GTM Sauce

A fitness app for people over 50 · B2C · host's own app

Have AI write ad angles and scripts from real competitor ads, then generate videos

Ads made; not run yet, host will report numbers later

Unknownpaid-social

What they did

Host worked entirely in Claude via MaxFusion's MCP: uploaded the app icon, asked 'what would be some good ads for this type of app that's targeting early 50-year-olds?', kept the angles he liked, then asked for more by awareness stage (solution-oriented, or problem-aware: 'you should be doing something for mobility as you age'). Claude scripted the ads, added an end card on request ('Start your habit today. Try it for free.') and synced the videos. Guest's advice: start from a research layer of real competitor ads (hooks, angles by funnel stage, offers, CTAs) rather than asking a chatbot cold, which gives generic scripts. Result: 8 AI videos plus 2 team-made ones, to run on Sunday-Tuesday, the app's best days.

What happened

Videos made; performance not yet known
6:20
“I've been using it to create my own ads for my app, and I'm going to run it this weekend. You know, for our fitness app, Sundays through Tuesdays generally do well”
Steve P. Young
27:42
“I was like, "Hey, what would be some good ads for this type of app that's targeting early 50-year-olds?" And so, it came up with different angles”
Steve P. Young
9:31
“So we added the research lab first because very important to have a data layer with real hooks, with real-performing ads, with real offers. Now you can feed that into Claude and get better scripts.”
Stav Zilbershtein

Related topics

Get tactics like this every Monday

New tactics from the week's founder interviews, each linked to where it was said.

More from this episode