Someone asks an assistant to recommend a person in your specialty. Your name comes up. And after that, almost nothing visible: in Google’s AI summaries, one in a hundred people opens the cited source, according to Pew Research (2025). It’s worth separating what actually happened from what we imagine happened.
What the machine did for you
It made the introduction. Someone who didn’t know you read a description of you, with your name on it, exactly when they were looking. That work used to be yours and it was expensive: a referral from a third party, a talk, a piece of content that reached the right person. Here it happened inside an answer.
What you won’t be able to see
None of it reaches you. There’s no notice, no log, no panel showing that you were named. If you don’t go look, you don’t find out.
That’s the uncomfortable part and it’s worth saying plainly: the effect exists and the evidence is thin. Anyone showing you an exact figure for how much business an AI recommendation brought you is estimating, not measuring. Which is why going to see what is answered about you today stopped being curiosity and became the minimum.
The one thing you can observe
Two things, actually, and both in your own conversations rather than on a dashboard.
The first is what the person who writes to you already assumes. When someone arrives with a formed idea of what you do and asks straight out whether you can handle theirs, someone gave them that idea first. It could have been a colleague. Increasingly, it could have been a machine.
The second is which idea they arrived with. If what they take for granted about you matches what you want said, the introduction went well. If it places you on ground that isn’t yours anymore, it went badly, and it went badly in the conversations that never arrived too.
It isn’t a metric. It’s an observation, and for now it’s the most honest thing available.
What’s at stake when the answer is wrong
Here there’s no ambiguity. A recommendation built on old facts introduces you to the wrong client, or leaves you out for the right one, and in both cases without your being able to step in. Silence and error have the same practical consequence: for that conversation, it wasn’t you.
That’s what changes. Not that more work appears by magic, but that a piece of your introduction moved into the hands of something you don’t control and can feed well.
If you want the ground underneath what it means for a model to name you, it’s in what it means to be cited by the AI.