Accounting and finance

When a buyer asks an AI assistant for an accountant, the answer is usually a listing site. The firm's own pages describe services in language that could belong to any practice in the city, and the partners then ask what they can defend before they ask what the tools can do.

Directories win the citation because they are structured, current, and written in the category terms the lists are titled with. The firm loses because it has never turned its specialism into a sentence a model can quote.

Meanwhile the internal AI conversation stalls after a demonstration, for the same reason: nobody will put their name on a tool without a written position.

On the site, we write the specialism in public: who you act for, which work you take, which work you send on, and how an engagement starts. We put those words on the page itself so an AI assistant can read them without clicking, and we treat the directories as a channel to be eligible for rather than as an enemy.

Off-site work is aimed at the lists those buyers already use. Inside the firm, AI governance produces a written go or no-go per workflow, because that is the conversation that is actually stuck. Operations builds only what that review cleared.

Measurement comes in when the partners can no longer tell which of the two, the site or the directories, is producing the work.

The work is not a scheme to outrank every directory, and it is not a chatbot that files a tax return. Firms that want to look like a general practice when they are not one will not like the copy, because the copy will be specific.

Firms that want AI in the building without a data-path decision will not like the governance, because several workflows usually get declined. The refusals are the point. A list with nothing refused on it is not assurance, and partners in this sector can smell that from the first page.