GateProof — Gateway 2 submission validator for HRB architects

1. One-liner

Pre-flights your Gateway 2 evidence pack against the BSR’s rejection criteria so it clears validation first time.

2. Trend signal — why now?

The UK’s Building Safety Regulator (BSR) moved to a validation-led model in early 2026: a Gateway 2 application for a higher-risk building (HRB) is now screened for completeness, clarity, and coordination before anyone reviews the actual design. Fail that screen and you’re bounced — not for bad design, for bad paperwork.

The numbers are brutal and well-documented:

• 29% of Gateway 2 applications are failing — and analysis shows the majority of failures are driven by “evidence architecture” (fragmented, uncoordinated, incomplete evidence), not design inadequacy. (London Construction Magazine, May 2026)
• The single biggest rejection category is Principal Designer (Building Regs) competence evidence — applications confirm the appointment but don’t evidence active engagement, competence, or outputs. (London Construction Magazine, May 2026)
• A rejection isn’t a slap on the wrist: “resubmission can add months to your programme, invalidate procurement processes and trigger significant commercial penalties.” (Construction Management, 2026)
• 18,000 homes have been stalled by Gateway 2, with architects named as the most affected discipline. (Architects’ Journal, 2026)
• Volume is real and recurring: ~740 building-control approval applications in the 12 weeks to 1 May 2026 (~247/month), 1,367 live applications, and 36,984 residential units sitting in live cases. (BSR / BCIS application data, May 2026)

What changed in the last 12 months: (a) the BSR flipped to validation-led screening, turning “evidence architecture” into the gating risk; (b) LLMs got good enough to read a multi-document submission and cross-check it against a published rubric; (c) the BSR and RIBA published explicit guidance on what a compliant submission looks like — a public spec a tool can encode.

Existing software in this space (SymTerra, BlockPro, Multivista, Matterport, Egnyte, Trimble) is about storing and capturing the golden thread for occupied buildings (Gateway 3 / in-life). Nobody is attacking the pre-submission validation gap that causes the 29% bounce.

Provenance:
  - Signal 1 (Demand): 29% Gateway 2 rejection rate, driven by evidence quality not design; PD competence evidence is the #1 failure category; rejection = months of delay + commercial penalties — https://www.constructionmagazine.uk/2026/05/gateway-2-rejection-rate-evidence-failures-london.html — 2026-05
  - Signal 2 (Feasibility): BSR operates a validation-led completeness/clarity/coordination screen since early 2026; RIBA + BSR published explicit submission guidance an LLM can encode and check against — https://www.riba.org/work/insights-and-resources/professional-features/what-can-the-latest-building-safety-regulator-figures-for-gateway-2-applications-tell-architects/ — 2026
  - Signal 3 (Economic): ~247 Gateway 2 applications/month, 1,367 live cases, 36,984 units in pipeline; 18,000 homes stalled; every stalled programme carries financing + penalty exposure — https://www.bcis.co.uk/news/latest-building-control-approval-application-data/ — 2026-05
  Category: Regulatory arbitrage

3. The opportunity

The BSR turned a design-quality regime into, functionally, a document-coordination exam — and 29% of professional applicants are failing the exam on a technicality. The incumbents sell filing cabinets (golden-thread record-keeping for buildings already approved). The gap is a pre-flight check: before you hit “submit” on the BSR portal, run your evidence pack through a tool that knows exactly what BSR validators bounce applications for and tells you the specific gaps — missing PD competence outputs, an uncross-referenced fire strategy, a design-decision log that doesn’t trace to the safety case.

This is not a disruption-of-incumbent play; it’s a white-space wedge. The “competitor” is the status quo: a £5,000–£15,000 specialist consultant doing a manual completeness review, or an architect crossing their fingers and eating a multi-month resubmission. AI does the cross-referencing in minutes that a senior consultant does in days.

