Headwind

Start

How it works

One review a week, one question per unclear role, and a forecast that changes only when evidence supports it.

The weekly flow

  1. Import. CSV files are checked row by row. Nothing is saved until every row is valid and you press Import.
  2. Match companies. Each client domain is looked up on Crustdata. It counts as matched only if exactly one company has that exact domain with high confidence. Otherwise it's blocked, not guessed.
  3. Check listings. For matched companies, Headwind searches for the role's public listing in a 30-day window.
  4. Assess. Fixed rules settle what the evidence supports. For anything unclear, a model writes a short note and a question, and a checker rejects any sentence that claims an outcome the evidence doesn't show.
  5. Forecast. Original, Reviewed and Downside are recalculated.
  6. Sync to cfo.ai. Reviewed and Downside are written to their own cfo.ai scenarios and read back. Only a matching readback shows as verified.

When it runs

A timer queues a weekly review every Monday at 07:00 UTC. You can also start one from the Overview. A pasted client reply queues a shorter reassessment that runs only the assess, forecast and sync steps.

The three scenarios

Original is your imported numbers. Reviewed keeps what the evidence supports and removes fees the client or your records have ruled out for this period. Downside also leaves out every role still waiting for confirmation.