Deal intake parser

Deal flow arrives as mess: forwarded emails, spreadsheet exports, notes from a demo day. The intake parser takes that raw material — pasted text or uploaded files — and returns structured company candidates, each scored against your stated thesis, in one batch.

How it works

1. 1Paste text directly, or upload files. File types are validated before anything is processed.
2. 2The AI extracts one candidate record per company it finds: name, sector, stage, geography, and whatever traction detail the source contained.
3. 3If a document is image-only (a scanned one-pager, a screenshot), extraction falls back to a vision model rather than failing.
4. 4Each candidate gets a thesis-fit score so you triage the batch by relevance, not source order.
5. 5From the results panel, add candidates to your watchlist or send an invite email directly.

Key rules & limits

• Paste-your-own-data only, by design. There is no email or CRM OAuth integration — Hockystick never connects to your inbox. What you paste is the entire input.
• Every run is saved to history and can be reopened later with its full results.
• Extraction quality follows source quality: a candidate parsed from two vague sentences will be sparse, and the parser leaves fields empty rather than inventing them.

What the AI does / doesn't