4. Target market

• Primary customer: Architecture practices and Principal Designer (Building Regulations) consultancies in England that submit Gateway 2 applications — typically 5–50-person practices with an HRB pipeline, plus the boutique “Gateway 2 specialist” firms now emerging (e.g. Folly Architects, Urbanist). The buyer is the practice director / lead PD who personally carries the rejection risk.
• Why they buy: In their words — “incomplete, uncertain or poorly coordinated submissions are not just delayed; they are routinely rejected.” A bounce burns the practice’s fee margin, blows the client’s programme, and damages the relationship. They will pay to not be in the 29%.
• Rough TAM reasoning: ~3,000 Gateway 2 applications/year flow through the BSR, concentrated among a few hundred active HRB practices and PD consultancies, plus the wider pool of fire engineers and contractors who assemble the same evidence. Add Gateway 3 and the swelling remediation pipeline. It’s a niche — hundreds-to-low-thousands of accounts, not tens of thousands — which is exactly the bootstrapped-operator sweet spot and far too small for a VC-scale player to bother with.
• Why now for them: The validation-led screen is new (early 2026), the rejection data is now public and frightening, and 36,984 units are stuck. The pain is at its peak this year.

5. Product sketch (MVP)

• Submission pre-flight: Upload the Gateway 2 evidence pack (drawings, design-decision log, fire strategy, structural rationale, PD/PC appointment + competence evidence, building regs compliance statements). GateProof returns a pass/fail-style readiness report.
• Rejection-risk flags: Each flag maps to a known BSR validation failure mode — e.g. “PD competence: appointment confirmed but no evidence of active design-safety engagement or outputs” — with the exact criterion cited.
• Cross-reference checker: Confirms the fire strategy, structural design, and design-decision log actually reference each other and the safety case (the “coordination” the BSR screens for).
• Completeness matrix: A live checklist of every document and data element the BSR validation stage expects, with what’s present, missing, or weak.
• PD competence pack builder: Guided assembly of the Principal Designer competence evidence — the #1 rejection category — into the format BSR validators expect.
• Resubmission diff: If you’ve already been bounced, paste the BSR’s clarification request; GateProof maps each point to the document that must change.
• Audit-ready export: A coordinated, indexed evidence bundle plus a readiness certificate the practice can show the client before submission.

6. AI angle — what’s load-bearing

Remove the AI and this is a static PDF checklist — useless against a 500-page, multi-document, cross-referenced submission. The load-bearing work is an LLM reading the actual contents of a heterogeneous evidence pack and judging it against the BSR’s validation rubric: does the design-decision log actually trace fire-safety choices to the strategy? Does the PD competence evidence demonstrate engagement or just assert appointment? That’s semantic comprehension and cross-document reasoning — exactly what got good enough in the last 12 months and exactly what a checkbox app can’t do. The AI is the product.

7. Localization angle (if any)

This is the localization play — England-specific regulatory arbitrage. The BSR, the three-gateway regime, the Principal Designer (Building Regs) role, and the validation criteria are unique to England’s Building Safety Act 2022. A generic global “construction document management” tool cannot compete because it doesn’t encode the BSR rubric. The moat is regulatory specificity. Adjacent expansion later: Scotland/Wales building-safety regimes, then the EU’s parallel building-safety directives.

8. Business model — path to $1M–$5M ARR

• Pricing: Hybrid. A practice subscription (£300–£600/mo) for unlimited pre-flights + a per-submission “certified readiness review” at £750–£1,500 per application (anchored well below the £5,000–£15,000 a human consultant charges for the same completeness review).
• ACV: A modestly active HRB practice runs 5–15 Gateway 2/3 submissions a year. At £450/mo base + 8 certified reviews × £1,000, ACV ≈ £13,400 ($17K).
• Rough math to $1M ARR (~£790K): ~60 active practices at ~£13K ACV. Plausible within the few-hundred-account market.
• Rough math to $5M ARR (~£3.9M): ~290 practices, OR broaden the unit to every evidence-assembling discipline (fire engineers, contractors, PD consultants) and add Gateway 3 / remediation submissions, which dwarf new-build volume. Requires capturing a meaningful slice of the active HRB ecosystem and expanding the document types covered.
• Expansion path: land on Gateway 2 pre-flight → expand to Gateway 3 (completion) and the golden-thread handover → sell per-seat across the practice → upsell remediation submissions (the larger, longer-tail pipeline).

