Production file-ingestion build

AI Building Surveyor Assistant

Upload PDFs, drawings/images, text files, or DXF/IFC-style CAD files. The app extracts text/OCR, classifies documents, flags missing information, and drafts RFIs for human review.

Real file upload

PDF + OCR + images + DXF/IFC text

Scanned PDFs/images use OCR when Tesseract is available. Binary DWG/RVT are surfaced as unsupported until a converter is added.

Drawing tag takeoff

C1/C2/C3 + evidence markup

First pass uses PDF text boxes + Tesseract OCR bounding boxes, coordinate deduplication, and annotated evidence images. It is detector-ready for YOLO/RT-DETR once labelled data is collected.

Manual text fallback