Process a deal
Add files to a deal already on your Deals page, process it, and read the results — by row id.
Loading documentation…
With a key made on your Deals page, the same processing works on a deal that already exists, addressed by its row id. Use these when your team also works the deal in Specter.
| Route | Scope | Does |
|---|---|---|
GET /v1/ingest/rows/{rowId}/files | read | The deal's files |
POST /v1/ingest/rows/{rowId}/files | rows:write | Declare a file; answers its upload URL |
POST /v1/ingest/rows/{rowId}/process | processing | Process the deal's documents |
GET /v1/ingest/rows/{rowId}/processing | read | The latest job, its steps and its documents |
GET /v1/ingest/rows/{rowId}/findings | read | The latest job's findings |
GET /v1/ingest/rows/{rowId}/financials | read | The deal's headline financials |
A deal's row id comes from reading the page, or from the answer when you add the deal with the same key.
A 201 answers the file in data and uploadUrl beside it. Upload the bytes exactly as for a
job — a PUT with the declared Content-Type, within 15 minutes
(Upload each file).
It processes every file on the deal, and answers like starting a job:
201 a new job, 200 the job these files already started, 409 would_supersede with the fields
a re-run would replace (send {"replace": true} to go ahead). The optional body takes
idempotencyKey and replace.
GET …/processing answers the job — status, events, documents with their extractions —
the same job object as Read results. A deal never processed
answers an empty job, not a 404. GET …/findings and GET …/financials answer the
findings and financials on their own.
These routes share the board routes' limit of 120 requests a minute per key — see Rate limits.