Get Extraction Data

Fetch just the extracted key-value data from an extraction with GET /v1/extractions/:id/data. Returns clean JSON by default or a downloadable CSV file.

The Get Extraction Data endpoint, GET /v1/extractions/:id/data, retrieves only the extracted field values from an extraction, without metadata, confidence scores, or processing details. This is the lightest-weight endpoint for consuming extraction output, and it is ideal for downstream integrations that only need the structured data as JSON or CSV.

# JSON (default)
curl -s https://api.talonic.com/v1/extractions/a1b2c3d4-e5f6-7890-abcd-ef1234567890/data \
  -H "Authorization: Bearer tlnc_your_api_key"

# CSV download
curl -s "https://api.talonic.com/v1/extractions/a1b2c3d4-e5f6-7890-abcd-ef1234567890/data?format=csv" \
  -H "Authorization: Bearer tlnc_your_api_key" -o extraction.csv

Response

{
  "vendor_name": "Acme Corp",
  "invoice_number": "INV-2026-0847",
  "total_amount": 14250.00,
  "due_date": "2026-03-15"
}

vendor_name,invoice_number,total_amount,due_date
"Acme Corp","INV-2026-0847",14250,"2026-03-15"

Most integrations call this endpoint to feed extraction output into downstream systems (CRMs, ERPs, data warehouses) that only need the raw key-value data. A typical workflow is to extract a document, then call this endpoint with the extraction_id from the response to get a clean data payload without metadata overhead.

The response is a flat JSON object where each key is a field name and each value is the extracted value, typed according to the schema (strings, numbers, dates, arrays). Use ?format=csv to download the same data as a CSV file — one header row of field names plus one data row, since an extraction covers a single document. String values are double-quoted with embedded quotes escaped by doubling; other values are stringified unquoted. To build a multi-document CSV table, export a data product or pipeline results instead of concatenating per-extraction files.

Pair this with [GET /v1/extractions/:id](get-extraction) when you also need confidence scores, locked field status, the normalized numeric magnitudes, or processing metadata. Note that the response shape matches the schema used during extraction — if no schema was provided, auto-discovered field names are used as keys, so downstream consumers should tolerate new keys appearing as documents vary.

Errors