Ten years ago a buyer asked Google whether a company was any good and got ten links. The buyer made up their own mind from the evidence.
Today a buyer asks ChatGPT or Claude the same question and gets one synthesized paragraph. The engine has already read the news coverage, the reviews, the Glassdoor entries, the court filings, the Reddit threads, the LinkedIn profiles, and the Wikipedia page. It compresses all of that into a few sentences and presents them as fact.
That paragraph is now the reputation. Not the press clip. Not the Yelp score. The paragraph.
Two things follow. First, the paragraph is measurable — across the five engines, across thousands of prompts, refreshed quarterly. Second, the paragraph is improvable. Source quality, citation graph, original reporting, structured data, schema, and authoritative coverage all change what the model writes the next time it is asked. This is the work of AI Communications, applied to the reputation problem.
Brands that do not measure the paragraph cannot improve it. Brands that do measure it discover, often for the first time, what the world's most-used reference tools are actually telling buyers about them.