Integrations
Documents reach Tavnit however suits the team sending them — dropped in the app, forwarded to an address, posted to an endpoint, or requested by an AI assistant. Extracted data leaves the same way.
Each route runs the same flows, so the schema, cleaning rules and review steps you configured apply no matter how a document arrives.
MCP Connector
Add Tavnit to claude.ai, Cursor or any MCP client so your assistant can run documents through your flows and query your extracted data directly.
REST API
Send documents programmatically with an API key. Multipart or base64 upload, with Python and JavaScript examples.
Forward or auto-forward documents to a dedicated address and have every attachment extracted without anyone opening the app.
Webhooks
Push typed results to your own endpoint the moment a run finishes, so extracted data lands in your systems automatically.
Zapier, Make, n8n and Power Automate
No-code recipes that call the same endpoints, for teams that would rather wire this up visually than write code.
Buckets
Keep extracted rows in built-in structured storage, query them, chart them, or export to CSV.
Which one should you use?
If people are sending the documents — an accounts inbox receiving supplier invoices, say — use email. Nobody has to learn a new tool; they forward as they already do.
If a system is sending them, use the REST API and have results pushed back by webhook. If you would rather not write the glue code, the same endpoints work from Zapier, Make, n8n and Power Automate.
If you are working alongside an AI assistant, use the MCP connector. It lets claude.ai or Cursor run documents through your flows and query what you have already extracted, without you moving files around by hand.
Start with one document
Build a flow, send something through it, and wire up delivery once you can see the output. Free credits to begin.
Start free