How my image AI project crossed $10K monthly revenue in 6 months

Hey all - I’m the developer of sinkin.ai, a service that hosts Stable Diffusion models on fast cloud GPUs.

I launched the site in February and revenue in July crossed $10K (so does August). I’m happy to share some experiences on how sinkin.ai was able to reach this milestone.

Be early

I did two other AI image projects before Sink In and they didn’t work out. During that process, I noticed the surge of custom Stable Diffusion models. For someone like me who doesn’t have a computer with powerful GPU, there was no easy way to run these models.

So the idea of hosting these models on cloud came to me. At that point, there were sites hosting the official Stable Diffusion models and some early custom models, but not these newly created models that were gaining popularity among SD community.

So I was early onto something.

Figure out distribution before building

Another problem I noticed was: model creators were not paid. They were doing free work.

I thought that should not be the case. These creators have great distribution channels: their model pages were getting a lot traction. We can partner up and form a win-win relationship.

So that led to the major distribution strategy for sinkin.ai: we share profit with model creators and they list our link on their model pages.

It turned out to be effective. People who are visiting these model pages are exactly the target users of Sink In. They’re interested in running these models but may not have powerful GPUs to run them. Sink In came in handy.

Launch fast

The idea marinated in my head for a couple weeks. When I decided to build, it took me about one week to launch the first version.

I copied a lot code from my previous AI image projects, that’s a big factor why I was able to launch in such a short time.

But the first version was also minimum, there wasn’t even terms of service or privacy page.

The idea here was: launch as fast as possible, only implement the parts that were absolutely required. Since if it’s not gonna work, all the efforts would be for nothing.

Launching fast also helped maintaining the edge gained by being early.

Adapt

The AI landscape has been moving very fast, so does the image AI world.

In the past 6 months, I made sure sinkin.ai always stayed at the head of the curve: covering the latest models, supporting ControlNet, supporting LoRA, etc.

Sink In was probably the first site that introduced some of these new technologies to a cloud service.

That helped us stay ahead of the competition and maintain the edge.

Final words

I’ll probably keep this post on those major points. There were other stuff that contributed, but I might save them for future sharing.

If you’re interested in my solopreneur journey, you can find me on twitter: https://twitter.com/boldjoeli

Thanks for reading!