PDF data extraction

Turn PDFs into tables you define.

Name the fields you need once. PdfParse reads every PDF you send it, fills one row per document, keeps each row linked to its source, and exports CSV, JSON or a SQLite database.

20 pages free · No credit card required

Animated walkthrough of the PdfParse invoices table: below the first two invoice rows, four more extracted invoice rows appear one at a time, then the record count and a highlight around the finished table.

How PdfParse works

  1. 01

    Describe the fields

    Name each field and say in plain words what to read. Repeating details, like line items, get their own linked table.

  2. 02

    Send your PDFs

    Upload them, forward them to an inbox, or call the API. Every document becomes a row.

  3. 03

    Check, then export

    Open any row beside the PDF it came from, then export CSV or JSON, or query the SQLite file.

Email inboxes

Forward invoices. Get rows.

Give each workflow its own address. PDF attachments sent there are extracted into your tables, and every message shows what happened to it, from received to dispatched.

  • Senders are denied by default. You choose who can send.
  • Repeat deliveries of the same file are skipped.
  • Anything uncertain waits in review instead of guessing.
How email inboxes work ↗
Animated walkthrough of the Accounts payable inbox in PdfParse: three new emails arrive on top of earlier messages, each with its status: No PDF attachments, Dispatched, and Processing for the latest.

Splitting and routing

One packet in. Every document in its place.

Vendors staple invoices, purchase orders and delivery notes into a single PDF. Describe the split in plain words, and each section goes to the table that fits it, at a confidence you set.

  • Splitters find where each document starts and ends.
  • Routing rules are written in plain language.
  • A section with no confident match is never guessed.
How routing works ↗
Animated walkthrough of a splitter test in PdfParse: a 12-page vendor packet is split into four sections, three are matched to destination tables and a delivery note is marked Needs review.

Human review

Sure ones run. The rest wait for you.

Every routing decision has a confidence score. When it falls below your threshold, or an extraction fails, the document waits in a review queue, so nothing uncertain reaches your tables.

  • See why each document is waiting.
  • Pick the table, then route and extract in one step.
  • Failed extractions save nothing until they're retried.
Read the human review guide ↗
The PdfParse human review queue: failed extractions and routing decisions below the 85% threshold, each waiting for a person to review.

API, SDK and MCP

Run it from your code, or your AI agent.

A project key gives your code the same tables, documents, jobs and rows as the dashboard. Call the REST API with cURL or the TypeScript SDK, try your code first in the API playground, or connect an assistant through the MCP server.

import { Client } from "pdfparse-sdk";

const pdfparse = new Client(process.env.PDFPARSE_API_KEY!);

const [invoices] = await pdfparse.tables.list();

const page = await pdfparse.rows.search({
  table_id: invoices.id,
  filters: [{ column: "vendor_name", op: "eq", value: "Atlas Industrial" }],
});

Output formats

One schema. Three ways out.

JSON

Consistent keys and nested records for an API or pipeline.

PDF to JSON ↗

Pricing

Start free. Upgrade as your volume grows.

Start with 20 free pages. Every plan includes extraction, inboxes, routing, review, and exports. Upgrade when you need more pages and storage.

Free

Try a complete document workflow on your own PDFs.

$0
  • 20 pages total
  • 50 MiB storage
  • Automated email intake & extraction
  • Classification, routing & PDF splitting
  • Human review
  • Custom schemas & queryable tables
  • CSV, JSON & SQLite exports
  • REST API, TypeScript SDK, webhooks & MCP
Start with 20 free pages

No credit card required.

Basic

Process documents regularly with room to grow.

$29.99/ month
  • 800 pages per month
  • 10 GB storage
  • Automated email intake & extraction
  • Classification, routing & PDF splitting
  • Human review
  • Custom schemas & queryable tables
  • CSV, JSON & SQLite exports
  • REST API, TypeScript SDK, webhooks & MCP
Get started

Monthly page allowance.

Pro

Handle larger monthly document volumes.

$79.99/ month
  • 3,000 pages per month
  • 50 GB storage
  • Automated email intake & extraction
  • Classification, routing & PDF splitting
  • Human review
  • Custom schemas & queryable tables
  • CSV, JSON & SQLite exports
  • REST API, TypeScript SDK, webhooks & MCP
Get started

Monthly page allowance.

Enterprise

Process high volumes with more storage.

$129.99/ month
  • 10,000 pages per month
  • 100 GB storage
  • Automated email intake & extraction
  • Classification, routing & PDF splitting
  • Human review
  • Custom schemas & queryable tables
  • CSV, JSON & SQLite exports
  • REST API, TypeScript SDK, webhooks & MCP
Get started

Monthly page allowance.

The free plan includes a one-time page allowance. Paid page allowances renew monthly. Prices shown in USD.

Guides

Learn the workflow step by step.

All articles ↗

Questions, answered

Before you upload your first PDF.

A few practical details about extraction, exports, and getting started.

Have a different question? Contact us ↗

The Free plan includes 20 pages total,50 MiB of storage, and custom extraction schemas. Create an account and verify your email to get started. No credit card is required, and the free page allowance does not reset each month.

Bring a PDF. Leave with a table.

Your first 20 pages are free. Describe the fields, extract your own documents, and check every row before you choose a plan.

Just one file? Try the PDF-to-CSV converter ↗
Start with 20 free pages ↗

One-time allowance. No credit card required.