The Conformance Tool is an AI-powered drawing review platform that automates the comparison of architectural and engineering submittal drawings against an organization's prototype or standard reference set. Built for firms that enforce consistent design standards across a portfolio of built projects — including hotel brands, retail chains, restaurant franchises, healthcare networks, and multi-site corporate tenants — the tool transforms what is traditionally a slow, manual, and inconsistent review process into a structured, evidence-backed audit trail.
At its core, the Conformance Tool ingests two PDF drawing sets: a prototype (the company's reference standard encoding brand, design, and technical requirements) and a submittal (the drawings produced by an architect, contractor, or franchisee for a specific project). The system then executes a multi-pass analysis pipeline. First, it pairs sheets between the two documents by drawing number, classifying each pair as matched, modified, unchanged, added, or removed. Within each matched pair, the tool extracts discrete drawing subjects — floor plans, elevations, sections, and details — and matches them across the prototype and submittal, flagging any subject that is missing from the submittal or appears as an unexpected addition. A secondary pair-swap audit re-validates same-category subject clusters to catch crossed matches between, for example, two adjacent elevations.
Once subjects are aligned, the tool runs structured change detection to surface MEP and architectural deviations within each matched subject. Every finding is captured with category, severity, confidence score, element type, description, and pixel-accurate bounding boxes on both drawings — so reviewers can see exactly where on each sheet the deviation occurs. An ensemble approach runs the detection across three independent AI drafts; findings flagged by all three are surfaced as high-confidence core issues, while single-draft findings are routed to a long-tail review queue, giving reviewers built-in triage prioritization.
Each finding is then linked on demand to the specific prototype criteria note that establishes the requirement being violated, providing instant justification without manual cross-referencing. The platform also extracts callout annotations (numbered markers and their referenced notes) from prototype drawings, persisting them with spatial coordinates for overlay rendering.
Reviewers navigate findings through multiple lenses: by sheet pair, by subject, by room (using extracted room metadata), or by discipline. The tool delivers a comprehensive structured report summarizing matched, modified, added, and removed sheets, broken down by discipline, with full pair-level detail. Real-time progress streaming, soft-delete with worker cancellation, and tenant-scoped access controls round out the production-grade infrastructure. The result: senior reviewers shift from hunting for deviations across hundreds of sheets to efficiently accepting or rejecting a pre-located, pre-justified list of findings — dramatically reducing review cycles while improving consistency across projects.
