Why I built it
Every analytics tool tells you why a video flopped after it already flopped. I learned retention the slow way, one post at a time. HookLab is the feedback I wish I had before hitting publish.
Calls I made along the way
Split the AI work in two. One model watches the video and pulls out structured data. A second one writes the advice from that data. So the report sticks to what's actually in the video instead of making stuff up.
Every note gets a timestamp and every score gets a rubric. Creators can check the feedback themselves, which is the only way they trust it.
Playbooks per platform, niche and game. A Warzone clip and a real estate tour need different advice. I even split Call of Duty into separate modes because Zombies and Warzone audiences want totally different things. Ten niche playbooks so far, with gaming as the deepest.
Shipped v3 off real feedback. After my first two customers I rebuilt the UI and upgraded the models. Then a paid Instagram campaign got creators with millions of followers using it within two days.
- Role
- Solo founder. Product, design, code, pricing, marketing.
- Status
- Live with paying users, on v3
- Pipeline
- VideotoGemini 2.5 FlashtoStructured JSONtoClaudetoReport
- Stack
- Next.js, Vercel, Supabase (Postgres), Stripe
- Up next
- Hook rewrites sorted by style, and connecting creators' accounts so feedback uses their real metrics