Steelflo
AI-powered steel cost estimating. Upload blueprints, detect members, export BOMs.
Project Overview
Structural steel takeoffs are still done by humans squinting at PDFs, manually measuring and counting every beam, column, and connection. Steelflo replaces that process with AI.
The system uses a hybrid pipeline — PyMuPDF handles structured extraction from engineering drawings, while Gemini Vision handles the visual interpretation that humans can barely read. Upload a blueprint, get detected steel members, verify the count, and export a full bill of materials or priced estimate.
Currently converting from an internal tool to a full SaaS product. Three user journeys complete: manual estimating, BOM-only extraction, and end-to-end upload-to-estimate.
Hybrid AI pipeline
PyMuPDF for structured data, Gemini Vision for drawings humans can barely read.
Three user journeys
Manual estimating, BOM-only extraction, and full end-to-end upload-to-estimate.
BOM → Estimate flow
Detected members feed directly into a pricing engine with real steel costs.