9. Go-to-market wedge — first 100 customers

• Scrape the failure list. The BSR/BCIS data and trade-press name affected projects and practices. Build a list of the few hundred practices and PD consultancies with live or rejected Gateway 2 applications. Cold email the director: “29% are getting bounced on evidence, not design. Send us your next pack — we’ll show you the flags free before you submit.” A free pre-flight on a real, live submission is an irresistible, zero-risk demo.
• RIBA / CIAT / FIA channels. RIBA, CIAT (Principal Designer Register), and the Fire Industry Association are all publishing anxious Gateway 2 guidance right now. Run a co-branded “Why 29% fail and how to pass” webinar; the audience is pre-qualified and terrified.
• Partner with the Gateway 2 specialist boutiques. Firms like Folly and Urbanist market themselves as HRB/Gateway 2 specialists — sell GateProof as their white-label QA layer so they scale beyond billable hours.
• LinkedIn + trade press. The 29% stat is doing the rounds in Construction Management, Architects’ Journal, RIBAJ. Publish a teardown of an anonymised rejected submission with the flags GateProof would have caught. The community shares it.
• Conversion math: A list of ~400 active practices, a free-pre-flight offer with a genuine “we found 6 things that’d get you bounced” payload, should convert far above cold-SaaS norms because the cost of not trying it is a six-figure programme delay. 100 paying accounts is a realistic 6–9 month target.

10. Build complexity — justification

Medium. The hard part is not engineering — it’s encoding the BSR validation rubric and rejection taxonomy accurately, which needs a domain expert (an architect or building-safety consultant who’s been through Gateway 2). Document ingestion (PDFs, drawings, mixed formats) plus LLM cross-document reasoning over long context is standard-but-non-trivial. No regulatory approval to build — it’s a private prep tool, not a submission gateway. A pair (one technical, one domain) ships a credible v1 in ~3–4 months.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	A private completeness-check tool; doesn’t submit to or impersonate the BSR.
Ethical — no harm / dark patterns	✅	Improves safety-case quality; aligned with regulator intent. Must be clear it’s an aid, not a guarantee.
Market exists (evidence above)	✅	29% rejection, 18,000 homes stalled, ~247 apps/month, public data.
1–5 person team can build this	✅	Pair (technical + domain) in 3–4 months.
Launchable with <$50K / ₹40L	✅	Off-the-shelf LLM APIs, standard web stack, no capex.

All five pass.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	16/20	Hair-on-fire for the affected practice — a bounce is months of delay and a margin-eating penalty. Not daily pain (submissions are episodic), which caps it below 17.
Demand evidence	15	13/15	Multiple independent, dated 2026 signals: hard rejection %, named affected discipline, public volume data, trade-press alarm. A skeptic nods.
Build feasibility	15	11/15	Standard stack + long-context LLM reasoning over messy multi-doc packs; real work to encode the rubric accurately. Pair in 3–4 months.
Distribution clarity	15	12/15	Named, finite, reachable buyer list; warm pro-body channels; irresistible free-pre-flight wedge. Slight uncertainty on conversion.
Revenue mechanics	15	10/15	Pricing anchors cleanly below human consultants; ACV healthy. But the account universe is genuinely small — $5M ARR needs ecosystem expansion, not just new-build practices.
Time to first revenue	10	7/10	A free pre-flight that finds real flaws converts fast; episodic buying cadence and trust-building on a high-stakes deliverable add some lag. 6–10 weeks to first paid.
Defensibility	10	6/10	Soft moat: accumulating rejection-pattern data + rubric encoding + pro-body relationships + workflow lock-in. Copyable in ~12 months by a determined competitor; the data flywheel and brand-in-niche are the defense.
Total	100	75/100

13. Qualitative modifiers

Founder-fit tags

domain-expertise-required · technical-heavy

This needs a co-founder or close advisor who has personally survived a Gateway 2 submission. Without that, the rubric encoding will be wrong and the tool will lose trust on its first false-negative.

