Opinions expressed by Entrepreneur contributors are their own.
Key Takeaways
- AI rankings are temporary; the framework machines use to judge vendors can last.
- You do not need to invent a category to lead one; define the problem, establish the standard buyers should use and explain how success should be measured before competitors get to write the rules.
In April, another agency launched a “Machine Relations” practice.
It announced the service on a national newswire, published its pricing and called Machine Relations “the emerging discipline of AI search visibility.” It didn’t mention where the name came from. I coined the term in 2024.
Plenty of founders would have spent that week fighting over credit or losing their minds. I saw it as the moment I had been building toward. A category doesn’t exist because the person who named it says so. It exists when someone else starts using the word as if it had always been there.
Today, other firms publish their own guides to the discipline, and some are doing their best to claim it. But ask AI search engines who coined it, and they still point back to me nearly every time. That’s because the sources they lean on all tell the same story: the official definition, the wire coverage, the public profiles and the research built on top of them.
I’m not telling you this to take a victory lap. I’m telling you because it points to the one part of AI visibility almost nobody is competing for, and most aren’t even aware of.
Most companies are fighting to show up in AI search. They should be. Getting recommended when a buyer asks an AI engine for the best option in their category is worth real money. But a spot on that list is rented, and the rent resets daily. You read that right. Daily.
Paralax, an AI search research publication I oversee, tracked the same questions across five engines every day for 89 days. On an average day in September, Gemini dropped 42.8% of the websites it had cited the day before. ChatGPT dropped 35.9%. That means you could win a spot on Monday and lose it by Tuesday.
But here’s the kicker. Not every question is treated the same. The Machine Relations Index, which my team also runs, tracks what six major AI engines cite. Within a single category, asking whether something is worth it instead of asking for the best tools made those engines more than four times as likely to cite Wikipedia—the internet’s default explainer. Same subject, same engines, same window.
Ask a machine for the best vendor, and it hands you names. Ask it whether something is worth it or how to choose, and it reaches for whoever explains the category most clearly and, more importantly, most consistently. That’s the part almost nobody understands: the company whose explanation it borrows has already framed the decision before the shortlist even forms.
None of this is new. The economics of categories have always rewarded that position. A widely cited analysis by category-design firm Play Bigger found that among venture-backed tech companies, category leaders captured 76% of their category’s market value. AI doesn’t change that math. It compounds it, because the one explaining your category to your buyers is now a machine.
This is where Machine Relations gets strategic. I define it as the discipline of earning AI citations and recommendations for a brand by making that brand legible, retrievable and credible inside AI-driven discovery. Most people hear that and think about rankings. The bigger prize is this: a buyer asks ChatGPT how to choose in your category and gets your criteria, in your words, before a single vendor is named. The result? Your competitors get judged by your yardstick.
You may be thinking you can’t invent a new word for what you sell. You don’t need to. Every market already has contested ground: what the problem is called, what separates a good solution from a risky one and how the outcome should be measured.
So whether you’re a founder, a marketer or the person running SEO and AI visibility, here’s how I’d go after it. It’s the same playbook I ran.
Claim the criteria before you claim a name
Most markets don’t need a new category. They need someone to state plainly what a buyer should look for. A cybersecurity vendor doesn’t need to coin anything. It needs to be the source that tells the machine which three questions a buyer should ask before trusting anyone with their data.
Start with three sentences: the problem, the standard and the measure. If your customers keep describing the same frustration in different words, the first sentence is already waiting for you.
I only coined a name because the shift handed me one. Years of living and breathing earned media at my company, AuthorityTech, showed me that the first reader of a placement was no longer a person. It was a machine absorbing the placement to answer humans. Public relations was quite literally becoming Machine Relations in real time.
Publish one definition and never let it drift
Machines reward consistency. If your website defines the idea one way, your executives describe it another and your press materials use a third phrasing, you’ve handed the machine three weak signals instead of one that compounds.
Write the definition once, date it, give it a permanent home and use the same words everywhere: your site, your bylines, your interviews, your public profiles. It’s boring, but it works. Simple scales.
This column is a live example. The definition I gave earlier matches my dated public record and the official glossary word for word. That’s the point.
Get it repeated where you don’t hold the pen
Your own site can define a term. It can’t make the market adopt it. A definition becomes the category’s when it shows up in coverage, interviews and analysis written by people with no reason to promote you. That’s the job earned media has always done for reputation, and now it does the same job for the machines.
Pitch the shift, not your company. Editors will cover a change their readers are living through long before they’ll cover your product. Every credible outlet that uses your language the way you defined it is one more source telling the machine whose idea it is.
Let your competitors carry your definition
When a competitor adopts your language, the instinct is to fight. Don’t. It is spending its own budget to distribute your category. That April newswire release carried Machine Relations to an audience I never paid to reach.
If you’ve done the work, your energy already went into the record. That’s why it matters so much to date your definition, make its origin easy to verify and keep every public profile saying the same thing. When the record is that consistent, any machine asked about the category finds the origin fast. Then every new adopter makes the category bigger, and the record keeps pointing back to you.
When it happened to me, I didn’t post a rebuttal. I simply reinforced the record by doing more of what I had already done right.
Ask the machines whose definition they use
Every month, ask ChatGPT, Gemini, Perplexity and Google’s AI Mode three questions about your category: what it is, how a buyer should evaluate it and who defined it. Don’t stop at whether you’re mentioned. Check whose language the answer uses and whose sources it cites. How often those sources are yours is what I track as share of citation. If the machine explains your market with a competitor’s framework, you’re competing inside its frame, no matter how many “best of” lists you show up in.
Rankings get reshuffled every time someone presses enter. Definitions get inherited. And once yours is, every ranking you win works harder.
The companies that own their markets in AI search won’t be the ones that show up in the most lists. They’ll be the ones whose frame the machine reaches for when a buyer asks what matters.
In April, a competitor started selling Machine Relations. Ask the machines today who coined it.
That’s the whole strategy.
Key Takeaways
- AI rankings are temporary; the framework machines use to judge vendors can last.
- You do not need to invent a category to lead one; define the problem, establish the standard buyers should use and explain how success should be measured before competitors get to write the rules.
In April, another agency launched a “Machine Relations” practice.
It announced the service on a national newswire, published its pricing and called Machine Relations “the emerging discipline of AI search visibility.” It didn’t mention where the name came from. I coined the term in 2024.
Plenty of founders would have spent that week fighting over credit or losing their minds. I saw it as the moment I had been building toward. A category doesn’t exist because the person who named it says so. It exists when someone else starts using the word as if it had always been there.
