UX will win the AI market
I recently wrote a piece about how I test ran five AI agent platforms. I had a few reflections on the experience:
Workflows are not intuitive and interfaces are uninspiring.
Onboarding is weak and often relies on off-platform learning.
Users are having to do too much to complete workflows.
Cognitive load is too high.
When issues arise, it is too hard or sometimes even impossible to correct them.
All of these issues could have easily been avoided with the proper application of UX research and design. I am not surprised that these corners were cut. Under the most common of circumstances teams will move without the data or design, and when it comes to AI tools, we are not experiencing anything common. Companies are in such a rush to get their products out onto the market that they are cutting everything, including staff, to do it. As someone who was part of a 60% RIF on my UX team, I’ve seen it firsthand. The goal is to move quickly and ship the most basic MVP before someone else does.
The problem with that is that now the market is flooded with products claiming to do the same thing, so that rush was not really all that worth it. I tested five AI agent tools, but there were more than double that. Look at so many AI products — assistants, vibe coding platforms, video tools, writing tools — and you will see just a slew of them available. Something has to help cut through the noise and find products of quality people will actually use.
In the end I could not get most to do what I wanted them to, and when I did find some success, it was hard-won with a lot of effort. That’s the exact opposite of what the world of AI is promising us. Needless to say, I did not adopt any of the platforms. I don’t think most people would.
So now we have a glut of AI tools with poor functionality that even the more tech-savvy of users will not adopt. The only recourse is to figure out how to make them easier to use. Companies that pushed out MVP fast now have a significant amount of tech debt to address. Those who wait a bit longer, do due diligence by their users, and ship something with better UX, will be more successful when they do.
