You take a photo of a construction site. An estimator writes a text description. An architect exports a Revit model. An AI agent takes any of these inputs, connects to a real construction database, and tells you how much it costs and how long it will take.
That’s not a concept — it works today. Free and open source.
Three Ways In
Send a Photo
AI identifies walls, floors, finishes — matches with 55,719 real work items — returns cost + labor hours.
Describe in Text
"Renovate 50m² bathroom with tiling and waterproofing" — full estimate with material, labor, equipment breakdown.
Upload CAD/BIM
Revit, IFC, DWG — automated 4D/5D estimate with quantities extracted directly from the model.
New in v0.1.1
- NEW UK and USA databases added
- NEW Now 11 regions with local currencies and standards
- NEW Examples in Python, JavaScript, Rust, and R
- NEW All free, all open source
I already know of several large companies — one of which is a major vendor — that are already testing these databases in their projects.
How It Works
AI decomposes what it sees into construction work items, searches a database of real norms (how many hours, how much material, what equipment), and calculates the price for your region.
Behind every estimate is a real database — not guesswork. Not abstract "per m² estimates" — real resource-based norms refined over 100 years.
What’s Inside Each Work Item
Labor
How many workers, how many hours per unit of work
Materials
What materials, exact quantities per unit
Equipment
What machinery, for how long
Regional Pricing
Local market costs for your specific region
The data schema has ~85 fields — codes, units of measurement, work composition, cost indicators. Load it directly into Python, Power BI, n8n, Claude Code, or any pipeline you use.
Everything is free. Ready pipelines and workflows included. 30+ code examples.
Download on GitHub
Star the repo to get notified about updates, new regions, and free tools.
