PDF
About PDFPipe
The PDF conversion API built for workflows that can't afford to break.
PDFs are everywhere - invoices, reports, contracts, government filings - but getting the text out of them programmatically is harder than it should be. Most tools work fine for simple, inline PDFs. But the real world is messier: tokenised download links, JavaScript-triggered attachments, redirect chains.
PDFPipe was built to solve that. One API endpoint that handles both inline and auto-download PDFs, returning the PDF's text in any of 10 formats - JSON, Markdown, CSV, a first-page image, and more. No browser automation to manage, no infrastructure to maintain.
Whether you're building an invoice processing pipeline, a document ingestion system, or connecting PDFs to Zapier and Make, PDFPipe is the reliable layer between a PDF URL and the data you actually need.
What we care about
Reliability first
PDF links come in every shape - inline, behind redirects, token-in-URL, JS-triggered downloads. PDFPipe fetches them so you don't have to.
Developer experience
One endpoint, one API key, 10 output formats. Nothing to install, no binaries to manage, no servers to maintain. A Node.js SDK and an MCP server if you want them.
Privacy by design
Documents are processed ephemerally. No long-term storage, no training on your data, presigned URLs scoped to the exact object. Your PDFs stay yours.
Transparent pricing
A free tier to get started (300 requests a month, no card), clear per-tier limits, and overage billing only on Business at a published rate. Upgrade when you need to.
How PDFPipe works
When you send a URL to PDFPipe, we first issue a HEAD request to determine how the PDF is served. If it's an inline PDF, we fetch it directly. If the PDF triggers an auto-download - common with enterprise document portals, government sites, and tokenised links - we spin up a headless Chromium browser to navigate the page, handle redirects and JavaScript triggers, and capture the downloaded file. We do not send cookies or credentials, so login-walled PDFs are out of scope.
Either way, the PDF goes through the same extractor (pdf.js via pdf-parse) and you get the same output: the text layer of each page plus the PDF's own metadata, in your choice of 10 formats. JSON gives you an array of pages. Markdown gives you one section per page for LLM pipelines. CSV gives you one row per page. Image formats render the first page as PNG, JPG, or WebP. There is no OCR, so scanned PDFs return no text, and no table or layout detection.
The entire process is ephemeral. PDFs are processed in memory, results are stored temporarily via presigned S3 URLs scoped to the exact object, and everything is cleaned up automatically. We never store your documents long-term and never use them for training.
Built for real workflows
Invoice processing
Fetch invoice PDFs from supplier download links and get their text as JSON, ready for your parser or an LLM to pull out totals and line items.
Document ingestion for AI
Convert PDFs to Markdown for RAG pipelines, vector databases, and LLM context windows.
No-code automation
Connect via Power Automate, Zapier, or Make to process PDFs from SharePoint, email, or cloud storage.
Data extraction
Pull the text of financial reports, research papers, and government filings into JSON or CSV, one row per page, for analysis.
Ready to try it?
Start converting PDFs in minutes - no credit card required.