AI Weekly Malaysia

Brooke Lacey, who has built a TikTok audience of more than 290,000, demonstrates a workflow where Claude Code drives FFmpeg and OpenAI's Whisper to turn a raw phone video into a captioned, jump-cut TikTok in roughly 15 seconds on her own computer. She describes teaching Claude her jump-cut cadence, hooks, and sense of what 'funny' sounds like, mining her TikTok comments for new scripts, and running a weekly short-form operation with no editor. The episode also covers where the workflow breaks — 'lane OS' and status files — and how she prompts Claude back on track.

Why: The reusable pattern here is not 'AI edits video' — it's wiring a coding-agent CLI to existing CLI tools (FFmpeg for cuts/captions, Whisper for transcription) so one prompt runs a full media batch job locally, with no per-seat editor subscription. The second transferable piece is codifying your style as instructions the agent re-reads (jump-cut cadence, hook format), plus status files to recover when the agent drifts — the same structure you'd want for any repeatable agent task, not just video.