Markitdown

Microsoft built this to convert documents into Markdown so LLMs can actually work with them. It handles PDFs, Office files, images with OCR, audio transcription, and even YouTube URLs. The Python API is straightforward, and you can optionally pipe in an LLM client for AI-generated image descriptions in presentations. Honestly most useful when you're preprocessing a pile of mixed-format documents for RAG or analysis. Ships with a CLI for quick one-offs and supports Azure Document Intelligence if you need heavy-duty PDF parsing. MIT licensed, installs via pip with granular control over which format dependencies you want.