OCR integrations for Python, AI agents, Nextcloud, Haystack and LangChain
Pick the tool you already use and read your PDFs and scans in it. Each connector is a small MIT-licensed project that calls the same API as everything else, so there is no second OCR stack to run.
OCR integrations for the tools you already use
OCR in Python and on the command line
One command or one call reads a PDF or scan and returns text, Markdown, hOCR, JSON or a searchable PDF. Install from GitHub.
Python OCR client →
OCR for AI agents (MCP)
Your agent reads a scanned PDF through six tools; a person confirms every job first. Install from GitHub.
OCR MCP server →
OCR for Haystack
A converter component turns each file into a Haystack Document, and a page that failed is flagged in its metadata. Install from GitHub.
Haystack OCR converter →
OCR for LangChain
A document loader returns one LangChain Document per file, with load or lazy_load. Install from GitHub.
LangChain OCR loader →
OCR for Nextcloud
Every PDF in a folder becomes a searchable PDF; with the pipeline for hard pages, which you choose, a Markdown copy too. Install from GitHub.
Nextcloud OCR app →
The source of every connector is public: ocr-software on GitHub.
Run an OCR integration on your own server
Each connector also works against the package on your own server: you change the API address, and the calls stay the same (see the install guide).
Production use, high volume or your own server?
Private use and tests are free. For a company in production, a server of your own or a pipeline built on your pages, send the kind of page and the volume, and get a written offer.
