Who hurts
Independent podcasters at 1k–50k downloads who can't afford an editor.
The problem
Recording is 10% of the work. Distribution is the other 90%, it's unglamorous, and it's the reason most shows die at episode twelve.
What you build
An upload becomes a review queue: five clip candidates ranked by a hook score, a thread, a newsletter draft and chaptered show notes. The creator approves or kills each one; nothing publishes without a tap.
Why now
Whisper-class transcription is nearly free and clip selection is now a solvable ranking problem rather than a taste problem.
Validate it this week
Pick ten shows in one niche, do it manually for free, and ask for a testimonial rather than money. If nobody will give you a testimonial, the output isn't good enough.
Why you'd keep winning
Niche-specific hook models. What lands for a true-crime show is not what lands for a B2B interview show, and generic tools average across both.
The honest risk
This is the most crowded idea on this page. Do not build it generically — pick one vertical and be visibly the best there, or don't start.
The prompt is written to make an AI argue with you before it writes code — that first round of pushback is worth more than the scaffold.