Thoughts after an AI-free summer

I had an AI-free summer and I could not let go of one thought: we are building a future on the assumption of the inevitability of AI. But will it really happen?

I must be honest: I did not use AI at all this summer, and I did not miss it. For sure there is a beauty in the technology, some help, but if it was taken away tomorrow, I would not scream in disappointment.

And yet, we keep on pushing, and racing toward a future where AI will inevitably permeate everything, where the super intelligence of tech-determinism will win, and there is nothing else to do than accept it.

Even when accepting means sacrificing a lot: from consent of data and intellectual property being taken, to data centers built in your backyard, to a really unclear benefit (for everyone but the handful of people pushing for that inevitable future).

And as a consequence we are seeing the revolts, because for the majority of people the pains are clear, but the gains are not.

What is clear in front of everyone are the mass layoffs, the push for getting more done with less, the constant pressure of up-skilling, or re-skilling while the rules of the game change at a breakneck pace, the data centers consuming drinkable waters, the intellectual property taken with no consent to train the models.

What we do not see yet is: what's in it for us? What is the problem that we are so desperately trying to solve with AI?

The more I think about it, the more I start to see this as a product problem: we keep on trying to find a better fit for the technology, and start from the solution, not the problem. And the solution itself keeps on evolving.

A clear example for me is the experiment that we've been running at Vend. Early in the year we ran a beta for a conversational search, betting that the same interface people love in LLM chatbots would be "the new way of searching". There were talks of "tech armageddons" in marketplaces, the word disruption was pronounced more times than I could count, and instead of waiting, we decided to lean into the tech, experiment, and learn.

And indeed we learned a lot, including that the habits people have built over (many) years don't shift overnight, neither they do for a solution that has potential, but only limited immediate improvement. And we learned that the feeling of control is as much important as the habit, and when someone is making an emotional decision like buying a house, a chat window doesn't automatically feel like the better way to search.

Two things stood out.

• First, a use case working brilliantly in one context doesn't mean it transfers, especially when the decision on the other side is high-stake and emotional rather than transactional. In other words, user behavior and expectations are not universally applicable for search.

• Second, the AI experiences that actually gained traction in our product are the ones where the technology disappeared into the background rather than being front and center. Our insight from a combination of usage and surveys is that nobody wants to feel like they are using AI, yet everyone want the benefit of it, seamlessly.

That's the pattern I keep running into: lead with the problem, show the gain, and let the technology stay invisible. The moment AI becomes the headline instead of the outcome, eyebrows raise and adoption stalls.

And beyond the product I work with, I see how this problem-solution fit is a a moving target.
The fact that the answer to "what will AI help with" can stretch from boosting productivity to curing cancer, depending on who's doing the pitching (or to whom), is a sign that nobody has actually landed on the problem. And a technology chasing a moving target will always struggle to prove itself.

And it is precisely from the lack of a clear problem-solution fit that we see the revolts, almost the repulse against AI. We say we do not want it, the chatbots deployed in almost all the products annoy the most, the AI slop of content is flooding everything content-related, and whenever we see a 6 page ai-slop document produced with absolute no care, the immediate response is: thanks, but no thanks.

At the same time we use it, and a looooooot of money (maybe to the point of no return) has been pushed into the industry.

I see the challenges, I do not see a neat resolution here. Starting from myself: I'm curious about the technology, skeptical of the hype, occasionally amazed, and often uneasy, sometimes all in the same week.

And the framing I keep coming back to is simple: the products and the visions that ask for our trust without solving a real problem are the ones that will not last. Trust isn't given by default, and neither is adoption. Both have to be earned, one useful, invisible, honestly-argued-for improvement at a time.

---

Extra listening:

For the ones interested in the revolts about AI, and on why the questions will really soon be political, especially in the US, I recommend this podcast from Ezra Klein: https://www.nytimes.com/2026/08/04/opinion/ezra-klein-podcast-jasmine-sun.html