) human QA on every low-confidence cluster. Hallucinated IDNs are blocked by format validators + optional ARBER lookup when public. False completeness is caught by comparing row count to gate event count (± tolerance). Customer attestation language makes clear GateTRU prepares; facility owns truth of on-site events.

15 Pricing, pricing legality, and unit economics

Hourly billing rejected. Contingency/success-fee not used (compliance filing, not recovery). Pricing is fixed outcome fees — legal and ordinary for administrative compliance vendors (Trucdocs precedent).

COGS per Quarterly Pack (launch assumptions)

• Model/OCR inference: $3–$12
• Hosting/storage: $1–$3
• Human review: 35–55 min @ $45–$65 fully loaded → $26–$60
• QA specialist: 10–15 min → $10–$20
• Support/follow-up: $5–$15
• Total COGS: ~$45–$110 on $490–$890 price → gross margin ~78–90% at launch on clean sites; messy sites pull toward 55–70% until automation rises.

Revenue/FTE target year 1: $350k–$500k with 2 operators + 1 specialist QA.

16 Nonlinear scaling plan

• Automation %: launch 45% → day 90 65% → year 1 80% of row production.
• Throughput: 4–6 facility packs/operator/day → 10–14 after templates/carrier graph mature.
• Cycle time: 3–5 business days → <48 hours for returning sites.
• Rework <5%; escalation <8%; quality failure <1% of filed packs.
• Margin expands as carrier alias graph and OCR templates reduce minutes/row.

17 Moat & Sam Altman test

Frontier models improve OCR/NER → lower COGS and higher completeness → GateTRU gets stronger. Commoditization threat: CARB builds a free portal with mobile capture. Mitigation: own the facility operating system (multi-dock intake SOPs, carrier graph, multi-site rollups, audit trails) and stay the trusted human interface. Generic ChatGPT cannot file a defensible pack across messy TMS schemas without this ops layer.

18 Buyer-specific go-to-market

Primary motion: Founder-led outbound to CA refrigerated WHDC / grocery DC EHS+Ops lists + warm intros via cold-storage associations + diagnostic lead magnet. Waitlist/AEO secondary.

• First 50 prospects: CA WHDCs 20k–150k sq ft in Inland Empire, Central Valley, Bay Area, LA County (built from business licenses + cold-storage directories).
• Triggers: Quarter close −21 days; new DC opening; CARB fee invoice; missed prior filing.
• Outreach wedge: Free “Q-1 Gate Completeness Diagnostic” (sample 2 weeks of gate data → missing-field scorecard).
• Credibility asset: Annotated redacted pack + CCR citation one-pager.
• Conversion path: Diagnostic → quarterly pilot ($490–$690) → annual managed.
• Sales cycle: 2–6 weeks. Activation = first on-time filed quarter.

Distribution proof table

First 30 days of content

• 10 posts: Option 1 vs 2; quarterly calendar; what is an applicable facility; IDN capture at the gate; USDOT pitfalls; penalty math; Trucdocs-priced alternatives; multi-site rollups; photo standards; seaport vs grocery differences.
• 3 teardowns: anonymized incomplete pack; OCR IDN failures; carrier alias collisions.
• 2 lead magnets: Gate Completeness Checklist; Quarterly Close Runbook PDF.
• 1 webinar: “15-day sprint to a CARB-ready quarterly TRU pack.”
• 1 outbound diagnosis template: missing-field scorecard email.

19 Pilot design & early-demand trap mitigation

• Pilot cap: 8 facilities / first quarter cohort.
• Incentive: 20% off first quarter if customer provides structured gate export.
• Success criteria: ≥7/8 on-time filings; median human minutes ≤50; NPS ≥40; zero CARB rejection for GateTRU errors.
• Custom work policy: No SWPPP-style consulting; no carrier refusal programs; only reporting pack improvements that become SOPs.
• Hardening gates: After 5 sites → freeze intake schema; after 10 → SOP + exception taxonomy; after 20 → pause until COGS/rework measured.

