Building the MVP
Kyle tweeted that he wanted to ‘tinker with AI’ in mid October, 2022. One week later, he announced that he was working on his first AI product idea.
He shared an ‘interesting’ example of the first prompt. Within hours, he got the stable diffusion prompts good enough for an MVP. Despite the improvements, he had issues with the amount of time it took to guess different prompts. He learned that using the same seed each time produces the same image, which made it easier to test prompts.
AI Logo Tool
Kyle tweeted that he was building an AI logo tool in late October. At that point, the output was an equal mixture of useful logos and complete garbage.
He tweeted a link to the prototype at AI Graphics. He described it as a logo generator. He also said that if he couldn’t get the app to consistently create consistent quality logos, he’d pivot to generating marketing/advertising images.
By October 28, the service had generated 700+ AI logos. He decided to test interest by limiting the amount of free logos people could get before asking for an email address. This increased signups by 42%.
Validating AIGraphics
Kyle had many feature ideas for the product, but hadn’t validated the business. One of the ideas was that the images didn’t have to be perfect; they could serve as inspiration for designers.
He bought the aigraphics.io domain name and expanded the concept. The service would now generate different types of graphics. He started testing different graphic types to validate demand. He built buttons for every feature with an email capture, and built the most requested features. At this point, he didn’t even have a pricing page.
He added subscription plans in early November. He tweeted a graphic showing the different plans. He got his first paying customer within a day – despite not having a landing page!