Distill

unlisted

by CarbeneAI

Turn any video into knowledge. Timestamp-aware AI that distills videos into summaries, wisdom, linked notes, and Anki flashcards.

Updated 1mo agoMIT
View on GitHub
Distill logo

Distill

Turn any video into knowledge.

Distill

License: MIT Obsidian Buy Me a Coffee

A faster, cleaner way to save and study YouTube videos in Obsidian. Think of it as MediaNote, rebuilt: one-click capture, a genuinely resizable player, click-to-seek timestamps, and AI notes, all in your vault.

What it does today

One-click capture. Install the bookmarklet and save any YouTube video to your vault as a clean note: a clean URL that always starts at 0:00 (no more resume-timestamp), the real video title as the filename (no junk index, no illegal characters), and tidy properties (channel, video id, source).

A real player. Open the video in a resizable, pop-out pane. Drag it as large as you like, or move it into its own window onto a second monitor. The tiny fixed embed is gone.

Click-to-seek timestamps. Tap a hotkey to drop a timestamp while you watch; click it later and the player jumps to that exact second.

AI notes, powered by Fabric. summarize and extract_wisdom, appended to the note automatically.

A note on transcripts (the honest part)

YouTube has locked down direct caption access, so a pure in-plugin transcript fetch isn't reliable yet. Today the AI summaries run best through a companion transcript workflow (for example a yt-dlp-based job) or your own transcript. A portable transcript path is the number-one roadmap item. Until then, capture, the player, and timestamps work for everyone, and the AI works wherever a transcript is available.

Install

  • Bookmarklet (capture): drag-to-install at carbene.ai/distill/install.html.
  • Plugin (player + timestamps): manual for now. Copy the build into .obsidian/plugins/distill/. A Community Plugins listing comes once it's polished.

Roadmap

We will flesh this out later. The short version:

  • Portable transcripts, so the AI works anywhere, on any device
  • Anki flashcards generated from your timestamps
  • Graph-linking: people and concepts become atomic [[notes]]
  • More sources: podcasts, articles, local files

Tech

  • Obsidian plugin (TypeScript)
  • Fabric patterns for the AI layer
  • Cloud LLMs or a local Ollama

Support

Free and open source under the MIT license. If it saves you time, you can fuel the build:

Buy Me a Coffee

Credits

Built by CarbeneAI. AI extraction powered by Fabric. MIT licensed.

For plugin developers

Search results and similarity scores are powered by semantic analysis of your plugin's README. If your plugin isn't appearing for searches you'd expect, try updating your README to clearly describe your plugin's purpose, features, and use cases.