Read documents as data

Captured documents are not just retrievable, they are readable as structured data. What Fryri extracts, your code can consume.

List and read#

curl "https://api.fryri.com/v1/documents" \
  -H "Authorization: Bearer fryri_sk_..."

The list includes each document's type, date and your source_ref if you sent one. Read one document in full:

curl "https://api.fryri.com/v1/documents/{file_id}" \
  -H "Authorization: Bearer fryri_sk_..."

That returns the extracted text plus the typed record: what kind of document it is, its date, and the fields found inside (amounts, parties, reference numbers). A download route returns the original bytes when they are stored with Fryri.

Structured facts#

GET /v1/facts returns the standing facts extracted across the whole library, and POST, PATCH and DELETE let you write and correct them. Use facts when you want durable statements ("the lease ends March 2027") rather than documents.

Both endpoints accept end_user_id to read a specific end user's library; see give your app memory.