I built an AI thumbnail generator for creators who hate starting from a blank canvas

Hey Indie Hackers 👋

I’m building Miniagen, an AI thumbnail generator for YouTubers, creators, agencies, and brands.

The problem I kept seeing is that thumbnails are one of the highest-leverage parts of a video, but they’re also slow and annoying to make. A creator can spend hours recording and editing, then rush the thumbnail at the end because opening Canva or Photoshop feels like another full task.

Miniagen is my attempt to make that workflow much faster.

You can start from:

a YouTube link
a video idea
an existing thumbnail or image

Then Miniagen generates multiple thumbnail variations designed for YouTube, Shorts, Reels, and TikTok-style content. The goal is not just “make an image with AI”, but to help creators explore stronger visual angles faster.

What I’m working on now:

better style selection
more useful variations
clearer editing controls
making outputs feel less generic and more platform-native

I’d really appreciate feedback from other founders and creators:

Would you use an AI tool for thumbnail creation, or do you still prefer manual design tools?

And what would make this useful enough for you to try it on a real video?

Here’s the project: https://miniagen.co