FLACRON INTEL

Evidence you can check yourself

Flacron Intel is built to be auditable. Here is exactly how we handle sources, scores and Super Intelligence.

Five kinds of statements, always labelled

  • Verified fact

    Copied or derived deterministically from an authoritative source record, with its link and retrieval time.

  • Reported claim

    Something a source states that we have not independently verified.

  • Super Intelligence interpretation

    A model-written summary or explanation. It must cite retrieved evidence or say it is uncertain.

  • Prediction

    A forecast. Always shown with assumptions and uncertainty — never as fact.

  • Evidence unavailable

    We could not retrieve or verify the information. We say so instead of guessing.

Provenance for every record

When we retrieve data from a provider we store the original payload as an immutable source record with its canonical URL, publisher, licence, publication date, retrieval time and a SHA-256 content fingerprint. If the source changes, we keep the old version, mark it superseded and link the new revision — so you can always see what changed and when.

Freshness and deadlines

Each insight shows when it was last verified. Records past their refresh window are labelled aging or stale. If a provider is unavailable we show the cached copy clearly marked as such — we never fill gaps with invented results. Federal grant deadlines are U.S. Eastern calendar dates and expired deadlines are flagged.

Separate, explainable scores

  • Relevance — how well an opportunity matches your profile keywords, categories and agencies.
  • Evidence quality — how complete the verified source information is.
  • Urgency — derived only from the time left before the deadline.
  • Eligibility fit — rule-based checks against the listing; shown as “unknown” when information is missing.

None of these is a probability of winning an award or making a profit, and we never present them that way.

Super Intelligence with guardrails

Retrieved documents and web pages are treated as untrusted content. They can inform an answer, but they can never trigger actions. Super Intelligence output must cite the evidence it relies on, and every external action — sending an email, contacting a partner, submitting anything — requires your explicit approval with the exact content shown first. We do not submit grant applications on your behalf.

Corrections

If you believe something is wrong, use “Report an issue” on the item or contact us. Corrections propagate to derived insights and everyone who saved or acted on the item is notified.

See the evidence for yourself

Create a free workspace and open any opportunity's evidence view: source, retrieval time, fingerprint and every eligibility check.