What if AI could help us pause — not just persuade us to buy?

We used to walk to a store to buy something. Then came websites, then apps, then one-click checkout, then saved cards, then “Buy Again.”

And now, apparently, groceries can arrive at your doorstep before you've even decided you need them.

Progress? Absolutely.

Impulse control? Decreased!

In fact, the algorithm and system experience is exactly helping us there — “Related buys.” “Frequently bought together.” “You might also like.”

And then there's the wonderfully distracting 99-store universe, where you opened the app for toothpaste and somehow you're now comparing snacks, chargers and a kitchen organiser you didn't know existed five minutes ago. Somewhere in that spiral, the price tag quietly became the loudest thing on the screen — not the toothpaste, not the need, just the number next to it.

Cross-selling is doing wonders. Last week it was a phone case. This week, apparently, a humidifier — for a room that has never once felt dry.

And honestly, there's plenty to love about the convenience — a few taps, and the day moves on. Nobody's asking for the five-minute checkout form back — least of all the OTP that autofills before you've even finished reading it.

But such connected experiences as above does raise a question —

Have e-commerce or grocery quick apps become so good at removing friction that they've started removing it from the one part of the journey where a little might actually help?

The decision.

Buying has become incredibly easy

A product is recommended, a discount appears, a notification reminds us, something similar catches our eye — and our payment details are already saved.

And suddenly:

Order placed.

The experience has done exactly what it was designed to do — reduce effort, reduce hesitation, get us there faster, and occasionally make us feel oddly proud of grabbing tomatoes at ₹9, cheaper than the vendor downstairs would've charged for the same walk down. And that's not a bad thing.

Increasingly intelligent systems have been built to help us buy more efficiently.

But here's the question worth sitting with:

If AI can get remarkably good at predicting what someone might buy next, could it also get good at helping them pause before they buy it?

Not stopping anyone. Not judging anyone. Just offering a little context.

There used to be a moment before the moment

Before everything became one tap, buying something usually involved a tiny pause :

Do you already have this? Did you buy something similar recently? Should you compare it? Do you actually need it? Maybe you should think about it. Maybe you should ask someone.

Nothing revolutionary — just a tiny bit of human friction.

Somewhere in there, "let me sleep on it" became an actual sleep-aid ad, shown to you at 11:47 PM, while you were meant to be sleeping on something else entirely.

And then digital commerce did what great digital experiences are supposed to do — it removed unnecessary steps.

Except... were all those steps unnecessary?

Maybe the problem isn't that friction disappeared. Maybe the wrong kind of friction got removed.

Not all friction is created equal

There's bad friction — forms that ask for a detail you already gave two screens ago, five checkout screens for what should be two taps, passwords you forgot five minutes ago, an app that logs you out right after you'd finally decided.

Then there's decision friction — that tiny pause before committing to something. Should you? Do you need it? Is there a better option?

There used to be a simpler version of this rule — wait a few days before buying, just to see if the want was still there. Most of the time, it wasn't.

Maybe don't remove all of that.

And then there's what's best described as protective friction — not a wall, not a warning siren, just enough context to make the decision a little more informed.

And this is where AI could get interesting.

What if AI became borrowed memory?

Nobody's working memory was built to carry an entire purchase history while scrolling at 11:30 PM. What did you buy last week? Do you still have it? Did you even use it? Weren't you looking for something else? And wait... didn't you already own this?

There's a yoga mat in the cupboard that can confirm this happens more than once. There are, in fact, two more behind it — and possibly a third somewhere in the storage room, waiting for someone to put in effort to find it.

That's not always a willpower problem. Sometimes it's simply a bandwidth and effort problem.

So what if the same algorithm could become something much simpler than a super-intelligent shopping assistant?

Something like Borrowed memory…not second brain :)

The bit of context we forgot to bring with us... surfaced exactly when we need it.

It's 11:30 PM. You're ordering snacks — you've ordered similar ones three times this week.

