Getting found in AI search terms

People are moving from Google over to AI. They open ChatGPT, Claude or Gemini and ask the thing they would once have typed into a search bar, and the tool hands back a short answer with two or three products named in it. Being mentioned in those places is very soon, I think, going to be more powerful than being found in Google. That is a big statement, but that is where the attention is going.

The difference from classic SEO is the shape of the result. A Google results page gives ten links and lets the person choose. An AI answer gives one paragraph, and whoever is named in that paragraph gets the click, or the trust, or both. If your product is not in that paragraph, you are not in the race for that question at all, no matter how good your page is.

The second difference is that the questions are longer and more human. Nobody asks an AI "build in public app". They ask something like "I want to document my creative process like a show, but don't know where to start without being a video editor. What should I use?" That is a whole sentence with a situation in it, and the answer it produces names whatever the model has read that fits that situation.

So the job breaks into two halves: find out which buyer questions are being asked and who gets named for each, then put something on the web that answers that exact question in your own words so there is something to name.

You can do all of this manually. Open ChatGPT, type a question a buyer would ask, read who it names. Do it again in Claude. Do it again in Gemini with web search on. Write down the answers. Do that for thirty questions, then do it again later to see whether anything changed. It is possible, and it is also the kind of job that quietly stops after one round.

The second problem is knowing which questions to ask in the first place. The useful ones are not the ones you would guess; they are the ones your buyers phrase in their own words about their own situation. Getting those right means knowing who your buyer is, what they are trying to do and what they call it.

The third problem is that knowing you are not named does not tell you what to do about it. A report that says "a competitor was named here" is a complaint. What you need is the question written out, what each tool actually said, and a clear next action that ends with something published under that question.

Click Promote at the top, then Found in AI answers, and it opens your answers. It might take a minute to run. At the top it says what it does: the questions your buyers ask an AI, put to ChatGPT, Claude and Gemini with their web search on, and whether the answers name you, cite your page or name a competitor instead. Your interviews never see it, so none of this leaks into the questions you get asked on camera.

When I opened it I was named in 1 of 17 answers. Straight away, great news. I am already being featured. Then it goes further: it gives me how I can get mentioned in all of the others. That is literally free traffic, being mentioned in these kinds of things.

The one I am named for is "is there a tool that interviews me by voice and turns my answers into a finished podcast episode?" ChatGPT names us and cites our page. It shows what each tool said, side by side, so you can read the actual wording. Claude does not name us or anyone we compete with, and says yes, a few tools are built specifically for this. Seeing the difference between the three tools on the same question is the useful part: one knows you, one does not know anyone, one knows your competitor.

Underneath that is the list that matters: nine searches where an AI names someone else. That is what we want to be found in.

Reading the nine is the uncomfortable bit and the useful bit. One of them is "I want to document my creative process like a show, but don't know where to start without being a video editor. What should I use?" That is literally what I need to come up for. Another is "what's a good alternative to paying a freelance editor to cut together founder content from raw footage?" It is just on the money.

These are not keywords, they are sentences a real person would type when they are stuck. Each one comes with what each tool said, so you can see who got named instead and in what terms. There is no point arguing with it. The answer is to go and make the thing that deserves to be named for that question.

There are so many in the list and the instructions are really simple. I can tick them off as I go: film this, film this, film this, put them out, document that, and they are out there.

For each question that names someone else, the report tells me what to do. Film your answer: say the question out loud, then answer it plainly in your own words. It sets the filming up with the title and the points to cover, which makes it so easy. No script to write, no research to do first, because the question and the points are already there.

Then put the answer on your own site, which Underway will do for you, under a heading that asks the question. If the search is "I want to document my creative process but don't know where to start without being a video editor", that exact sentence becomes the heading on the page, and underneath it is your answer in your own words.

Underway edits it all together: an intro, an outro, cut so it answers that question specifically. Then Post to all: your YouTube, your directory page, your Twitter. One answer becomes a video, a page and posts in several places, all pointing at the same question.

That is the whole loop. Question in, episode out, page published, posted everywhere. You start ranking and bringing in free traffic. The work on your side is talking for a few minutes about something you already know the answer to, because it is your own product and your own buyer.

Filming one answer is minutes, not hours. You say the question aloud, answer it plainly, and the episode comes back finished with captions written. There is no timeline and no retakes. The realistic cost is the decision to sit down and do it, repeated across a list of questions.

Ranking is the slow part, and it is worth being honest about that. When the marketing brief looked at search, the site had 325 impressions, 6 clicks and no page-one searches, with almost all traffic coming from paid social. Search is a slow one. Articles take months, and you do not get to rush it.

That is why the AI answers report is a better scoreboard than Google positions early on. The named or not named count moves on its own schedule, question by question, and you can see which of the three tools shifted. The brief's plan for the week is simple enough: list the searches where AI names a competitor instead of you, film three of those answers, publish each on its matching page, and check the report again at the end of the week.

It also means volume matters less than specificity. One page that answers one long buyer question exactly is worth more here than ten pages about a category.

It suits people who are building something now and already know the answers to their buyers' questions, but would never sit down and write thirty articles. If you can talk for three minutes about why someone would use your thing instead of hiring an editor, you have everything the method needs.

It suits people documenting work outside software too: a shop, a studio, a band. The questions come out of your own world, written from your own one-line description and your site, so the searches are the ones your buyers would actually type.

It does not suit someone who does not want to speak. If you would rather write, write; a written journal will suit you better. It also does not suit anyone expecting a fast result. Being named in AI answers follows from having published a real answer in the first place, and publishing is on you.

And it is worth saying what the report is not. It does not post for you, it does not write in anyone else's voice, and it does not feed into your interviews. It tells you what was asked, who was named, and what to film next.