The buying journey moved
Buyers used to type a category into a search box and open ten tabs. Now they describe their problem to an AI assistant in a sentence or two and ask who they should consider. The assistant answers with three or four names and a reason for each.
The shortlist is made in that conversation. The website gets visited afterwards, if at all, and mostly by buyers who already know what they want to confirm.
The shortlist is made in the conversation. The website is visited afterwards.
AI recommends what it can understand
An assistant recommends companies it can describe with confidence: what they do, who they serve, and how they differ from the alternatives. That confidence comes from what it can read about the company, and from how clearly that material is written.
A site that talks around its product gets summarised vaguely, or skipped. A company whose positioning lives in a sales deck and a founder’s head does not exist to the assistant at all.
Silence is not neutral
When a buyer asks and the client is not in the answer, a competitor is. Every recommendation the client is missing from is a deal that went elsewhere without anyone at the client ever knowing the question was asked.
This is the part that changes how clients see the problem. It is not a marketing channel they can choose to ignore. It is a room where their buyers are already talking about them, and someone else is doing the describing.
It can be measured
Ask the assistants the questions the client’s buyers actually ask. Count how often the client appears. Note who appears instead, and what the assistant says about each. Repeat next month and compare.
That is the whole discipline. It takes an afternoon by hand, which is why almost nobody does it twice. The AI visibility measurement guide covers how a Haycion workspace runs it on a cadence instead.
What a consultant can do about it
The good news is that this is a positioning problem, and positioning is the work consultants already do. The difference is that the audience now includes a machine that reads literally.
- Say plainly what the client does, and for whom. One sentence an assistant could repeat without hedging.
- Make the differences legible. Not “best in class”, but the specific thing a competitor does not do.
- Keep the public picture current. A pricing change or a new segment that only lives in a deck does not exist to AI.
- Check what the assistants actually say. Ask the buyers’ questions, read the answers, fix what is wrong at the source.
- Report the change. Recommended more often this month than last is a number a client understands.
What Haycion does about it
A client agent is a public, structured profile of the client that Claude, ChatGPT and other AI assistants can read, plus a monthly check of whether they recommend the client and who wins instead. You set it up from the client’s workspace, within guardrails you and the client agree.
- Grounded in the workspace. The agent says only what the profile you checked says.
- Checked every month. The buyers’ questions, asked again, with the answers compared.
- Reported under your name. The client hears it from you.
What the client receives
- A plain account of why AI visibility now decides who makes the shortlist, in language you can forward as-is.
- The five moves that fix it, in order, starting with what to say plainly and ending with what to report.
- A client agent that keeps the public profile current and checked, so the fix does not go stale.
How it shows its work
Every claim above is checkable, not asserted: ask the assistants the buyers’ own questions and read what comes back. The measurement guide shows the same discipline run automatically, with the citation rate, the cited pages and the competitors named instead, so “recommended more often this month” is a number built from probes you can read yourself.
What a result looks like
A client who has never checked typically finds out they are missing from more answers than they expected, and that a competitor with a weaker product is not. Saying plainly what they do, tightening one or two differentiators, and publishing a client agent usually moves the needle inside a month: not to universal recommendation, but to being in the conversation their buyers are already having.