Sinun AI
fi

Features

Local RAG — your own knowledge base without a cloud

Drop thousands of PDFs, emails and notes into a local folder. Sinun AI indexes them in a vector database — and answers questions from your own material offline.

sqlite-vecLanceDBOffline

RAG means augmenting an LLM's answers with your own material. Local RAG runs on your own hardware: thousands of documents, millisecond semantic search, not a byte in a cloud.

How it's built

Stack: an open embedding model (e.g. bge-m3), sqlite-vec or LanceDB as the vector database, SQLite for metadata. Indexing runs in the background. When you ask something, the assistant pulls relevant chunks from your material and hands them to the model as context.

What kinds of material?

PDF papers, Word documents, emails, meeting notes, audio transcripts, source code, contracts, board material. Typical scale: 10,000–1,000,000 pages.

Frequently asked

How long does indexing take?
On an M-series Mac, 10,000 PDFs index overnight. For large datasets a workstation (Framework Desktop, DGX Spark) speeds things up significantly.

Updated 2026-04-21

Related

Features

Local-first sync — no cloud, no leaks

Local-first sync over a SQLite + sqlite-vec + CRDT stack. Phone, laptop and workstation stay aligned without a central cloud — end-to-end encrypted.

Features

Offline AI — your work keeps going

Offline AI means AI that works without a network. Local open models (Viking, Qwen, Gemma) deliver full functionality on a train, on a plane, or in the field.

Features

End-to-end encrypted device sync — keys stay with you

End-to-end encrypted sync for AI memory and documents across your devices. Servers only see ciphertext; keys stay on your devices.

All resources

Want your own local AI assistant?

Tell us about your work and hardware — we'll map the right model, the right hardware tier and the right sync configuration.

Get in Touch