Start a document processing session

POSThttps://studioapi.kriyam.ai/models/docwise/v2

Content type multipart/form-data

Uploads a set of documents and starts reading them. It returns once the documents are stored and their pages rendered, before any of them have been read — results arrive at your webhook. Affordability is checked against the whole pipeline first, so a session you can't pay for never starts.

Request

curl -X POST "https://studioapi.kriyam.ai/models/docwise/v2" \
  -H "X-API-Key: $KRIYAM_API_KEY" \
  -F "files=@invoice.pdf" \
  -F "capability=extract" \
  -F "processing_mode=fast" \
  --form-string 'extraction_schema={"type": "object", "properties": {"invoice_number": {"type": "string", "description": "The invoice number printed by the seller"}, "invoice_date": {"type": "string", "description": "Date of issue, as YYYY-MM-DD"}, "seller_gstin": {"type": "string", "description": "GSTIN of the seller, not the buyer"}, "total_amount": {"type": "number", "description": "Grand total payable including tax"}, "line_items": {"type": "array", "description": "One entry per billed line", "items": {"type": "object", "properties": {"description": {"type": "string", "description": "What was billed"}, "amount": {"type": "number", "description": "Line total before tax"}}}}}}'

Response

{2 items"session_id":"e3a91b47-0c25-4d6f-9b8a-71f4c2..."
"documents":[1 item0:{4 items"id":"1d7c4f92-8b3a-4e50-a6c1-9f28b5..."
"original_filename":"invoice.pdf"
"status":"uploaded"
"sequence":1
}
]
}

Authentication

Send your key in the X-API-Key header. See Authentication.

Request

Body

Name	Type	Required	Description
`files`	file[]	Required	The documents — repeat the field once per file, up to 5. PDFs and images.
`capability`	string	Required	A step to run, repeated once per step: `parse` reads the text and layout, `classify` identifies what each document is, `extract` pulls structured fields out of it.
`processing_mode`	string	Optional	`fast` or `accurate`. Accurate costs more per page and is worth it on poor scans. Default `fast` fastaccurate
`extraction_schema`	object	Optional	The fields `extract` should fill, as a JSON Schema object sent as JSON text. Every field needs a `description` saying what goes in it. Requires `extract` among the capabilities. Omit it to have Docwise generate a schema from the documents. At most 200 fields in all, nested 5 levels deep, 64 KB.

Response

200 OK

Field	Type	Description
`session_id`	uuid	The session. Everything else is read or reported by it.
`documents[]`	object[]	The documents as stored, in the order you sent them. None of them have been read yet.
`documents[].id`	uuid	One document within the session.
`documents[].original_filename`	string	The filename you uploaded.
`documents[].sequence`	integer	Its 1-based position in the session, which is how documents are ordered — files uploaded together share a timestamp, so that can't order them.
`documents[].status`	string	Where this document has got to on its own.

Errors

Status	Code	When
400	—	No file was sent, more than 5 were, no step was requested, or a requested step is unknown or not currently available. Also when `extraction_schema` is sent without `extract`, isn't a JSON Schema object, or has a field with no `description` — the message names the field.

Every keyed call can also get401402403404429— see Errors.