Why AI Can't Recommend What It Can't Find
For twenty years, being findable meant being on the first page. With an AI assistant, the first page is one name long. That's a much smaller place to stand.
By NoTrouble · March 15, 2026
Google gave them ten. ChatGPT gives them one
When someone asked Google for a plumber, Google gave them ten. They picked one. You had a chance to be one of the ten, and as long as you were on the first page, you were in the conversation.
When someone asks ChatGPT for a plumber, ChatGPT gives them a plumber. Not ten. Not a list. One name, sometimes two, with a reason.
That's a quieter change than it sounds. It changes what "being findable" means. With Google, you could be in the consideration set without being the answer. With an AI assistant, those are the same thing.
What AI actually looks at
Large language models don't browse like people do. When they answer a question about who to hire, they're pulling from indexed pages where the meaning is unambiguous, mostly by reading something called structured data.
Structured data is a small block of code hidden in the page that says, in machine-readable terms: this is a person, this is what they do, this is where they work, this is what other people say about them. Schema.org is the shared vocabulary for that block. It's what tells a model that a paragraph about "Sandra" is about a financial advisor in Halifax with 47 reviews and a 4.9-star average, and not about Sandra Bullock.
The interesting thing isn't the technology. It's that very few small-professional pages on the internet have any of it.
What your Linktree page tells an AI about you
If your current presence is Linktree, here is the entire profile an AI system can confidently build of you from your page.
You have a name. You have some links.
That's the file. It doesn't know what you do for work. It doesn't know where you are. It doesn't know your reviews exist, because they don't live on that page. It doesn't know your credentials, your services, or the kind of client you're a fit for. When someone in your city asks it "who's a good X near me," you aren't in the answer, because there's nothing for the model to be confident about.
Linktree isn't doing anything wrong here. It was built for a different job, before AI search was a job. Most of the link-in-bio category is in the same shape. The page works fine for human visitors who already know what they're clicking. It just doesn't say anything to a machine that's been asked to make a recommendation.
What being legible looks like
A profile that an AI can recommend tends to answer a handful of small questions clearly.
What kind of professional is this. What do they specifically offer. Where are they. What do other people say. What are their credentials. These are the same things a human prospect wants to know on a first visit, written in a way a machine can also read.
This is where the boring middle of NoTrouble shows up. When you set up a profile, you pick a category from one of around fifty professional types. You add your services. You connect Google so your reviews flow in. You attach a custom domain so the page belongs to you. None of that feels like SEO work, because none of it is, but underneath it, the page is being marked up with the structured data that lets a model parse what it's looking at.
You don't need to think about schema. You need to think about what you do for a living. The rest of it gets written for you.
Why this matters now, not later
AI search isn't every search yet. It will probably take a while before "ask the assistant" replaces "open Google," and waiting is a reasonable position to start from.
The argument against waiting is narrower than it sounds. Models are training on the internet that exists right now. The professionals who have legible profiles in 2026 are the ones who get cited in 2027. Recommendations, once they're encoded, tend to be sticky. The early answer to "who's a good X in my city" becomes the boring default answer, the way "who's a good plumber" used to be answered by whoever bought the back of the phone book.
SMBs with structured data on their sites are already showing up to 35% more often in local results, by Google's own published numbers. That's the visible version of the shift. The invisible version is happening in chat windows you'll never see.
AI can't recommend what it can't find. For the first time in the history of small business, the question isn't whether your customers can read about you. It's whether the machine answering their question can.
Your professional presence, ready in 5 minutes.
36 design combinations. 15 content types. Auto-syncing reviews. AI discoverability. Free to start.