Key assumptions to validate (3–5)

1. Assumption: Practices will trust an AI pre-flight enough to act on its flags before a high-stakes submission. How to test: Run free pre-flights on 10 real, live packs; measure how many flags the practice agrees are genuine and would fix.
2. Assumption: The BSR validation rubric is stable and encodable enough that GateProof’s flags correlate with actual rejection reasons. How to test: Back-test against a set of already-rejected applications — does GateProof catch the same gaps the BSR cited?
3. Assumption: Willingness to pay ~£1,000/submission + a monthly base. How to test: Offer paid certified reviews to 15 practices after the free pre-flight; measure close rate at the price.
4. Assumption: The account universe (hundreds of practices) supports a $1M+ business at this ACV. How to test: Build the actual named list from BSR/BCIS data and pro-body directories; count reachable, qualified accounts.

Risk flags

1. Regulatory risk (timing): The BSR is actively reforming — a “batching pilot” already cut London approval times from ~48 to ~12–14 weeks, and the 29% rate may fall as the regime matures. If validation gets easy, the pain shrinks. Mitigate by expanding into Gateway 3 / remediation, where the backlog is larger and longer-lived.
2. Liability / trust: A false “you’re ready” that precedes a real rejection destroys credibility. Position firmly as a decision aid, cap claims, and let the human PD sign off — never imply a guarantee.
3. Small TAM: This is a niche by design. $1M–$3M ARR is very achievable; $5M needs disciplined expansion across disciplines and gateways. Wrong idea for someone chasing scale; right idea for a bootstrapper.
4. Platform dependency: Relies on the BSR’s published criteria and the practice’s willingness to share sensitive packs — handle data security as a first-class feature, not an afterthought.

14. Structured verdict

Score:                  75/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder paired with a building-safety / architecture domain expert who has run a Gateway 2 submission
Time to revenue:        6–10 weeks (free pre-flight → paid certified review)
Capital to launch:      £15–30K ($20–40K) — LLM API spend, a domain advisor, standard web stack
Top 3 assumptions to validate first:
  1. GateProof's flags match real BSR rejection reasons — back-test against already-rejected applications
  2. Practices trust and act on AI flags before submitting — run 10 free live pre-flights, measure agreement
  3. WTP at ~£1,000/submission + monthly base — convert 15 free pre-flights to paid reviews
Kill criteria:
  - Abandon if <40% of flags on back-tested rejected applications match the BSR's actual cited reasons
  - Abandon if <3 of 15 free-pre-flight practices convert to a paid review
  - Abandon if BSR reforms push the rejection rate below ~10% before launch and the Gateway 3 / remediation pivot shows no equivalent pain

15. Next step — 1-week validation sprint

• Day 1–2: Pull the BSR/BCIS application data and trade-press coverage; build the named list of practices with live or rejected Gateway 2 applications. Collect every public BSR + RIBA validation-guidance document into a single rubric draft.
• Day 3–4: Get 3–5 real Gateway 2 packs (anonymised) — at least one already rejected. Hand-run them against the rubric (no product yet, just a checklist + an LLM in a chat window) and see whether the flags would have predicted the actual rejections.
• Day 5: Email 30 directors offering a free pre-flight on their next submission. Go/no-go: proceed only if (a) the manual back-test catches ≥40% of the real cited rejection reasons on the rejected pack, AND (b) ≥5 practices say yes to a free pre-flight. Falsifiable: if the rubric can’t predict real rejections, the core premise is dead — stop.