Two products · Live in NYC

Stop refiling the same document.

PermitCheck AI and TitleCheck AI read the PDFs that real estate runs on, from permit sets to title commitments to surveys, and catch the issues that send filings back. Every finding lands with the exact code section behind it.

40–60%

permits rejected first pass

50K+

code sections checked

3–6 wks

saved per rejection

PermitCheck_AI / Floor_Plan_A2.pdf
Live

Sheet A-101 · Rev 03

Compliance Punchlist

3 issues

Corridor width below minimum

2nd floor corridor measures 38" clear. Minimum 44" required.

BC §1005.1

Exit separation distance

Required exits are 18' apart. Need ½ diagonal min.

BC §1007.1.1

Door swing into egress path

Stairwell B door reduces corridor to 32" when open.

BC §1005.2
TitleCheck_AI

Variance Report

Easement (Sched B-II.4)Mismatch
Legal descriptionDelta
Encroachment lineMatch
The math we kept hearing
Filed · Rejected · Refiled

One rejection. One missed variance. Real money.

Every expediter and attorney we talk to knows this math by heart. One avoidable mistake that reaches an examiner's desk costs more than a year of pre-checking ever will.

Cost of one rejection cycle

Rework & resubmission$500–$2,000+
Project delay3–6 weeks
Idle project carrying cost$5K–$50K+
Client trust erosionIncalculable

Either of our tools costs a fraction of a single rejection cycle. One caught violation pays for itself.

The portfolio

Two products. One thesis.

Compliance is a search problem. Whoever knows the rules and reads the documents fastest wins. We built two tools that put that knowledge into software.

Index of sheets
01PermitCheck AINYC DOB pre-check
02TitleCheck AITitle & ALTA variance
PermitCheck AI · NYC DOB pre-check
Product · Live

PermitCheck AI

Know what DOB will reject — before you submit.

Reads architectural plan sets and checks them against NYC building, zoning, and fire code. Returns a punchlist with verbatim code citations and suggested fixes in minutes.

  • Catches exact violations an examiner would flag, with §code citations
  • Hybrid deterministic + LLM checks (corridors, egress, clearances, fire)
  • Works with standard PDF plan sets — no BIM, no CAD, no integration
  • Designed for first-pass DOB approval, not after a 3-week rejection cycle

Built for

Permit expediters · Architecture firms · GCs filing in NYC

Walkthrough · ~2 min
PermitCheck AI

40–60%

NYC permits rejected first try

Minutes

from upload to punchlist

§Citations

verbatim, with page refs

TitleCheck AI · Title & ALTA survey variance
Product · On request

TitleCheck AI

Title + ALTA Survey Variance Analyzer

Find every variance between the commitment and the survey.

Cross-references title commitments against ALTA surveys and supporting documents. Surfaces mismatches — easements, encroachments, legal description deltas — with page-level evidence and citation back to source.

  • Side-by-side variance report with page-level evidence
  • Detects easement mismatches, missing exceptions, legal description deltas
  • Drafts attorney-grade objection letter scaffolding
  • Reduces hours of manual cross-reference to a single review pass

Built for

Real estate attorneys · Title companies · Lender counsel

Walkthrough
TitleCheck AI

Page-level

evidence + citations

ALTA

commitment + survey aware

Letter draft

attorney-grade scaffolding

Audience · Who we built it for

People who pay for first-pass approval.

Permit expediters

Your business depends on first-pass approvals. PermitCheck AI catches the issues that send filings back.

PermitCheck AI

Architecture firms

Run a pre-check before handing plans to an expediter, or before signing off internally.

PermitCheck AI

Real estate attorneys

Find variances between commitments and surveys without manually cross-referencing both documents.

TitleCheck AI

Title & lender counsel

Surface mismatches with page-level evidence so your team can clear exceptions faster.

TitleCheck AI
Trust · Why you can act on a finding

Not a chatbot guess. A reviewed punchlist you can file behind.

Both products combine deterministic code rules with a model that reads context — then a human checks the findings before they reach you. No paraphrase, no invented sections, nothing you have to double-check.

01

Hybrid rules + LLM

Deterministic code rules catch the black-and-white violations; the model reads the context, drawings, and exceptions the rules can't.

02

Verbatim citations

Every finding links to the exact code section or document line — no paraphrase, no invented rule, no hallucinated section number.

03

Reviewed by us

A human checks the findings before they reach you. You get a punchlist you can act on, not a draft you have to double-check.

Questions

Honest answers about both products.

Have a different question? Book a call and we'll walk you through it live.

Book a call
Field notes · The story

What we built. Why it works.

Scrute Labs is a small team that shipped products for two distinct customer segments in under a year. The story matters because it shaped what we built: hundreds of conversations with the people who actually file and review these documents, baked into every check the tools run.

Where we are today: Both products are live and gaining users. Compliance sales cycles are long and we're committed for the duration — if you file permits in NYC or review title work, we'd genuinely like to talk.

  1. 1

    Early 2025

    Started with title work

    Scrute began as a Title Commitment + ALTA Survey Variance Analyzer — later renamed TitleCheck AI — built for real estate attorneys and title companies. The document-pairing problem we could see most clearly.

  2. 2

    Mid 2025

    Pivoted to permits

    After several months of customer development, we redirected to permit expediters and built PermitCheck AI for NYC DOB submissions — same underlying problem (compliance is a search problem), different document set.

  3. 3

    Today

    Live and onboarding

    Both tools are live and in use. Compliance has a long sales cycle and we're in it for the duration — onboarding new customers directly and shaping the roadmap around the teams who file and review every week.

Contact · Send us a document

Send us a real document. We'll run it through.

A recently rejected plan set. A pending title commitment + survey pair. We'll run it through the relevant tool and send back what we find.

No cost · No commitment
PermitCheck AI for NYC plan sets
TitleCheck AI for ALTA commitments + surveys
Verbatim citations on every finding
Findings reviewed by us before they reach you

Get in touch

Tell us what you file or review and we'll point you at the right tool.