Try the full caging flow.

Upload a check image and watch the complete pipeline: OCR extraction → warehouse lookup (occupation, employer, past contributions) → filing-ready donor record. All FEC-required fields auto-filled. No manual typing.

Click a sample check to run it

Synthetic data for demo. Click any check below to send it through the full flow — OCR, warehouse lookup, filing-ready output.

Or upload your own

What this demo shows

  1. Real OCR via Anthropic Vision (Haiku 4.5) — extracts payer name, address, amount, date, memo, check number from any check image.
  2. Warehouse lookup — finds the donor in our 261M FEC + 48M state record warehouse and returns their occupation, employer, and past contribution history. (Demo currently uses mock warehouse data; real implementation hits campaign-warehouse-api.)
  3. Filing-ready record— combined data structured for direct write to your committee's ledger. No manual entry of occupation or employer.

What happens to your image?

  • • Held in memory for the OCR call only — never written to disk.
  • • MICR strip (the bottom band with routing/account info) is masked before OCR sees it.
  • • Only payer name, address, amount, date, memo, and check number are returned.
  • • Account number, routing number, and bank name are stripped at the API boundary.