Agent lab
Video AI Agent
An agent that watches a long video and gives you the parts that matter: chapters, a summary, quotes with timestamps, and clips cut for social.
- tools chained
- 4tools chained
- max input
- 90 minmax input
- source walkthrough
- Opensource walkthrough
What you get
How this one is different
Transcribe and diarise
Speech to text with speaker labels, so a two-hour panel becomes searchable text.
Chapter automatically
The agent segments by topic shift rather than fixed intervals, then titles each section.
Pull quotable moments
Ranked highlights with timestamps you can jump to, plus a one-paragraph summary.
Cut short clips
Vertical crops with burned-in captions, ready for a feed.
Detail
How it is built
- 01Whisper-class transcription with speaker segmentation
- 02Topic segmentation over the transcript
- 03LLM summarisation with citation back to timestamps
- 04FFmpeg clipping driven by tool calls
- 05A queue for long jobs and partial results
- 06Evaluation set to catch hallucinated quotes
Questions
Before you start
Members get access during the agent lab, and the full build walkthrough is part of Module 4.
JOIN NOW
Begin the first module
It is free, it is the real curriculum, and if it is not for you, you have lost nothing but an evening.
Join any time · Build AI skills at your pace