Exploring the potential of AI in prototyping

Hey... in the last few weeks I have been asking myself a lot about how AI is going to change the way we work in Business design, and more specifically in my field of work: Prototyping. One of the key things in prototyping is speed. We have a lot of uncertainties and hypotheses that we need to reduce and test as quickly as possible when we are building something new. This is what we do with prototyping.

Especially when building MVPs, there are a lot of steps involved in the process: We have to come up with a design (ui/ux) and we have to build the code for the front end and the back end. AI already helps us write code (hello Github Copilot). But can AI help us with the design and the frontend? Most of the time we start with a blank canvas. We have to come up with color schemes, visuals, images, layouts and so on. Can the AI help us with that already?

There are a number of text-to-image AI models out there that can do image generation. The most popular are Stable Diffusion, Dall-E and Midjourney. These are trained on billions (Stable Diffusion: 2.3 billion) of images from the web (Pinterest, Flickr, DevianArt...). I knew they were pretty good at generating art and people. But can we use them to generate landing pages and software user interfaces? I took Midjourney for a spin, and after a bit of experimenting with the prompts, I was able to generate the following images for an input such as "modern landing page for a coffee company":

Hm...I'm still a bit overwhelmed and don't really know what to think. But I can clearly see how it's going to help me in what I'm doing in the future. Sure, there's a lot of stuff on there that doesn't make sense. But I think it can still be helpful to get inspiration for color schemes, image composition, layout ideas and even some "ready to use" visuals. This can definitely speed up the process of building "high-fidelity prototypes" quite a bit.

In the past, it took a lot of energy and time to build ONE solution. With these tools, it takes seconds to build unlimited solutions. Now you can spend your valuable time and energy to think more about what's important to you. What fits you. Your focus can change. And that's a good thing!

This works not only for landing pages. It also works for other types of UIs such as dashboards, mobile apps, etc. You can also use it to generate app icons, logos and other image assets.

The fact that any kid with a smartphone can create all this in 2 minutes still blows my mind. And if you think about it: These models are being trained on random images taken from the internet. We will soon see models trained purely on landing pages and software UIs. And this is just the beginning. This will all get so much better over time.

It's still early days, but I'm very excited about how this will be a huge help to all of us in the very near future.