First Your Teenager Talks Back. Now Your Ads Will.

For most of my career, digital advertising has worked on roughly the same principle.

We interrupt you.

Hopefully at the right moment, with something vaguely relevant, and ideally with enough targeting data that the interruption doesn't feel completely random.

Then we ask you to click.

That model has become incredibly sophisticated. But fundamentally, it's still a billboard with better data.

This week, something happened that I think points towards a much bigger change.

OpenAI announced it is testing Sponsored Agents inside ChatGPT. Instead of clicking an advert and being sent somewhere else, you can start a conversation with an AI agent sponsored by the business itself. (OpenAI)

Think about that for a second.

The advert can now talk back.

The click might be becoming obsolete

Imagine I'm looking for a new road bike.

The old internet gives me an advert.

The next internet might give me Trek.

Not the website. Not a banner. An AI representative of Trek that I can interrogate.

Which bike suits Geneva? What's the difference between these two models? Will this frame fit me? Why should I buy yours rather than a Specialized? Can you find one nearby?

Suddenly advertising isn't just persuasion.

It's a conversation.

And conversations require something advertising has historically been rather good at avoiding.

Answers.

Check it. 👇

The interesting question isn't whether AI makes advertising more personalised.

We've been doing personalisation for years.

It's what happens when the distance between claim and challenge disappears.

Your advertising may have to defend itself

Traditional advertising gives brands enormous control.

You choose the image. The claim. The 30 seconds. The landing page.

A conversational advert is different.

The customer chooses the next question.

And then the next one.

That changes the creative problem completely.

The best advert may no longer be the one with the cleverest headline. It may be the brand with the deepest product knowledge, clearest evidence and most useful answers.

That's particularly interesting because consumers appear surprisingly open to AI when it actually helps them.

Canva's 2026 research across seven countries found 68% of consumers don't mind AI in advertising if it makes ads more helpful or relevant. But 70% said AI-generated advertising often feels like something is missing. (Canva)

That tension makes perfect sense to me.

People don't necessarily hate AI.

They hate crap.

AI may turn advertising from a contest over who can make the most persuasive claim into a contest over who can give the most useful answer.

And that could have consequences well beyond media.

Because if your AI representative can't explain why your product costs more, substantiate your sustainability claim, understand the customer or answer an uncomfortable question, that isn't really an advertising problem.

It's an organisational one.

For years we've built increasingly sophisticated systems for getting people to click.

Maybe the next competitive advantage is what happens when they start asking questions instead.

That sounds like a much more interesting internet to me.