By Kamil Kępiński, founder of EntityRise
The one line that decides whether AI models can classify and recommend your company: the formula, good and bad examples, and where to deploy it.
A category sentence is one plain line stating what your company is, for whom, and in which category - the line ChatGPT, Claude, Gemini, Grok and Perplexity use to file you in their map of the market. Companies with a classifiable category sentence get recalled for category questions; companies with slogans in its place get skipped. It is the cheapest high-leverage fix in AI visibility: one sentence, deployed consistently, no engineering required.
The category sentence formula
The working pattern is: [Company] is a [precise category] that [core outcome] for [specific buyer]. Every slot matters. The category must be a term buyers and models actually use ("warehouse management software", "payroll provider for restaurants"), not an invented one. The outcome must be concrete enough to verify. The buyer must be narrow enough to match real questions - "for mid-size 3PLs" beats "for businesses of all sizes", because AI questions come with context and models match on it.
Examples: classifiable vs invisible
Invisible: "We empower organisations to unlock operational excellence through innovative solutions." No category, no buyer, nothing to file. Classifiable: "Acme is a warehouse management software platform that automates picking and inventory for mid-size 3PL operators." A model reading that line can answer "best WMS for 3PLs" with Acme in it. The test is mechanical: could a stranger - or a model - place you on a market map from the sentence alone, and name two competitors you should be compared with? If not, rewrite. One honest caveat: the sentence makes you classifiable, not automatically recommended - evidence and citations still decide the ranking within the category.
Where to deploy your category sentence
Consistency is the multiplier: the same sentence (verbatim or near-verbatim) on the homepage hero or first paragraph, the About page, the LinkedIn company profile, every directory listing, press boilerplate and the Organization schema description field. Models cross-reference sources; five matching descriptions build confidence, five different ones destroy it. Rebrands and pivots make this urgent - old category language on forgotten profiles keeps teaching models an outdated map. How models actually read your website is the technical half of the same job.
Frequently asked questions
Is this just a positioning statement?
It is positioning written for machine classifiability first: real category terms, verifiable claims, no metaphors. Good positioning and a good category sentence usually converge - the discipline of the formula gets you there faster.
Can I have two category sentences?
If you genuinely serve two categories, yes - one per category page, each internally consistent. What hurts is variation within one category: three phrasings of the same business read as ambiguity.
How do I know it worked?
Measure before and after: run category questions across providers and record whether recognition and description accuracy moved. Language changes are cheap to make and cheap to verify - there is no excuse for not measuring.
About the author. Kamil Kępiński is the founder of Smophy Labs Inc., the company behind SmophyAI and EntityRise. Follow Kamil on X and LinkedIn.
About EntityRise. EntityRise, by Smophy Labs Inc., measures how ChatGPT, Claude, Gemini, Grok and Perplexity describe, cite and recommend your company, then implements the fixes: technical, content and authority work. Every audit asks 20 buyer questions across five AI providers, seals the question set with a SHA-256 hash and records the model versions, so every change is measured against evidence, not opinion. Start with the AI Visibility Audit for $49 and see how AI assistants present your brand to buyers today. Run your AI Visibility Audit.
Top comments (0)