Your customer isn't searching for you anymore

Say there's a guy called Rohit. He lives in Shimla.

His old car has finally given up and he needs a new one. Five years ago he'd have typed "best SUV under 15 lakh" into Google, opened twelve tabs and watched three YouTube reviews at 1.5x.

This time he opens ChatGPT and types:

"I live in the hills. I need a car that handles steep roads well and still gives decent mileage. What should I buy?"

Read that again. No brand. No model. Not one keyword a marketer would ever bid on.

He didn't search for a product. He described his life.

And somewhere, a car company's marketing team spent the whole quarter trying to rank for "best SUV in India".

The search bar is turning into a conversation

This shift didn't start yesterday. But it's picking up speed.

People are now asking ChatGPT and other AI tools the stuff they used to ask Google, and the stuff they used to ask their friends. And they ask it differently. A search box trains you to talk like a robot: "running shoes flat feet men". A chat window lets you talk like a person:

"My feet are completely flat and my knees hurt after a 5k. What should I be wearing?"

"I want a snack that won't leave me bloated."

"My neck has been killing me for a week. Is there a good doctor near me?"

Nobody types "healthy chips". They type their problem. Then the AI goes looking for whatever out there best matches that problem, and recommends it.

Which leaves every business with an uncomfortable question: do you actually know what your customer is typing?

The chips brand that never gets mentioned

Picture a small brand making genuinely good, light, roasted chips. Their website says "healthy chips", "guilt-free snacking", "100% natural". Fine.

Now someone asks ChatGPT for a snack that's filling but doesn't feel heavy. Or something that won't bloat them before a 4pm meeting.

If the brand has never written a single word about bloating, about feeling heavy, about who their chips are actually for, the AI has nothing that connects them to that question. Their product might be the perfect answer. It just never got introduced.

Same with a shoe brand that makes shoes for flat feet. If their site reads like every other shoe site, they're invisible to the exact person they built the shoe for.

This is what people call GEO, Generative Engine Optimization. Most of the talk around it is about content. Write more blogs, add FAQs, structure your pages so AI can read them. All true. But I think that's step two.

Step one is knowing what to write about. And that only comes from one place.

What I learned building a GEO agent

I built a GEO agent. It reads a business's publicly available data, finds the gaps in its online presence and suggests what blogs and content it should create to show up better in AI answers. It works, up to a point.

But building it taught me something I didn't expect.

The writing is the easy part. You can hand it to an AI today. Hundreds of blog posts? A weekend's work. Soon everyone will be able to do that, which means it stops being an edge.

The hard part is the input. What is your customer actually worried about? What frustrates them? Why did they really buy from you, not the reason in your pitch deck, the real one? What words do they use when they explain their problem to a friend at 11pm?

An agent can't know that on its own. It can guess, and that's exactly the problem. Feed it a shallow understanding of your customer and you get a thousand shallow blogs. Confident, nicely formatted, and about nothing your customer ever asked.

So here's the irony. The more AI writes our content, the more the advantage shifts to the one thing it can't do for you.

Knowing your customer.

Now add agents to the picture

It gets more interesting.

Right now a human asks ChatGPT and reads the answer. Soon a lot of people will have their own AI agents doing the searching for them. (I wrote more about that shift here.)

"Find me running shoes for flat feet, under 6,000, that don't need breaking in."

The agent goes off, reads, compares, maybe even buys. It never sees your billboard. It doesn't watch your Instagram reel. It cares about one thing: does this product match my owner's exact problem, and is there enough information out there to back that up?

You can pay to get in front of that agent. Ads are coming to AI platforms and that's a fair route. But paid distribution gets expensive fast. Showing up organically, because you genuinely are the best answer and you've said so clearly, is the much bigger win.

The small guy might finally get a shot

This is the part that excites me most.

For decades, distribution has belonged to whoever has the biggest budget. Big brand, big ad spend, big shelf space, big wins.

But an agent doesn't care how big you are. It cares how well you fit.

If you're a tiny company making shoes only for flat-footed runners, you've always lost to the giants on reach. In a world where agents match specific problems to specific solutions, you could win on relevance. The person with that exact problem might get sent straight to you.

Maybe the world splits into lots of small, sharp businesses, each solving one problem for one kind of person really, really well. I'd love to live in that world.

The job that just got more important

If knowing the customer is the edge, the people who do that for a living just became a lot more valuable.

UX researchers. The people who sit with customers, run interviews, dig through feedback and notice the gap between what people say and what they do. In the AI era I think their work moves from "nice to have" to the foundation everything else stands on.

And I can see where it goes next. A virtual UX researcher. An AI employee that talks to your customers all day, asks good questions, listens to how they describe their problems and keeps telling the company what people actually want. Companies might even start paying customers for honest, detailed feedback, because that feedback becomes the raw material for everything else.

Now picture the loop. One agent talks to customers and learns their language. It hands what it learned to another agent that writes the blogs, the FAQs, the product explanations. Every week the company understands its customer a little better, and every week it explains itself a little better.

That's a GEO engine. Not a content farm.

Back to Rohit

Somewhere in the hills, Rohit is still waiting for his answer.

The car that gets recommended to him won't necessarily be the one with the biggest ad budget. It's more likely to be the one whose maker understood people like Rohit (steep roads, cold mornings, a fuel bill that actually matters) and bothered to say it out loud, in plain words, somewhere the AI could find it.

Anyone can generate a thousand blogs now. That's not the moat.

The moat is knowing your fucking customer.