An indie app developer ran an OpenClaw agent on an old gaming PC to research hooks, generate six-slide TikTok carousels and upload them as drafts. He says it passed 500,000 views in under a week.
- First reported
- x.com · 12 Feb 2026
- Added
- 13 Oct 2026
- Source checked
- 13 Oct 2026
What they did
The agent, running Claude as its model, generated images through an image API, wrote its own code for text overlays, wrote captions and uploaded each slideshow to TikTok as a draft through a scheduling tool. It reused one detailed scene description across all six slides so the images stayed consistent, and logged every lesson in memory and skill files. He picked a trending sound and published each post himself, which he says took about a minute.
What happened
The poster says the slideshows passed 500,000 views in under a week, with four posts above 100,000, and brought in paying subscribers across two apps.
Keep in mind
Early posts failed on landscape images, tiny text and self-focused hooks, and local image models looked too artificial. The figures are self-reported in a post with affiliate links.
Try it with your agent
Each day, research what hooks are working for <app category> on TikTok and write three hook options for <your app>. Generate a six-slide portrait slideshow that reuses one scene description so every slide looks consistent, with large text placed clear of the app interface. Upload it as a draft to <scheduling tool>, and log what worked and what failed in <lessons file>.
Replace the <angle brackets> with your own details.
Original source
x.comUser storySomeone who uses the agent described their own setup. Figures are theirs.
