01. Intro

what is this about anyway?

The thought of using AI to design felt beneath me. I was a designer, I didn't want AI taking my job. So why use it?

Why do I keep reading posts and articles about managements pushing it so hard on other designers?

Are people's talents just not good enough anymore...?

It was a fast transition. I resisted as much as I could, but finally the words were said out loud:

"We got a Claude subscription. Claude Design seems like it has a lot to offer."

I winced internally and said I'd check it out.

The more I looked into Claude and using it to design, the more I realized that if I didn't jump on this, I was going to fall behind. There was only one other time I'd felt this way, and that was when Figma was introduced to the scene. I'd been using Photoshop for mockups, and dabbled in InVision and Axure. Then suddenly I was at a new job and they were using this thing called Figma. What the heck was a Figma? But as I learned its benefits and got past the learning curve, it became, and still is, my go to mockup and prototyping tool. So now AI is the new Figma? Fine. Better to try it out before I'm left behind again.

But I needed a purpose to use it. The thought of dropping an already fully designed and developed product into it and prompting it to "make it better" wasn't something I was a big fan of. So I started thinking about how I could explore this with something new instead. Problem was, there were no new projects on the horizon that would warrant a brand new interface. So I decided this was actually a great time for a design exercise I had been thinking about, one that would benefit me, and if the company wanted it, them too.

At Within3, we have our star product, Launch Intelligence Platform. It officially launched about a year ago, and since then it's grown and added more and more into its teeny little one-page UI. The growing pains were becoming pretty apparent, and I could see that eventually we'd need a redesign. Of course, no one was quite ready for that conversation except me, so I figured this was the perfect opportunity to play around with AI design. I had my list of customer and sales feedback, management's eventual to-do list, and a dream. No questions, no deadlines, no problems..

And so I started to prompt...

02. First Impressions

never a second chance for these

As the prompt began to generate, I was immediately struck that it was building out these pages with HTML and CSS and interactive components. I hadn't expected that but was intrigued. At best I was expecting maybe a flat image of what a page could look like and maybe a prompt to generate some other pages it suggested to use for inspiration.

Usually to find inspiration I'd have to scour aross sites like Dribbble or BeHance to see if anything seemed remotely similar to what I was working on. Making sure I was keeping up with latest design trends without being too overly trendy. Most of the time I just found some pretty, but not very practical designs which felt kind of like a waste of time. Having this come up with a few different ideas based on what I'm looking for was actually... kinda nice. I've never had the ability to just find examples of something I was looking for that easily. And to then easily give me three more different types of design, that was neat.

03. Challenges

challenge: accepted

So the generated page was ready to download or view.

There seemed to be elements that were usable, but overall it really needed help. There were so many components grabbing my attention that I really didn't know where to look or what to focus on. The eye fatigue started hitting right away, with nothing to guide my eye. The insight itself was an afterthought, and that was supposed to be what we wanted users to focus on. There were so many data points thrown at you. I take pride in the usability of my designs and making sure, first and foremost, that they are usable first and good looking second. You couldn't give this to a client and expect a user to know how to use it. This didn't solve what the user needed, it didn't solve the problem, it merely threw data at you and hoped something would stick.

But I confess, it wasn't all negative. I did like how it laid out some of the charts and KPIs, so I kept that idea in my back pocket. That wasn't a bad suggestion at all.

Since I wanted to expand this product from one page to multiple pages, I started to form an idea of the information architecture I wanted to support for this redesign. Filling multiple pages from just a single page of content seemed like a lot of work to pull off. Would I have enough information? So back to Claude I went, prompting away on how to fill these new pages with usable, organized data.

04. And How it Ended Up

it's both the journey and destination?

As the AI page examples generated, so did the list of ideas I actually really liked. There was one page Claude generated that would list all the available reports and generated reports, and I remember audibly saying "Oooo" when I clicked the "View page" button on Claude's chat interface..

It finally clicked.

This AI stuff wasn't very good at creating layouts, but boy, did it have some absolute bangers of inspiration and ideas for how to organize the data. I could feel that AI wall I had built up so high starting to crumble after working on this project. Not only did it show me what could work, it also showed me what absolutely did not work. I actually became more confident knowing that my job as a designer was safe. If all websites and applications start using AI, nothing will stand out anymore. Products will lose their identity and what makes them special.

Long term, AI may be a new tool, but it won't replace the designer. It will just become my job to tame this wild AI beast, to enhance my projects and use it for good.

Below are some of the pages that Claude and I created in the end. I used my design system with some of Claude's ideas, and I actually really loved how it turned out. It was a great design exercise, and I'm more prepared for the future because of it.