Showing up in AI answers

The new zero-click surface

For twenty years, search meant ten blue links and the game was ranking among them. Answer engines changed the shape of the page. The model reads the sources, writes a paragraph, and cites a couple of them. So the question isn’t just “do we rank?” anymore. It’s “when an assistant answers a question in our category, does it mention us, and is what it says right?”

None of this is replacing search, and paid media isn’t going anywhere. But it’s a real and growing place where buying decisions start. Most brands have no idea what the assistants currently say about them.

AEO and GEO, briefly

Two acronyms are doing the rounds. They’re aiming at the same thing from slightly different angles:

• AEO, answer engine optimization. Being the clear, citable source an assistant pulls from when it answers a question.
• GEO, generative engine optimization. Being present and correctly represented inside the generated answer itself, across assistants.

You don’t need to care about the labels. What matters is whether the machine can find you, follow what you do, and quote you without getting it wrong.

What actually moves it

There’s no ranking dashboard for this yet. But the inputs aren’t a mystery. The assistants reward the same things a careful human researcher would:

• Be crawlable by the AI bots. If your robots policy blocks GPTBot, ClaudeBot, PerplexityBot and the rest, you’ve opted out of the answer. Let them in.
• State facts as clean, self-contained sentences. Models lift passages whole. Copy that spells out what you are, who you serve and how you price, in plain declarative language, is copy a model can quote without guessing.
• Give machines a structured version. Schema.org data plus a plain-text summary the assistants can read make your key facts unambiguous. This site ships an llms.txt and in-page structured data for exactly that reason.
• Make the entity unambiguous. One consistent name, a clear description, a real company behind it. If an assistant can’t tell which brand you are, it won’t risk citing you.
• Watch share of voice across assistants. Ask them your category’s real questions and read the answers back. That’s the baseline you work against.

If a model can’t quote you cleanly, it quotes whoever wrote it plainer. Usually a competitor.

How we run it

Presence is one of the six stations in the loop. The agent scans your AEO, your schema and your share of voice across AI answers, then fixes what’s weak. A strategist signs the surface before anything ships. It runs continuously, because the answers keep moving as the models and the sources do.

None of it is a trick to game a model. It’s the unglamorous work of being findable, legible and described correctly. Done every week, not once a year.