# Show HN: RevisePulse – AI for CAD and Revit Coordination > Source: > Published: 2026-10-15 05:24:46+00:00 We replaced brute-force bounding-box intersection with semantic AI reasoning. The result: fewer tickets, real issues, faster coordination. Llama-3.1 analyzes element properties and spatial context — not just bounding boxes — to flag real coordination errors. Compare architectural and structural IFC models side-by-side. AI detects what changed and what actually conflicts. Built on ifcopenshell. We parse every wall, column, slab, and beam with its full property set and global coordinates. Upload your models and get a prioritized coordination report before your coffee cools. No overnight batch processing. Traditional clash detection buries you in thousands of false positives. AI filters the noise and surfaces only real issues. Every element is stored as a 3D geometry point. Proximity queries run at database speed with GiST spatial indexing. Drop your architectural and structural .ifc files. ifcopenshell extracts every element with its coordinates and properties. PostGIS finds proximity candidates. Llama-3.1 evaluates each pair semantically — is this a real conflict or an expected junction? Get a prioritized list of real issues with severity, description, and suggested fix — formatted as JSON for easy integration. Simple, transparent pricing. Cancel anytime. Your card is never charged during the onboarding period. For independent draftsmen and small studios. For growing firms with multi-discipline models. For large contractors with custom requirements. Stay updated as we roll out broader discipline file processing capabilities.