DNAX turns architectural drawings, specifications and project data into a spatial graph that AI can query and reason about.
Explore the demo View technologyDrawings hold far more than lines. DNAX extracts the structure and keeps it queryable.
Walls, rooms, doors, windows, dimensions, structural elements.
Furniture, fixtures, zones and circulation paths.
Adjacency, access, daylight exposure and area per space.
Briefs, specifications and standards linked to the spaces they govern.
Computer vision understands the geometry. Claude reasons about the project. Deterministic checks validate the result.
PDF, DWG, IFC, images, briefs, specifications.
CV model: wall mask, openings, room segmentation, object classes.
Shapely and NetworkX build rooms, edges and metrics.
Project documents and standards via pgvector.
Claude API receives graph, brief and context.
Schema and rule checks before anything reaches the user.
DNAX is designed around the Claude API as its reasoning layer across spatial data, specifications and requirements, returning structured output that is validated before use.
{
"project_summary": {},
"spaces": [{"id":"living","area_m2":18.4}],
"issues": [{"severity":"critical",
"space":"bedroom"}],
"recommendations": [],
"brief_compliance": {},
"confidence": {}
}Sample project with curated output. It illustrates the workflow and is not a live model run.
DNAX is in private beta. We are onboarding a small number of design teams.
Request early access