{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/40ee2d2267d644b9a5d6f9e17bd30fa5\" frameborder=\"0\" width=\"1724\" height=\"1293\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1293,"width":1724,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1293,"thumbnail_width":1724,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/40ee2d2267d644b9a5d6f9e17bd30fa5-bc83c8e6800e97a8.gif","duration":881.826,"title":"Turn YouTube Videos into Agent Knowledge","description":"This Loom explains how Cade productizes an MCP-based system to streamline indexing YouTube videos into searchable, timestamped knowledge for AI agents. He describes starting with tools like ChatGPT, YouTube transcript extraction, and earlier work on his open source Clip Finder, then building a self-hosted web app that ingests video URLs, creates transcript chunks, and returns a TLDR with timestamps. He highlights that the workflow removes the previous siloed notebook or session problem, replacing it with agent-agnostic context that persists across platforms. He also shares examples like extracting key five points, generating flashcards, and syncing whole YouTube playlists; indexing a video previously took about five minutes for agent context."}