June 2026 • 7 min read
Tags: AI, Product, UX Design
TL;DR: AI is no longer new enough for its mere presence in a product to be interesting. Users have tried it, been excited by it, been disappointed by it, adopted some capabilities and become skeptical of others. So the question is not whether we added AI, but whether it actually fits the user’s goal, replaces an existing usage pattern in a better way, or solves a problem that was not easy to solve before.
Recently, I opened Monday and saw a huge, well-designed message about the new agent they had added. They already have AI that helps with simple tasks, and now apparently something bigger was added.
I closed the message and moved on.
Why? Because for the specific tasks I do there, it is not worth going deeper into it. It is not interesting enough to me. I open Monday to do something specific, and if the existing way works well enough, a festive AI announcement does not necessarily change my priorities.
Something similar happened to me with Evrit. I like reading digital books, which means I also buy and borrow a lot of books. I always go to Evrit because that is where there are the most reviews for books in Hebrew. This week I saw that they added a button with a GPT icon, meant to help me understand whether a book is right for me.
I ignored that one too.
Partly because I have already tried getting personalized book recommendations from GPT, and they were not reliable enough. And partly because this is not what I am looking for when I go to Evrit. I am looking for ratings and reviews from real users.
For the purpose of this post, I went back to Evrit and clicked the new button. It did not work. It led to a broken link.
Users enter interfaces with a specific goal: to get information, apply for a job, buy a book, create a timesheet report, understand a status, complete an action.
If the product, form or website lets them achieve that goal easily, that is what matters. AI that does not fit into that goal is not interesting just because it is AI.
This is an easy point to miss inside the hype.
When a company adds an AI feature, it sees a new capability. The user sees another thing asking for attention, a decision, trust and a small change in behavior. If the value is not clear enough, very often they will simply close it, ignore it or return to the familiar pattern.
Nielsen Norman Group also wrote about AI feature discoverability, and the point there is important: even AI features that can provide value will not be adopted if people do not notice them, do not understand them, or cannot connect them to their mental model and the task they came to complete.
AI is no longer brand new. We have been dealing with it a lot for about three years. That means we have tried it, been excited by it, been disappointed by it, abandoned it and adopted it.
On one hand, we fell in love with getting a precise answer from a chat, and we have already changed usage patterns. Google search is no longer always the default, and who would have believed that would happen so quickly.
On the other hand, we have developed a very high level of skepticism toward the exaggerated hype around every new product and every new feature. We have already encountered many of the current limitations of these tools. We have seen where they help, where they confuse, where they invent things, and where they are simply not good enough to replace what already works.
That does not mean interest has gone down. It means the standard has gone up.
A 2026 study on AI fatigue in human-AI interaction describes AI fatigue as a combination of cognitive overload, emotional overload, behavioral disengagement and physical fatigue. In other words, it is not just that “we are tired of the hype.” When AI appears in more and more places, it also asks users for more attention, more decisions, more trust and more effort to understand when it is even worth using.
At the same time, Gartner published the 2026 Hype Cycle for GenAI, describing a field that continues to evolve at an unusually fast pace, but is still surrounded by hype that makes it difficult for leaders to decide which capabilities are actually worth investing in.
This feels much more aligned with the current stage: we are not past AI, and we are not indifferent to it. We are simply past the stage where saying “there is AI here” is enough to be impressive.
We can all see how the market is changing because of AI. One of the drivers of this change is that if anyone, including someone non-technical, can create interfaces that solve simple problems for themselves, then the mere ability to create an interface is no longer enough.
It is not necessarily that impressive anymore.
And without clear value, there is no reason to pay money for it.
Where AI adds value, makes work more efficient, solves real problems or enables something that was hard to do before, that is where things get interesting. But in places where it is just another layer on top of a simple action, another button next to an existing choice, another chat asking me to ask something when I actually came to read reviews from real people, the value is not clear.
Beyond the hype that pushes companies to add AI everywhere, because no one wants to say they have not implemented AI, it is worth pausing and asking a few simple questions.
Am I solving a complex problem here, or can the user solve it themselves without much effort?
Does the user actually want to ask AI, or do they want to see human responses, a rating, a review, a history or another reliable source?
Is the right solution actually a pre-AI solution, such as encouraging users to leave more high-quality reviews instead of adding a generated answer?
Does the AI fit into the existing usage pattern, or does it require the user to stop what they are doing and learn a new way that is not clearly better?
And is it good enough to replace the current way the user solves the problem?
The race toward artificial intelligence has led companies to push it almost everywhere, add a lot of features and show that we are in the new era. Fine.
But if it is not convenient enough to fit into existing usage patterns, or good enough to replace the current way of working inside the product, who needs it?
What does it give us besides checking the box that we added AI?
You can look at AI as a process accelerator, both in the good sense and in the less comfortable sense. It accelerates the run toward a deliverable and the checking-off of features, but it also accelerates the exposure of things nobody really cares about.
Because if a feature does not solve a real problem, AI will not necessarily save it. Sometimes it will only make that clear faster.
The smart thing is not to add AI. The smart thing is to figure out how to really add AI in a way that solves a problem, fits into existing behavior, builds trust and gives the user something they actually wanted or needed.
And that requires more expertise and deeper understanding, not less.