Intelcraft is a word I’m making up. Yep. It’s the age of generative ai, and if there are going to be some new generations, it may as well come from bio AI, like us.
One of my goals of intelcraft (the open source education /community for people who are learning/teaching AI) is to come up with one prompt that create a production grade system.
Unfortunately , the type of production grade systems our company has design, built, rescued for Fortune 50/500 companies , specifically global platforms, the tools don’t exist yet to do that.
Luckily, we do have Claude Code / Cursor for creating relatively small systems.
graph TD
dpf[document-perception-framework] --> providers
providers --> lib & api
lib --> pymypdf4llm[pymypdf4llm<br>/pdf] & ocrmypdf[ocrmypdf<br>/pdf/ocr] & pytesseract[pytesseract<br>/*/pdf/ocr]
api --> llama-parse[llama-parse<br>/*/pdf/agent/ocr] & mistral-ocr[mistral-ocr<br>/*/pdf/llm/ocr]
https://github.com/xingh/docsray-mcp
TODO: screenshot
I made a Document perception and analysis MCP server in about a day, and about a day of testing, we were ready to publish it. This is how: