AI brand monitoring: what the assistants say about you, read word for word.

AI brand monitoring means keeping a record of what AI assistants write about your business: whether they describe the right company, who they name instead of you, whether they link to your site, and in what tone. The form above gives a free first reading on ChatGPT, Gemini, and Perplexity. Doing it every day is our paid product.

Why this is not ordinary brand monitoring

Ordinary brand monitoring watches for things people publish: a review, a news story, a post. Something was written once, it has an address, and you can read it, answer it, or ask for a correction.

An AI answer has none of that. It is composed for one person at the moment they ask, it has no address, and it is gone when the conversation closes. Nobody is notified. If an assistant tells a buyer that your firm is a different firm with a similar name, the only way you find out is by asking the same question yourself.

So monitoring here means something specific: asking a fixed set of questions on a schedule, keeping every answer, and reading them for five things.

Five things worth monitoring

The list is our own framework. For each item we say what our product checks automatically and what it leaves for you to read.

1. Identity: is the answer about your business at all?

This is the failure people do not expect. The assistant uses your name and then describes someone else: another company with the same name, in another city or another trade. A name made only of everyday words is the kind our scan treats as a candidate for it.

Our scan looks through the answers that use your name for wording that signals one of three outcomes. A second model call has to confirm it before the answer is labeled:

• Confused. The answer describes a different business that shares your name.
• Not recognized. The engine says it cannot find or identify your business.
• Ambiguous. The engine asks which business you mean.

An answer with one of those labels is not counted as a mention of you, so a namesake cannot inflate your numbers. One guard runs the other way: an answer that links to your own site is never relabeled as confusion.

In the free report this is the section called “How AI describes you when asked by name.” If an engine mixes you up with another business of the same name, or cannot find you, the report shows it with the engine's own words. It appears only when there is something to show.

2. Description: what does it say you do?

Once the answer is about the right business, read how it describes you. Is the category right? Are the services current? Is it repeating something from an old version of your site?

Here we are plain about a limit. Our product stores every answer word for word and shows it to you, and it does not grade the details. Nothing in it decides whether your opening hours or your price range were stated correctly. That reading is yours. Start with the answers to the questions that carry your name.

3. Who is named instead

Most buyer questions do not contain a brand. “Who should I hire for this?” gets a list, and you are on it or you are not. When you are not, the names that are there are the most useful thing in the answer.

The free report has a section for this, “Who AI named instead.” The businesses the engines named on the questions where you were missing, each one checked in the answers themselves. The same check runs on the daily scan.

4. Source: is your site where the answer came from?

An assistant can describe you from your own pages or from what others wrote about you. For each answer we store the links that point to your domain and every other source the answer cited. Over time that shows which outside pages the engines lean on when they talk about your category, and whether your own site is among them. If a description of you is wrong, the cited sources are the first place to look for where it came from.

5. Tone: on the higher plans

Being named is not always good news. An answer can name you and advise against you.

On Solo Pro and Agency, each answer that names your business is classified into one of four tones: positive, neutral, negative, or hostile, where hostile means the answer warns people away. Only answers that name you are classified, and an unclear one is filed as neutral. The free audit and the Solo plan do not classify tone. The text is still there to read.

How to read what you find

How often, and on which engines

Our paid plans ask every day. Solo asks 100 questions a day and Solo Pro asks 200; Agency asks 25 per client on all 5 engines. In a paid question list, at most one question in ten names your business. The rest are the open questions buyers ask before they know any brand, because that is where most of the names-instead evidence is.

From the second scan on, the latest answers sit beside the previous ones, question by question, so a description that changed is visible as a change.

One engine is worth a note for brand work. Claude is available on Solo Pro and Agency, and it is read differently from the others: model mode: how the model understands your brand long-term (labeled, never counted as a live capture). The four live engines show what a buyer is told today.

If you would rather do this by hand, it can be done. Write ten questions, two with your name and eight without, ask each engine in a fresh conversation, save the answers with the date, and repeat on a schedule you will keep. The discipline that matters is using the same questions each time.

What monitoring cannot do

• Show you every conversation. It records the answers to the questions on its list. Assistants adjust what they say to each person, and nobody outside the engine sees those conversations.
• Check the facts for you. As said above: identity is checked automatically, the details are not.
• Change the answer. There is no form to submit to an assistant. What changes an answer is changing what the engine reads.
• Tell you how many people saw it. The engines do not publish how often a question is asked.

The free first reading

The form at the top runs our free audit: 10 questions on 3 engines (ChatGPT, Gemini, and Perplexity), 30 answers, in about three minutes. It needs a website and an email, no account, and no card.

For brand monitoring, know how the 10 are split. 2 of them name your business, a question about whether you can be trusted and a question about whether you are a fit, so you get 6 answers about you by name. The other 8 are open questions with no brand in them, which is where you see who is named instead. It is one reading on one day, and it does not classify tone.

Daily monitoring starts with the 7-day trial, which asks for no card and runs on the Solo plan. That means the trial includes the identity check, the names-instead record, and the cited sources, and does not include tone, which starts at Solo Pro ($299 a month). Plans are on the pricing page.