The smartest systems will not simply anticipate. They will know when to hand agency back to the human. That may become one of the most important design principles of the AI era.

For the longest time, the promise of digital was simple: remove the friction.

One click instead of three
A payment in seconds instead of minutes
A recommendation instead of a search
A message instead of a phone call

We built an extraordinary digital economy around making life easier. But I wonder if we have reached an interesting inflection point. What if the next big challenge isn't removing friction, but knowing which friction is worth keeping?

Not all friction is bad. Sometimes friction facilitates judgement. And as technology becomes capable of doing more and more on our behalf, judgement may become the most valuable human capability we need to design for.

We optimised the click. Now we need to optimise the choice.

Digital advertising is a good example. For years, our ambition was to reduce the distance between seeing something and buying it: discover, click, convert. But the abundance of content, commerce and personalisation has created a new problem: choice overload. Consumers don't necessarily need another 50 recommendations. They need a reason to choose one.

Spotify’s Wrapped is interesting because it doesn't simply give users more content. It turns an enormous amount of behavioural data into something people can understand, enjoy and share: a story about themselves.

The intelligence isn't just in the data. It’s in what the brand chooses to do with it. That distinction will matter enormously in the age of AI. Because when everyone can generate more, the scarce resource will no longer be content. It will be curation.

AI changes the question from “Can we?” to “What should we?”

We have become fascinated by what AI can produce. A campaign in minutes. A film without a shoot. A strategy from a prompt. A thousand creative variations. All impressive. But if everyone has access to the same capability, capability itself stops being the advantage. The competitive advantage shifts to judgement.

Nike, for instance, has used data and technology to make digital experiences more individually relevant. The larger lesson isn’t simply “use AI to personalise”. It is that technology works best when it strengthens a distinctive understanding of the consumer rather than replacing the brand's point of view.

This is the uncomfortable question businesses need to ask: If AI can make almost everything possible, who decides what is worth making? That's a leadership question, not a technology question.

Fintech has already discovered that friction isn’t always the enemy

Payments have been built around making transactions almost invisible. Scan. Pay. Done. But there are moments when making something slightly harder can prevent something much worse. A warning before an unusual transaction. Additional authentication. A prompt asking you to reconsider a suspicious action.

Google Pay and PhonePe, among others, have invested in fraud detection and consumer warnings because the fastest transaction isn't necessarily the best transaction when the consumer may be at risk.

This points to a new design principle: Good design isn't the absence of friction. It is the intelligent placement of friction. The best products will know when to say, Go ahead. And when to say, Wait. Are you sure?

Sometimes an interruption is the experience

Cybersecurity makes this even clearer. For years, we designed technology to be invisible. The best security was supposed to be security the user never noticed. But AI-generated impersonation, deepfakes, social engineering and increasingly sophisticated scams are changing that equation. Sometimes the responsible experience needs to interrupt us: the banking notification asking whether we authorised a payment; the platform asking us to verify our identity; the warning that stops us from clicking.

These moments feel like friction. But they are actually digital guardrails. I think this presents a fascinating opportunity for brands: to stop treating every interruption as a failure of UX. Sometimes an interruption is the experience.

Personalisation should also know when to step back

There was a time when personalisation meant: We know you! Your name. Your purchase history. Your location. Your preferences.

AI takes this much further. Soon, the question won't simply be whether a brand can predict what you want. It will be whether the brand knows when not to predict for you. Because personalisation can become presumptuous. Automation can become overreach. Convenience can become dependency. The smartest systems will not simply anticipate. They will know when to hand agency back to the human. That may become one of the most important design principles of the AI era.

The new leadership equation

Digital has trained a generation of leaders to value speed. Move fast. Test. Learn. Scale. Repeat. I've benefited enormously from that mindset myself. But speed is useful only when you know where you’re going. The leadership challenge ahead isn’t simply moving faster than the market. It is knowing what deserves to move fast and what deserves to be thought through. I increasingly think of leadership as managing three things simultaneously: Speed: What can we accelerate? Judgement: What should we question? Agency: What should we leave to people?

That framework applies equally to a marketer launching an AI campaign, a fintech company designing a payment journey, a cybersecurity team building fraud protection, or a CEO deciding where automation belongs.

The next digital revolution may be about restraint.

This sounds counterintuitive. After decades of celebrating disruption, innovation and frictionless experiences, our next competitive advantage may come from knowing when not to optimise.

When not to automate.
When not to personalise.
When not to add another feature.
When not to remove another step.
When to let a human decide.

Technology is rapidly making the question What can we do? almost irrelevant. There will be very little we cannot do. The more important question is: What should we do and what should we leave beautifully, deliberately human? That, to me, is where the next chapter of digital leadership begins. Not in making technology disappear, but in making sure that technology knows when to step back.

The author is Global Strategy Director, Tonic Worldwide and Head, GIPSI (HI+AI research division of Tonic Worldwide). Views expressed are personal.