The Next Front Door to a Brand Is a Person Who Doesn’t Exist

What happens when a brand stops telling everyone the same story and starts remembering yours?

Imagine you miss a week of training. When you open your running app again, a character you’ve come to know remembers the 10K you wanted to run and the knee that made you nervous about starting. It doesn’t congratulate you for “showing up”. It doesn’t offer you 20% off some shoes. It asks if twenty minutes would feel possible today.

That could be useful.

It could also be the most persuasive advert you’ve ever met.

The character isn’t a person. It belongs to the sports brand behind the app. But you know how it speaks. It remembers what you’ve told it. You’ve followed its story while it has followed a little of yours.

Brands have always tried to get our attention. I’m interested in what happens when they can hold a conversation—and pick it up again next week.

One character. Millions of stories.

Social media made a particular relationship enormous: one creator tells one story to millions of followers. Followers can respond, but the creator cannot know each of them. Much of the relationship happens in the follower’s head.

An AI character could have one recognisable personality and backstory, while sharing a different ongoing experience with every person who talks to it. Its own story might develop in public. Its conversations with you could continue in private.

That is a bigger idea than putting your name in an email or choosing which advert you see.

Personalisation may stop meaning personalised advertising and start meaning personalised storytelling.

The visual form is almost beside the point. It might look human, animated or unlike anything we have a name for yet. What matters is whether it has a character worth spending time with, and whether it can respond to what happens in your life without losing its own identity.

There are early pieces of this already. Duolingo describes how Lily, its conversational language-learning character, has a defined personality and backstory, and how information from one call can inform the next. Tavus offers persistent memory for conversational characters. Neither example proves that people want a relationship with a brand. They show that character and continuity are becoming practical design choices.

You can’t prompt a personality into existence

This is the part I find most interesting as a creative director.

A brand character isn’t made by choosing a realistic face and writing “cheeky, funny and into fitness” in a brief. That gives you a costume. It doesn’t tell you how the character should behave when someone misses a goal, gets frustrated or asks a question the brand would rather avoid.

A sports brand could create a relentless fitness obsessive, a patient coach or a mischievous training partner. Each might be entertaining. Each would build a different relationship. A joke about a missed run could make one person laugh and make another close the app.

So the first question is not what does the avatar look like? It is who is this brand asking to be in someone’s life?

What does the character believe? What is its backstory? What can it be funny about, and what must it take seriously? When does it push, when does it back off, and when does it say nothing? What should it remember? How does it admit it was wrong? Those decisions have to come from the brand’s point of view and from an understanding of the people it hopes will return.

Its tone can adapt without its principles changing. A good coach speaks differently to someone celebrating a personal best and someone struggling to begin. That doesn’t make the coach two different people. If the character changes its beliefs to please whoever is in front of it, there is no character left to know or trust.

In an advert, you can edit the line that didn’t land. In an ongoing conversation, the brand reveals itself through hundreds of small decisions. The personality isn’t the voice you give the avatar. It’s the pattern of behaviour people come to recognise.

Would you come back if it had nothing to sell?

That is the test I’d put to any brand proposing one of these characters.

Imagine the sports character following your progress, helping you adjust a goal, or making the dull middle weeks of training more interesting. Perhaps it has its own unfolding story that you enjoy returning to. A product recommendation might occasionally make sense within that experience. But if every conversation bends towards a sale, you haven’t made a character. You’ve given a sales funnel a face.

The most interesting version crosses boundaries we currently keep separate. You might first encounter the character through a story. Later it helps you choose something. After you buy, it helps you use it. The same intelligence could eventually live in a product, car or service rather than disappearing when the campaign ends.

A website would still matter. People need somewhere to check facts, compare terms and get things done. But the first place they turn to might be a conversation.

Brand positioning used to be something you communicated. AI may turn it into something people interact with.

If a brand says it is encouraging, the character has to be encouraging when you’ve failed at something. If it claims to understand you, it has to know when to stop talking. A beautifully written manifesto won’t rescue a character that behaves badly on a difficult Tuesday.

This is not just a new type of influencer, either. An influencer brings a personality a brand can borrow for a while. This character could become part of the brand’s service and its products. Writers, designers, service teams and product teams would all have a hand in it. Nobody could simply hand it back to the campaign department when the launch was over.

Would it tell you to buy from someone else?

I think it has to be able to.

If another brand makes a better product for your needs, a trustworthy character should be able to say so—or at least explain the difference honestly. It should also be able to tell you that you don’t need to buy anything. People already ask general AI assistants to compare their options. Why would they trust a brand character that suddenly loses its judgment whenever a rival is mentioned?

That is difficult for brands. They have spent years controlling what their representatives say. A character designed to agree with every brand claim might stay on message, but people would soon learn that its advice has limits. The relationship would have limits too.

Honest advice wouldn’t mean pretending the character is independent. It would still belong to the brand, and people should know that. But a brand strong enough to acknowledge when a rival is better could earn something more valuable than one sale: a reason to be believed next time.

That principle has to shape how the character works, not sit in a statement about trust. If the only measure of success is how often it converts a conversation into a purchase, it will be pushed to sell. Someone inside the business must be willing to defend the occasions when it gives good advice and the customer leaves without buying.

The friend-shaped problem

I can see why this idea makes people uncomfortable. It should.

A synthetic character could remember what you want, what you fear and when you’re most open to encouragement. It could entertain you, help you and earn your trust. It would still be working for a company with commercial interests.

Social media made one-sided relationships with public figures scalable. AI could make those relationships responsive. But responsive is not the same as reciprocal. The character can ask how your knee feels. It does not care about your knee in the way a friend does.

People should be able to see what it remembers, decide what it may keep and tell it to forget. They should know when a recommendation is commercially motivated. The US Federal Trade Commission has opened an inquiry into how AI companions use personal information and make money. That inquiry is not a verdict on brand characters, but the questions belong in the design from the start.

The advertising industry has spent a century learning how to persuade people. AI may force brands to learn how to advise them.

The next front door to a brand may not be a website. It may be someone.

The hard part won’t be making that someone believable. It will be building a character the brand is brave enough to let tell the truth.