I’m sitting in front of an insurance company’s website. Ten minutes. Twenty. I’m trying to work out whether the policy covers water damage too or only fire damage. The buttons are too small. The text is written in a courtroom font. I feel like a criminal trying to sneak through a system that doesn’t really want me to succeed.
And then I think: in two years, nobody will be doing this anymore.
Not me. Not you. Nobody.
Because in two years, your agent will be doing it for you.
So what, AI does everything?
No. That’s not it.
And this isn’t the claim that “AI will replace us” either. That claim is too old, and it’s also inaccurate.
The new claim is far more interesting:
The user will no longer be you.
Think about that for a moment. For twenty years we built the entire digital world around one simple assumption: a person sits in front of a screen, reads, clicks, fills in, compares and decides.
Every UX tool, every conversion funnel, every A/B test — they’re all built around the human eye. Curiosity. Fatigue. The urge to buy. Fear of missing out.
But what happens when the party making the decisions is no longer a person?
From UX to AX
In 2024 everyone was talking about artificial intelligence.
“It changes everything,” they said.
And we, of course, thought that meant we’d use AI to write posts and generate images.
Cute.
But that was only the beginning.
The real revolution isn’t happening in chatbots.
It’s happening quietly, on the other side of the screen.
I’m talking about digital agents — entities that represent you online. They aren’t “smart assistants” waiting for a command. They act independently. They browse, compare, verify information, negotiate and execute transactions.
Without you seeing a screen.
Without you clicking a button.
And here’s the point almost nobody understands yet:
When the agent is the user, all the rules of the game change.
A site built for human beings needs beautiful images, a compelling headline and a red button.
A site built for an agent needs a clear data structure, an open API, prices that can be read without parsing HTML, and a refund policy defined logically rather than legally.
This is no longer a question of prettier design.
It’s a question of a new language.
We are moving from UX (User Experience) to AX (Agent Experience).
And this isn’t an upgrade.
It’s a revolution.
A site without eyes
Imagine for a moment that you want to buy a new car.
Today you visit five websites, compare prices, read reviews, call a dealer, fill in a form, upload your ID and sign a contract.
Hours.
Days.
A headache.
Tomorrow you say to your agent:
“Find me a family SUV for up to 150 thousand shekels, with full insurance and a garage in Tel Aviv.”
And the agent sets off.
It doesn’t “visit” websites.
It doesn’t “see” screens.
It receives data directly — prices, warranties, availability and reviews — through open information layers.
It compares a thousand options within seconds.
It negotiates with the garage’s agent.
It checks the warranty terms in the code, not in a legal document.
And it signs on your behalf — because you already gave it permission to do so.
At the end it reports:
“Found it. This car, at this price. The terms are 12% better than the market average. Approve?”
You click “yes”.
Or not even that.
Because you already specified that any transaction up to 160 thousand shekels can be executed automatically.
Where was the screen in this whole process?
Nowhere.
This isn’t science fiction. It’s already happening.
I know what you’re thinking.
“This sounds like a distant future.”
But it isn’t.
It’s already happening — it’s just not consolidated yet.
Today there are already tools that search for flights for you, compare prices and book automatically when the price drops.
Today there are already AI agents that manage investment portfolios, buying and selling shares according to rules you defined.
Today there are already bots that fill in tax forms, check reconciliations in bank accounts and identify duplicate charges.
The difference between “already today” and “tomorrow” is only a question of how many of your digital actions are carried out by an agent — and how many are still carried out by you.
And the answer?
That number is growing.
Fast.
So what’s left for human beings?
That question troubles a lot of people.
I understand why.
But I think it’s simply the wrong question.
The right question is:
What do human beings actually want to do?
Because let’s be honest.
Most of what we do online today consists of tasks nobody really wants to perform.
Nobody wakes up in the morning and says:
“Wow, what fun. Today I’m going to fill in an insurance form.”
Nobody dreams of comparing air conditioner prices for two hours.
Nobody misses the bank’s hold music.
Agents aren’t taking away something we love.
They’re taking away something we hated anyway.
And what’s left?
Decisions.
Values.
Choices.
Should I buy this car at all?
Is this the right investment for my family?
Does this company align with what I believe in?
The human being moves from task executor to task definer.
And that, if you ask me, is a tremendous upgrade.
The problem nobody has solved yet
But let’s not be naive.
There is a real problem here.
And I see it every day.
If the agent is the user — who builds it?
Who defines its priorities?
Who makes sure it isn’t influenced by covert advertising?
And who guarantees that it’s really working for you, and not for whoever developed it?
This is a question of trust.
And it’s far more complex than any UX question we’ve known until today.
Because when you scroll a website, you can sense that something is off.
But when the agent scrolls for you — you no longer see anything.
You simply trust.
And that’s frightening.
I think the great challenge of the coming years won’t be building smarter agents.
The challenge will be building the transparency that allows human beings to trust the decisions agents make on their behalf.
That’s an entirely new layer of design.
Not screen design.
Trust design.
The third generation
Let’s talk about history for a moment.
The first generation of digital connected people to computers.
You sat in front of a screen, typed and received an answer.
The second generation connected computers to people.
The internet, apps and social networks started reaching out to you — with notifications, recommendations and reminders.
The third generation — and it has already begun — connects computers to computers.
An agent talks to an agent.
An API talks to an API.
A transaction is executed between two digital entities, with human beings at the edge of the process — if at all.
And that changes everything.
Because when the flow takes place between machines, the human no longer needs to understand the language.
They only need to trust.
So what are we building now?
I’ve been working in this world for years.
And there’s something I relearn almost every day:
The best technology is the one you don’t feel.
Not because it’s hiding something.
But because it’s so natural that you no longer think of it as technology.
The next generation of digital systems won’t be another screen you have to understand.
It will be a decision you have to make.
Not a form you have to fill in.
A result you have to approve.
Not a user interface.
An intent interface.
We are moving from a world of “how many buttons are there on the screen” to a world of “how many decisions do you actually need to make”.
And that, in my view, is the greatest hope of the digital world.
No more systems that add screens.
Systems that erase them.
The end (or a real beginning)
I go back to the insurance website.
Twenty minutes have passed.
And I still don’t know whether water damage is covered.
And at that moment I understand:
It’s not that I’m stupid.
It’s the system that wasn’t built for me.
It was built so that I’d spend time in it.
So that I’d get confused.
So that I’d call.
So that I’d pay more.
The next generation of digital won’t allow that.
Not because there will be a law forbidding it.
But because no agent will agree to be confused for twenty minutes on a frustrating website.
It will simply move on.
And that is exactly what needs to happen.
Because in the end, digital was never meant to be a maze.
It was meant to be a bridge.
And the best bridge is the one you don’t feel yourself crossing.
So the next time someone says:
“We need to improve the site’s UX.”
Ask them:
“For whom exactly?”
Because if the answer is “for human beings” — it may already be too late.
And what do you think?
In five years, will most of us even visit websites, or will agents already be doing it for us?
Related guides
- 88+ AI tools worth knowing
- Implementing artificial intelligence in your business: the complete guide
- Information systems requirements analysis: the complete guide for businesses 2026
- UX/UI user interface design: the complete guide
- Loss of sovereignty: why the modern organization no longer controls itself