AI Visibility Partners

Case file · Local service businesses · US · 2026

What is AI telling your customers right now?

Field Notes is where the measurements publish. Every entry is a real prompt run from a clean, signed-out session, with the engine's answer logged verbatim, dated, and screenshotted. Some entries cut against our own pitch. They publish anyway, because a record that only shows the flattering answers is not a record.

Written by Austin Brewer · Index updated 19 September 2026

Which entries are published?

One so far, with two more scheduled. The first covers remodeling, the first trade we measured; notes on other trades follow as they are run. Each entry is listed with its type, its market, and the date the prompt was run, which is the date that matters: an AI answer logged in September says nothing about what the same engine says in November. Scheduled entries publish on the dates given whatever the measurement shows.

Field Note 01 · Prompt run · Charlotte, NC

Who does ChatGPT name for a kitchen remodel in Charlotte?

Real local firms, not directories. A single dated answer from a clean temporary chat that shows the opposite of what a vendor selling urgency would want you to see. It also shows what the answer leaned on to pick them.

Austin Brewer · Run 25 August 2026 · Read the entry

Field Note 02 · Re-measurement · Scheduled

What moved after 14 days of structural fixes, and what did not?

A client panel re-run against its registered baseline, with the noise floor stated. Publishes on the re-measurement date with whatever the sheet shows, including a null result.

Austin Brewer · Publishes after client sign-off

Field Note 03 · Market survey · In production

Who holds the answer space for local services in ten US metros?

Twenty-five buyer prompts, four engines, ten metros, three passes per cell. The share of answers naming an independent firm, a directory, or a national chain, reported by metro and by engine.

Austin Brewer · Publishing 2026

How is every entry structured?

The same parts in the same order every time: a named author, a run date, the direct answer first, the evidence reproduced verbatim as a labelled exhibit, and a short FAQ. That structure is what we tell clients makes a page quotable by an engine, so our own pages hold to it.

Named author and run date
Austin Brewer, the date the prompt was run, and the date the entry was last updated, all visible and repeated in Article schema.
Direct answer first
The first paragraph answers the headline question plainly. No preamble.
Exhibit
The engine's answer reproduced verbatim, labelled with engine, market, session protocol, and date. Annotations sit in the margin and are marked as ours.
What it does and does not show
One run is one run. Every entry states its own limits before drawing anything from it.
FAQ
Two to four questions a reader would ask next, matching FAQPage schema word for word.

Want the entry about your market?

Send your business name, website, and city. We reply with one real AI answer about your market, logged the same way, no charge and no call required.

Email us your business name

hello@aivispartners.com · Reply within two business days