illostrationAI

One current issue with illostration.com is that my target group is too broad. I have to define a more precise customer group.

There are multiple options now:

• Students who need illustrations for their presentations
• Product managers who need illustrations for presentations
• Founders who need illustrations for their landing pages
• Founders who want to create logos for their next startups
• ...

I'm not interested in students because I want to focus on the B2B segment. Therefore, I need to figure out if product managers or founders need illustrations more urgently. I'll figure this out next through user research.

I posted the project to some startup directories and communities. I learned that most directories have a waiting queue, which can be multiple months. They all offer the option to skip this by purchasing a package, but I think the project is not in the stage to do so. The best directory so far is https://owwly.com/ where I got some users.

I also created a post in the IndieWorldWide community where I got some feedback about the user group.

The MVP is built with Next.js and Tailwind CSS for the frontend. The data is stored in Supabase, which also handles authentication and login with Google OAuth. I use Replicate.com as my AI provider, where I use the Stable Diffusion model. The website is hosted on Vercel.

I came across new AI tools for image generation. Instead of building the next avatar, I decided to focus on a specific type of image.

I had the problem of finding a nice illustration for a website. So currently you have to search through a lot of illustration websites, and you only get illustrations that have already been created. The problem I had is, that you find suitable illustrations for 4 out of 5 sections, but there is still one sections where you can not find a suitable illustration.

The solution I came up with is a website where you can create AI-generated illustrations. You can select your desired style from a list of presets and describe your scene. The AI will generate an image based on your inputs.

Mathias Michel