For developers

Populate.ai API & MCP

Turn any bill, receipt, invoice, scan, or photo into structured data with one HTTP call — or plug Populate.ai straight into Claude, ChatGPT, Cursor, and other AI assistants via MCP.

API keys

Sign in to create an API key

API keys are tied to your Populate.ai account. Once signed in, you can create, name, and revoke keys here.

Sign in

Quickstart

Create an API key above, then send your first request:

curl -X POST https://populate.info/api/v1/parse \
  -H "Authorization: Bearer pop_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "filename": "invoice.pdf",
    "mimeType": "application/pdf",
    "fileBase64": "<base64-encoded file>"
  }'

Endpoints

MethodPathDescription
POST/api/v1/parseExtract a table (rows/columns + per-cell confidence) from a document or audio dictation.
POST/api/v1/reconstructIntelligent OCR — return a formatted document (headings, paragraphs, lists, tables).
GET/api/v1/documentsList your recent saved uploads.
GET/api/v1/documents/:idFetch one saved document by id.
GET/api/v1/healthHealth probe — no auth required.

All authenticated calls take Authorization: Bearer pop_live_…. Files are sent as base64 in the request body — the API returns the same JSON your browser sees.

MCP server

Add Populate.ai as a Model Context Protocol server. Your AI assistant signs in with your Populate.ai account and can call the same tools on your behalf.

MCP URL
https://populate.info/mcp
Claude Desktop / Web

Settings → Connectors → Add custom server → paste the MCP URL. Sign in with the Populate.ai account you already use.

ChatGPT (Developer mode)

Settings → Connectors → Add → paste the MCP URL. Authorize with your Populate.ai account when prompted.

Cursor

Settings → MCP → Add server → HTTP → paste the MCP URL. Sign in to complete the OAuth handshake.

Any MCP client

Point the client at the MCP URL above. Populate.ai uses OAuth 2.1 with dynamic client registration — most clients discover it automatically.

Available tools

  • parse_document — extract rows/columns from a document.
  • reconstruct_document — Intelligent OCR into a formatted document.
  • list_documents — list your saved uploads.
  • get_document — fetch one saved upload.
Limits during beta: Table extraction is capped at 120 rows per document and Intelligent OCR at 10 pages per PDF. API and MCP calls both act as the signed-in Populate.ai user, so they bypass the anonymous 5-upload cap.
Populate.ai

Receipts in. Tables out. Privacy-first AI data entry for accountants and small teams.

Product
Company
Get started

Drop a receipt, snap a photo, or dictate aloud. Free up to 15 MB per file and 15 MB per upload session.

© 2026 Populate.ai. All Rights Reserved.TermsPrivacy