RAG Converter
Turn everyday business files into AI-readable chunks and embeddings. The free converter runs in the tab. Developers POST the same document to /v1/convert and get the same JSONL back.
Last updated: August 29, 2026
Everyday business files, AI-readable
PDF, Word, PowerPoint, Excel, Markdown, HTML, CSV, JSON, images, and audio (WAV, FLAC). The converter splits on paragraph and sentence boundaries into overlapping passages — about 256 tokens with roughly 32 tokens of overlap — and writes JSONL you can load into SpaceDB, Chroma, Qdrant, pgvector, LanceDB, Weaviate, Pinecone, or Milvus.
Scans and screenshots go through Carmenta (OCR). Recordings go through Mercury. Images and video frames go through Argus. One schema, one load script. The long form is Convert business files to AI-readable RAG.
Browser tab or developer API
The free path is WebAssembly in your browser. Files are read by the page and never sent to a server. The only network call is a quota check that carries a number. No account.
The paid API is for pipelines: POST /v1/convert returns the same JSONL a person would download. It runs the full-size models a tab cannot — OCR at native resolution, speech, video — and has no file-size theater. Sign-in is mID / Sovereign ID. API docs live on ragconverter.com/api.
Built from the Remade with Rust stack
No LibreOffice, no OpenCV, no system packages. Word, PowerPoint, and Excel go through rusty_xml. Images through rusty_jpeg / rusty_png. Audio through rusty_flac. Allocation through rusty_alloc, including in WASM. The AI engines are the FFAI trio — Carmenta, Mercury, Argus. Storage seam is SpaceDB.
Why the browser path stays private · Mercury · Freedom Guide