20 Competitive landscape

• Trucdocs: Public $980/facility/year DFY — validates budget; lighter AI ops story.
• Be CARB Compliant / general fleet consultants: Broader TRU/ARBER fleet registration; less facility-gate specialization.
• Big EHS firms: Expensive, slow, not productized for quarterly packs.
• DIY: Free CARB template — high labor, high miss rate.
• GateTRU whitespace: AI-native facility-gate extraction + fixed quarterly outcome + multi-site portfolio OS.

21 Regulation, compliance, and licensing boundary

• AI may: Extract, classify, draft template rows, flag missing fields, draft submission email.
• Operators may: QA packs, chase missing evidence, operate portals under customer entitlement.
• Facility authorized signer must: Attest accuracy; remain responsible party under 2477.17/2477.19.
• GateTRU must not claim: To be CARB, to certify TRU equipment compliance, to provide legal defense, or to guarantee no enforcement action.
• Controls: Written MSA + attestation, audit logs, data retention ≥5 years, CA privacy posture for yard imagery.

22 Compact founding team and expert map

23 Exhaustive risk register

24 Tech stack & build plan

• Day 0: Google Workspace + shared Drive, Notion SOPs, Typeform/intake, Stripe billing, secure S3/GCS uploads.
• Week 1–2: Python ETL + vision OCR (portable), carrier alias DB (Postgres), template writer (openpyxl).
• Week 3–4: Exception queue UI (simple internal Next.js or Retool), confidence scores, audit log.
• Day 90: Customer portal for status only (not a co-pilot); Zapier/email automations for T-14 reminders.
• Do not build: public self-serve AI product; CARB portal clone; yard hardware.

25 Metrics & KPIs

• Throughput packs/operator/day; cycle time; rework %; gross margin; COGS/pack; revenue/FTE.
• Escalation %; automation %; evidence completeness %; quality failure %; customer acceptance %.
• Diag→pilot %; pilot→annual %; on-time filing %; logo retention quarterly.

26 What could kill this

1. CARB launches free mobile capture that makes DFY unnecessary.
2. Facilities systematically pick Option 2 despite perishable risk.
3. Gate data quality so poor that minutes/pack never fall below ~90.
4. Trucdocs + consultants saturate mid-market before differentiation lands.
5. Enforcement lull removes urgency for 2+ years.

27 90-day validation and launch plan

Days 1–7

• Legal MSA/attestation draft; intake schema; 30 prospect list; diagnostic landing page.

Days 8–30

• 3 design partners; OCR pipeline MVP; first paid quarterly pack; measure minutes/row.

Days 31–60

• Expand to 8 pilots; harden exception taxonomy; publish 2 teardown posts; association outreach.

Days 61–90

• Convert ≥4 to annual; freeze custom work; decide kill/continue on COGS <$110 and on-time ≥95%.

Kill criteria: Cannot get 5 paid facilities in 90 days; median human minutes >90 after 3 iterations; CARB rejects >2 packs for GateTRU errors.

28 Sources

1. CARB — Applicable Facility Requirements
2. CARB — Applicable Facility Requirements FAQ
3. CARB — TRU App F Applicable Facility Determination (~6,119 facilities)
4. 13 CCR § 2477.17 — Applicable Facility Owner Requirements
5. 13 CCR § 2477.19 — Non-Compliance and Penalties
6. 13 CCR § 2477.21 — Fees ($45 / 3 years)
7. Trucdocs — DFY TRU facility compliance pricing ($980/facility/year)
8. Benesch Law — Heightened CARB TRU enforcement 2026
9. CARB — TRU Forms (registration / templates)
10. CARB — 2022 Amendments to the TRU ATCM fact sheet
11. Be CARB Compliant — TRU / facility reporting services
12. CARB — TRU ATCM FAQ (reporting schedule Table 4)

Hard-to-fool blueprint. Claims labeled Verified / Inferred / Unverified. Unverified volume assumptions must be measured in pilots before scale.