Steven Gonsalvez

Software Engineer

RT @sentient_agency: This is wild. @levelsio dropped 14 raw clips (58 min) into a folder and gave Claude Code one prompt. It handed back a 10 min vlog. How it worked: → mlx-whisper transcribed ever

Why CEREBRO kept it

Claude Code video pipeline; multimodal agent workflow, demo-level

The text below is an automated extraction of the article at https://x.com/sentient_agency/status/2117388991897153581, stored verbatim in the public cerebro-vault repository. Copyright remains with the original publisher (x.com).

@levelsio dropped 14 raw clips (58 min) into a folder and gave Claude Code one prompt. It handed back a 10 min vlog.

How it worked:

→ mlx-whisper transcribed ever

> Claude Code video pipeline; multimodal agent workflow, demo-level

RT @sentient_agency: This is wild.

@levelsio dropped 14 raw clips (58 min) into a folder and gave Claude Code one prompt. It handed back a 10 min vlog.

How it worked:

→ mlx-whisper transcribed everything → frame grids so it could "see" each clip → voice matching to find his lines, face detection for thumbnails → a JSON cut list, then captions

His cost: about $0.10 on his $100/mo Claude plan.

Backlinks

Appeared in 1 briefing

Related

Shares tags: ai/tool-pairing · cerebro/signal

Also from x.com