All recipes

Automate

Build a RAG knowledge base from a channel

What it does

Raw transcripts make muddy retrieval units. The distilled archive chunks cleanly — claims already carry their sources, entities their relations — so your RAG pipeline starts from structure instead of noise.

How it works

  1. Get the archive

    Buy the channel's archive — entities, claims, topics, and transcripts in one zip.

  2. Chunk it your way

    Feed it to your pipeline: the structure means clean chunks with real provenance.

  3. Retrieve with citations

    Retrieval hits carry their source video and claim — no orphan quotes.

What you'll need

With an AI agent

  • A purchased skill archive
  • Your RAG or embedding pipeline

Offline, from the zip

  • A purchased skill archive
  • Python or any JSON-processing toolchain

How to use it

Connect your agent, then start with a prompt like:

Chunk this channel's knowledge graph into retrieval units, preserving claim provenance.
Find a channel to distillSee pricing