The commerce experience might quite reasonably say:

“Buy again?”

Your personal AI could say:

“You've ordered similar items three times this week. Still want them?”

That's it. No “This is unhealthy.” No “You shouldn't be doing this.” Definitely no virtual parent confiscating the phone at 11 PM.

Just:

“Here's something you might want to consider.”

You still decide. AI simply remembers what you might have forgotten.

You're looking at another pair of sneakers. The store knows you like them.

But someone needs to make the connection between what's on your screen and what's already sitting on your rack — by looking through the lens of your history.

“You bought something quite similar recently. Want to compare before adding another?”

Maybe this really is the one. It's probably not. Somewhere in the back of the wardrobe is a pair still in the original tissue paper, bought with the exact same conviction.

But that's between you and your wardrobe.

The point isn't to stop the purchase. It's to bring the missing context into the moment.

And then there's electronics.

You're looking at the latest phone. Faster processor, better camera, exchange offer, EMI, limited-time price — and then... the price.

Did the price just jump significantly? The last three phones were each bought largely "for the camera." Between them, roughly four photos were taken before life returned to normal and the camera went back to being a camera.

If you're going to upgrade, shouldn't the question be more than:

“Is this phone better?”

Maybe it's:

“Is this upgrade worth what I'm paying for it?”

A personal AI could potentially say:

“Your current phone still meets most of your usage needs — the slowdown might just be the battery ageing. A replacement costs a fraction of a new device, and it could feel new again. Want to check that first?”

That's not telling you don't buy. It might actually save the buy altogether — more value for the same intent.

And suddenly the AI isn't a purchase brake. It's a decision enhancer.

The goal isn't to make buying harder

This is the distinction worth returning to: friction that doesn't need to exist should go, but perhaps not the friction that helps people remember, compare and reconsider.

So: bad friction? Remove it. Decision friction? Respect it. Protective friction? Design it carefully.

Because perhaps the goal isn't:

“Make buying harder.”

It's:

“Make forgetting harder.”

And here's the beautiful part: nobody should have to maintain any of this

Nobody really wants another app asking them to log what they bought, how much they spent, what they already own, when they last reordered something. There are enough unfinished lists already.

At some point, the tracker or history needs tracking.

The interesting opportunity is that much of this trail already exists — purchase history, usage patterns, previous behaviour, preferences.

With the right permissions, AI could potentially connect the relevant pieces — not necessarily collecting more information, but making better use of the context already available.

History → Pattern → Trigger → Context → Choice

And crucially, it doesn't need to interrupt constantly. It waits — a meaningful pattern appears, it surfaces one useful piece of context, then it gets out of the way.

That's probably a better experience than another notification badge screaming for attention.

The interesting future isn't AI telling us what to do

It's AI helping us notice what we're already doing. What if it remembers what we forgot? What if it connects things we never connected? What if it surfaces the right context... at the right moment?

Not “here's another thing you might like,” but occasionally “you bought something similar recently.” Or “you haven't used the previous one much.” Or “the slowdown might just be the battery.” Or simply — “want to take a second look?”

The system doesn't become the decision-maker. It becomes the context-provider. And perhaps that's a healthier relationship between personalisation and autonomy.

One final thought

Digital experiences have spent years getting faster, and AI is about to make them dramatically more personalised.

But personalisation doesn't have to mean:

“I know you well enough to sell to you.”

It could also mean:

“I know your context well enough to help you decide.”

That's the part worth watching — because if technology is getting better at understanding what we might want next, perhaps it can also get better at helping us understand whether we actually need it.

Comfort now. Cost later.

AI doesn't have to become better at selling us things. It could become better at protecting us from ourselves.

Maybe the future of good experience isn't making every decision faster.

Maybe it's knowing which decisions deserve a pause.

Part 1 of a 3-part series exploring AI, intent and the experiences we design around human behaviour.

Next → From Screen Time to Life Time.
What if your personal AI could understand where your intentions are going? And elaborate on that purposefully :)