Launching Artwurk.AI
In 2022, I launched Artwurk.AI which back then was called aialbumcover.com. The idea came to me after poking around with OpenAI's Dall-e model. As a music producer/artist, I thought this would be good for people like me who just want a quick photo to use as the cover art for a song. Previously, I used free stock photos from websites like Pexels.com, but it would always take too much time to find the perfect photo. With aialbumcover.com I thought, I would simply ask the AI what I wanted and get exactly that or something close to that.
I launched the website early in the year; it was pretty basic. Essentially, it simply made an Api call to OpenAI, received the image or images, and displayed them to the user. I was actually excited about it. I imagined all the users that were going to love this. I thought, if I find this useful, I'm sure other producers/artists will too. I dm'ed a number of music producers and artists on Instagram and told them about the website. I got a few of them to actually try it.
I gave every new user 10 free images with the option to buy 100 credits for just $5. I thought, $5 is nothing if you get to generate 100 cool images, right? Guess how much money I made? $0. Nobody bought any credits. The website had just under 100 users who would use the 10 free images and never return.
After a while, I kind of shifted focus on other projects. Then one day, I had a thought. What if I just make it free? I could build up a database of users and then figure out how to make money later.
Make it free to make money?
I reworked the website a bit. I updated the UI and made the site completely free. I also decided to display every single generated picture on the landing page (I copied Lexica.art's design). I figured this way I can at least build some time of stock photos platform by letting users create these images that I can figure out how to monitize later. I didn't really bother with marketing it. I just kind of moved on to other things.
After a while I noticed that the number of photos on the landing page kept on growing. People were actually using this. I checked the database, and the users table had grown to about 300 users. Most were just one- or two-times users and never came back. A handful of them were returning users who had probably bookmarked the website and would go to it when they needed an image.
What is this new thing?
In 2023, I noticed some accounts that I follow on X post about Stable Diffusion. I immediately had a new idea. Artwurk.AI can let artists create their own AI models and generate song/album covers with their face on them. They will no longer need to hire an expensive photographer or a graphic designer to edit their picture into a cover art. This also sounds much better than generating random AI images for an album cover. This actually sounds like it has a real use case. That's it! I was going to build this into Artwurk.AI.
I started working on adding Stable Diffusion to the site but made no real progress. Between constantly getting distracted by new ideas that never saw the light of day, working a full-time job, and becoming a dad, I made no real progress. Another thing that made it hard to complete the project was seeing all the hacks other people had to do just to get the model to spit out decent looking images. I had no motivation to see this through.
Evolving Artwurk.AI
Fast forward to 2024, everyone is talking about this new model called Flux. It apparently just works! No hacks needed. I log into my replicate account quickly, create a new Flux model and generate a few photos. I was impressed. I immediately thought about my abandoned project. I wasted no time and got right to work.
After a few weeks of work, I finally had it working. Then I was faced with yet another thought. Why just artists? Why only song covers? Isn't this a bit limiting? The tool is so powerful that limiting it to just album covers seems like a waste. So, I said fuck it! I decided to make it into a website that just lets you create awesome realistic photos of yourself. Additionally, I would add templates that people can use to generate specific photos without having to enter a prompt.
Are people actually going to use this
After successfully making all the changes to the site, It is finally online and fully functional although I'm still addressing a few things and improving it every day. Yet I couldn't help but invite new thoughts into my head. Are people actually going to pay for this? Since launching it, I've had a few users log into the app. Some even created a model using the free credits that every new user gets but no one has paid for additional credits.
It left me wondering yet again, are people actually going to pay to create AI images of themselves? Does this even solve a real-world problem?
On a encouraging note, I come across a different ad about a different AI app that creates photos of you in 1970 every single day. There must be a market for it then, right? I guess we'll just have to find out.