Making your company readable by AI. Outside, and in. Understand · Answer engines
What do answer engines actually do?
This section explains in plain English how answer engines such as ChatGPT produce what they say, what an AI visibility figure actually measures, and what a provider can honestly promise you. It is written for a business leader who needs to decide without being a technician: each page answers one question, cites its sources and carries its date. We apply to our own figures the rules we ask others to follow, including the zeros.
The four pages are designed to be read in the order below, and each one stands on its own.
The four pages, in reading order
01
What are AI visibility figures really worth?
A visibility figure does not describe your company. It describes a measurement procedure: this page sets out the four decisions a tool makes on your behalf, and the seven questions to ask a provider before subscribing.
02
Why can nobody guarantee that you will be cited?
The engine decides whether to cite you each time it produces an answer: this page explains why no guarantee holds, what a provider can commit to instead, and the four phrases that should stop you when you read a quotation.
03
When does ChatGPT actually search the web?
Of thirty questions asked during the night of 14 to 15 August 2026, the application searched the web for only eleven: this page shows which ones, and why they were precisely the questions that decide a purchase.
04
How does a language model work?
The machine does not consult a directory of companies. It composes its answer piece by piece: this page explains the mechanism plainly, and the four things a language model does not do.
Page published on 16 August 2026; updated when the facts change. Each page in this section carries its own date, its own sources and its own limitations.