All resources
The basics·7 min read

Intelligent data automation, in practice.

Pulling fields out of a document is only half the job. The other half is getting that data into the systems that act on it — your ERP, your CRM, your spreadsheet, your custom internal tool. This guide walks through the three integration surfaces DocParse ships and the workflows teams build with them.

Three integration surfaces

DocParse exposes three ways to wire extracted data into the rest of your stack. They overlap on purpose — pick whichever matches your team's preferred level of code.

  • REST API — for backends and scripts. Plain HTTP, bearer-token auth, JSON in and out.
  • Signed webhooks — for event-driven systems. POSTed to your endpoint the moment an extraction finishes.
  • Native Zapier app — for non-developers. Trigger downstream actions in 6,000+ apps without writing code.

REST API

The REST API is what powers the dashboard, the Zapier app, and your own code — there is only one API. Three endpoints cover most workflows: createExtraction (define a schema), uploadFiles (push documents to it), and getBatchResults (read structured output).

Authentication is bearer-token: pass the dp_... key in an Authorization header. Keys are SHA-256-hashed at rest and only their prefix is visible after creation, so rotate them anytime — old keys are revoked instantly.

Signed webhooks

Polling for results works, but for high-volume pipelines you want push semantics. Register a webhook endpoint in the dashboard, paste your URL, and DocParse will POST a signed delivery the moment any extraction in that account finishes.

Every delivery follows the Standard Webhooks specification: webhook-id and webhook-timestamp headers identify the event, and webhook-signature: v1,<sig> is an HMAC-SHA256 over `{id}.{timestamp}.{body}` using your per-endpoint secret. Verify it on receipt to confirm the delivery came from DocParse and hasn't been replayed (we reject timestamps older than 5 minutes on our side too).

Failed deliveries are retried with exponential backoff. Every attempt is logged in the dashboard with response code, body, and timing.

Native Zapier app

For workflows that don't want any code, the Zapier app is the fastest path. It ships with two triggers, two actions, and a search:

  • Trigger: New File Processed — fires when an extraction completes, with extracted fields flattened so you can map them directly to downstream apps
  • Action: Upload File — push a file from any Zapier-supported source (Gmail, Drive, Dropbox, email attachment) into a DocParse extraction
  • Action: Create Extraction — provision a new extraction template on the fly
  • Search: Find Extraction by Name — chain Find &rarr; Upload when the extraction already exists

Common patterns

A few workflows we see often:

  • Gmail label → Upload to invoice extraction → New File Processed → Update Sheets row
  • Drive folder → Upload to resume extraction → Append to Airtable / HubSpot
  • Slack file share → Upload to receipt extraction → Post extracted data back to the channel
  • Backend → REST createExtraction + uploadFiles → Signed webhook → Postgres upsert

Reaching tools without a native integration

DocParse doesn't ship native connectors for Make.com, n8n, or specific ERPs like NetSuite or SAP — but both Make.com and n8n accept generic HTTP webhook nodes, so you can point them at the DocParse payload. The same applies to any internal tool that can receive an HTTP POST.

Automate one workflow this week.

Start with 100 free pages — wire DocParse into the rest of your stack in an afternoon.