opentakeoff
Open-source PDF construction takeoff engine driven natively by AI agents via MCP.
Measure construction plans in your browser or hand the engine to your AI agent.
OpenTakeoff is an open-source PDF takeoff and estimation engine built for preconstruction. It provides both a browser canvas for human estimators and 58 native Model Context Protocol (MCP) tools for AI agents.
What it is
It parses construction PDF plan sets, lets you set drawing scales, trace rooms on wall faces, execute symbol sweeps, and calculate material quantities. The engine exports marked-up PDFs, reports, DXF files, and structured cost estimates.
Who it's for
It suits construction estimators, preconstruction teams, and developers building AI workflows for blueprint analysis. It does not suit casual users looking for general document readers or non-technical estimators who want a fully managed cloud SaaS.
Key features
Dual Human-Agent Access
Use the browser canvas yourself or let an AI agent drive the exact same math and scale gate over MCP.
Proposal Safety Contract
Agent work lands as dashed proposals and cannot become approved ink until a human reviews and accepts it.
Strict Scale Enforcement
The engine reads scale notes but never applies them automatically, refusing to measure unscaled sheets.
Marked-Up PDF Export
Burns measurement layers and a legend cover with totals directly into the original plan drawings.
Up and running in 3 steps
Our honest verdict
What's great
- Runs entirely in your browser with zero data uploads or sign-ups required
- Offers 58 dedicated MCP tools covering every stage from plan opening to pricing and RFI creation
- Maintains full provenance by recording scale, drawing method, and human versus agent authorship
- Includes built-in pricing rules, Excel estimates, and letter-size bid proposal PDF generation
Where it falls short
- Requires Node 20 or newer and technical setup to connect and operate AI agent MCP clients
- AGPL-3.0 copyleft license imposes strict sharing obligations if you modify and network the software
- One-click automated area calculation is temporarily gated while the flood engine undergoes re-validation
Pricing & license
The software is free and open source under its license. The only cost is running it yourself locally or hosting your own static copy.
Open remote AI-training roles
870 open remote AI-training roles, updated November 14, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Does my plan data get uploaded to a cloud server?
No. The browser canvas and engine run entirely locally in your browser, and nothing phones home.
How do AI agents interact with the takeoffs?
Agents connect via the Model Context Protocol using standard I/O, utilizing 58 specialized tools to measure, check, and propose shapes.
Can an AI agent finalize a bid without human approval?
No. All agent work stays a dashed proposal until a human reviewer manually applies the approved seal on the canvas.
