OpenTakeoff
Specialized verticals · tested 2026-08-09 · re-test due 2026-11-09 · by the Hlido desk, not the vendor
In short: Browser-local construction takeoff that records how every measurement was made — by a person, by one click, or by an agent — which is the right provenance model for a document that ends up in a bid.
4 PASS · 3 FAIL of 7 public-surface claims
Quick answer
OpenTakeoff scores 70/100 (STEADY) on Hlido’s independent, hands-on test (reviewed 2026-08-09). STEADY (70) for solving a real trade problem with a genuine constraint respected (plan sets never leave the browser), Apache-2.0 licensing, and an unusually sound provenance model that records whether each measurement ca Pricing: Open source (free entry point documented).
OpenTakeoff is an Apache-2.0 quantity-takeoff tool that runs entirely in the browser: drag in a PDF, an image or a whole zip plan set, set or auto-detect the scale, and trace areas, cut-outs and markups against named conditions. Two decisions stand out on the captured surface. First, 'nothing leaves your browser' — for contractors, plan sets are commercially sensitive and often under NDA, so a local-only tool removes a real adoption blocker rather than a theoretical one. Second, and more interesting for an agent register: the app states that every measurement keeps its scale and how it was made — a person, one click, or an agent — over one shared engine. That is provenance at the measurement level, and it is exactly what a takeoff needs, because a quantity that feeds a bid has to be auditable regardless of who or what produced it. Most tools bolting on AI assistance treat human and machine output as interchangeable; recording the difference is the more honest design. The surface is the running application rather than a marketing page, so what Hlido could verify is the UI itself: conditions, waste percentages, fill styles, scale setting, voice command, zone and snap aids, and a sample medical-centre floor-finish plan for first use. What is not visible anywhere is the agent interface it advertises — no MCP endpoint, API documentation or integration guide was reachable — along with no accuracy evidence, no version and no adoption. The agent claim is therefore stated but unverified.
Why STEADY
STEADY (70) for solving a real trade problem with a genuine constraint respected (plan sets never leave the browser), Apache-2.0 licensing, and an unusually sound provenance model that records whether each measurement came from a person, a click or an agent. Held below VITAL because the agent-facing path it advertises is not documented anywhere reachable — 'point an AI agent at the same engine' is a claim with no visible interface behind it — and because there is no accuracy evidence, version or adoption on the surface.
Public-surface checklist
- PASS Homepage loads (required)
- PASS Primary value prop (required) — 'free, AI-native takeoff for construction' — measure a plan by hand or point an agent at the same engine
- PASS Cta present (required) — 'Open your plans' drop zone plus 'Load sample plan'
- PASS Pricing or access — Apache-2.0 open source, free, no account
- FAIL Docs present (required) — In-app hints only; no reachable documentation site
- FAIL Agent interface documented — Agent capability advertised but no MCP/API interface reachable
- FAIL Third party validation — No accuracy evidence, version or adoption
What we saw
1 screenshot captured by the Hlido engine during the reviewed run (run-5ab1555bb2361361-opentakeoff-netlify-app). Our own captures — not vendor marketing material.
What it does well
- Runs entirely in the browser — plan sets never leave the machine, which is a genuine constraint in construction bidding
- Records how each measurement was made (person, one click, or agent) alongside its scale — measurement-level provenance
- One engine shared by humans and agents rather than a separate AI mode
- Apache-2.0 open source
- Practical takeoff features: conditions with waste percentages, area and cut-out tools, fill styles, markup, zone and snap aids, scale auto-detect
- Multi-sheet plan sets with user-controlled ordering; accepts PDF, image or a zip set
- Sample medical-centre floor-finish plan lets a new user complete a takeoff without supplying their own drawings
What it fails at
- The advertised agent path has no documented interface — no MCP endpoint, API reference or integration guide was reachable
- No accuracy or validation evidence for measurements
- No version, changelog or adoption evidence
- Browser-local by design means no collaboration, server-side storage or team workflow
- Not tested hands-on by Hlido against a known-quantity plan
Red flags
- 'Point an AI agent at the same engine' is presented as a core capability, but no MCP endpoint, API documentation or integration guide is reachable from the captured surface.
Best for
- Estimators and small contractors who need takeoff without sending client plan sets to a vendor
- Anyone wanting an open-source alternative to per-seat commercial takeoff software
- Teams that need to know whether a quantity was measured by a person or produced automatically
Not recommended for
- Agent-driven workflows today — the integration path is advertised but not documented
- Multi-user collaboration or centralised project storage
- Work requiring validated measurement accuracy or a vendor support commitment
Pricing & access
- ModelOpen source
- Free entry pointYes — a free tier or open-source edition is documented
- Pricing findable on the public surfacePASS Apache-2.0 open source, free, no account (tested 2026-08-09)
Derived from Hlido-held evidence only (engine checklist + editorial text); quotes are verbatim from the scorecard; not vendor-supplied; re-derived daily. Verify current prices on the vendor's pricing page. Last verified 2026-08-09.
Related agents
Agent relevance
No programmatic surfaces
Agentic-Commerce Readiness 32/100 · SURFACE-ONLY
Independent readiness for agent delegation & transaction. How it’s scored · check live
The application advertises that an AI agent can be pointed at the same measurement engine as a human, and it records agent-produced measurements distinctly — but no MCP endpoint, API reference or integration guide is reachable from the public surface, so the path cannot be confirmed or used today.
Agent-friendly score: 3/10
Score over time
The longitudinal record — every point is the score as published on that date. Raw series.
Evidence
- Browser-local takeoff — 'Nothing leaves your browser'; accepts PDF, image or zip plan sets — source (2026-08-09) verified
- Every measurement keeps its scale and how it was made — a person, one click, or an agent — source (2026-08-09) verified
- Apache-2.0 open source, by Kentucky AI — source (2026-08-09) verified
- Conditions with waste percentages, area and cut-out tools, markup, zone and snap aids, scale set or auto-detect — source (2026-08-09) verified
- Sample medical-centre floor-finish plan provided for first use — source (2026-08-09) verified
- A documented agent interface (MCP endpoint, API reference or integration guide) — source (2026-08-09)
- Measurement accuracy evidence, version or adoption — source (2026-08-09)
