The pain: saves that rot
You bookmarked the video. You queued the podcast. You saved the article. None of it is searchable by what was actually said. Playlists are graveyards. The one quote you need is buried in a 90-minute interview you can't skim.
What BrainTube does
- Auto-captures from YouTube with segmented transcripts.
- Transcribes podcasts and any audio you save.
- Ingests PDFs, EPUBs, web articles, Notion (4-hour sync), meeting audio, and screenshots.
- Builds per-item semantic embeddings and pulls out entities.
- Links everything in a Postgres knowledge graph with PageRank-style retrieval.
The result: you can ask "that thing the founder said about pricing leverage" and land on the exact segment — across every source you've ever saved.
Query it from any AI
BrainTube exposes your library through a live MCP server (in the official MCP registry and on Smithery). Claude, ChatGPT, Cursor — and any future MCP-capable model — read straight from your corpus. No copy-paste, no re-explaining. Here's how the MCP memory layer works.
Why nothing else does this
Note apps make you type. Bookmark apps store URLs, not meaning. BrainTube is the only memory layer where the index fills itself from media — and is queryable from every AI you'll use, not just one.
