AI Changed What Your Users Expect. Most Products Didn’t Get The Memo.

Why user behavior, not your roadmap, decides whether your product actually succeeds

I want to start with something I believe more strongly every year I do this work: your product's success was never really up to you. It's up to how well your product matches what your user already expects, before they've even opened it.

That sounds obvious when I say it out loud. But if it were actually obvious, I wouldn't be looking at well-funded, well-established products every week that are still built for a version of user behavior that quietly stopped existing a while ago.

Behavior moved. A lot of products stayed still.

Here's what I keep seeing. AI has reset what "normal" feels like for your users, and it happened fast enough that most product teams haven't caught up. People now expect a system to understand intent, not just accept input. They expect a search bar to feel like a conversation, not a query field. They expect friction to be the exception, not the default state they have to work around.

And yet. I open products every week, some of them from companies that have been around for a decade, and I find interfaces still asking the user to think like the system instead of the system adapting to think like the user. That gap didn't used to matter as much. Now it does, because your user has a dozen other tabs open where something else already closed that gap for them.

If your product isn't built around how people are actually expecting it to behave right now, it doesn't matter how solid your engineering is underneath it. You're not being judged against your last release. You're being judged against the best experience your user had this morning, on a completely different app.

The framework I keep coming back to

This is where I want to bring in Don Norman, because decades before AI, he was already describing exactly the failure I'm talking about. Norman calls it the Gulf of Execution and the Gulf of Evaluation, and once you understand it, you'll start seeing it everywhere.

The Gulf of Execution is the gap between what a user wants to do and what the system actually lets them do easily. It's the distance between someone's intention and the actions the interface demands of them to make that intention happen. Every extra click, every unclear label, every moment where a user has to stop and figure out "how do I even do this" widens that gulf.

The Gulf of Evaluation is the other side of it: the gap between what the system just did and whether the user can actually tell what happened. Did that save? Did that request go through? Is it thinking, or is it broken? If your user has to guess, that gulf is wide open, and every open gulf is a small tax on trust that compounds every time they use your product.

Here's the part I want you to actually sit with: AI didn't create these two gulfs. Norman named them long before any of this. What AI did was raise the bar for how narrow those gulfs are allowed to be before your product starts feeling outdated.

Why AI is the thing that finally closes these gulfs, not just describes them

For years, closing the Gulf of Execution meant better information architecture, clearer labels, smarter defaults, a lot of careful human design work. Necessary, but slow, and it required your team to predict every way a user might approach a task.

AI changes that equation completely. Instead of designing one fixed path and hoping it matches most users' mental models, you can build a system that interprets intent directly, lets someone say what they want in their own words, and figures out the execution path for them. That's not a small UX improvement. That's the Gulf of Execution shrinking in real time, per user, per session, instead of shrinking once per design sprint.

The same goes for the Gulf of Evaluation. AI-driven systems can now explain themselves as they act, confirm what just happened in plain language, and flag uncertainty instead of leaving a user staring at a spinner wondering if anything is happening at all. That's not decoration. That's you finally closing a gap Norman was describing back when the closest thing to AI feedback was a beep.

The actual question for your product

So here's what I want you to ask yourself honestly, not about your competitors, about your own product. When your user opens it today, is the distance between what they want to do and what they can actually do getting narrower, the way their expectations already have? Or is your product quietly asking them to do the translating, the way products used to get away with a few years ago?

Because your users aren't measuring you against where you started. They're measuring you against the gulf they didn't have to cross somewhere else this morning.

Close that gap, or someone else will close it for your users while you're still shipping the next feature nobody asked to think harder for.

Before you go

• Please take a moment to like the post and follow the writer!
• Did you know that over 400,000 developers share what they're building, learning, and discovering across our platforms every month? Learn how you can contribute here