Skip to content
CapexSignal.

Industrial project intelligence

Build a project pipeline your team can verify.

Search and monitor industrial-project evidence from public sources. Every held fact carries the document and exact quote behind it, and every answer says when source depth is incomplete. Each country opens independently only after its live source, canary and evaluation gates pass.

No card or sales call to open a free GB sample. Paid checkout opens only for launch-ready geographies. Pricing is on the site, which is more than any incumbent will tell you.

GET /v1/launchindependent by geography
GB · NESO TEC
adapter ready; gate pending
US · EIA-860M
adapter ready; gate pending
BR · ANEEL SIGA
adapter ready; ODbL gate
IN · CEA RE
adapter ready; permission + identity gates
DE · AE · SA
public-source depth pending

Adapter-ready is not paid-ready. A country also needs legally usable customer data, seven daily production canaries and a recent independent scorecard with useful cases.

7
countries tracked through independent launch gates
4
implemented public project-register adapters
0
countries made sellable by configuration alone
0
evidenced facts stored without a quote

What teams use it for

Why our data is different

Three differences that come from how the data is stored, which is why a competitor cannot add them without rebuilding.

You get the sentence, not just the field

Ask for a capital cost and you get the figure, the document version and the words that state it. When two sources disagree you see both and how it was settled — never an average.

Nothing is ever overwritten

A date that moves does not erase the old one. Ask what we said in March and get an answer — which is what makes the record usable in diligence, and a developer's said-versus-did measurable.

We publish our accuracy

Scored per geography when an adjudicated corpus has actually been published, and never averaged into one flattering figure. An absent scorecard renders as absent, not zero.See the numbers.

Fits the systems you already run

A versioned REST API, bounded SQL, and Parquet export straight to your warehouse — plus an MCP endpoint, so your own AI agents can query the evidence directly rather than scraping a portal.

Alerts fire on new matches only, into a webhook or your CRM. Nobody wants to run the same search every Monday morning.

Every response names its own scope, what it cost and what was left out, so an incomplete answer never looks like a complete one. See the full endpoint list, or the terms this site defines, in the glossary.

See the API →

GET /v1/claims?subject=proj:GB:example:1&predicate=capital_cost

{
  "facts": [{
    "value": "£560 million",
    "confidence": 0.69,
    "evidence": [{
      "quote": "the revised capital cost of £560 million",
      "document": "9a3f9f41…",
      "source": "gb_public_register",
      "asserted_at": "2026-08-01"
    }]
  }],
  "bounding": { "complete": false, "scope": ["GB"],
                "note": "one source is acquired but not yet mapped" }
}
Illustrative response shape, clearly marked: not a live customer record or accuracy claim.

Try it on a project you already know

The fastest way to judge this is to look up something you can check. The free account is limited to the current GB corpus, needs no card, and shows both evidence and coverage gaps.