How it works
How our AI reputation management works, from the first call to the fix.
An engagement runs in four steps, with a later reading as a fifth if you commission it. Nothing is commissioned until you accept a written proposal.
A free consultation.
We learn about your brand, what you see in AI answers today and the outcome you want, then agree the scope and depth of the work and the budget it should fit. During the call we show you a first reading of your own website, as a sample of what the analysis finds. We also check that we are the right firm for the matter, and tell you if we are not.
A written proposal.
It sets out the subjects, the questions to cover, the timing and the fee. You can commission one report or readings through the year.
The analysis.
Our analysts work from our own insights platform. For each brief we write tens of thousands of questions, phrased as your buyers phrase them, and run them through the latest AI models. We read every answer against your own pages and record the sources it drew on, including pages you do not own. Your first report arrives within 3 days of the agreed brief.
The correction.
Each finding ends with the wording and the page we propose, and the report ranks them in priority order. We work through the list with your in-house team or agency, so the changes are published promptly and reach AI engines sooner. You decide what to publish.
A later reading.
If you commission one, we put the same questions to the same models again. It shows which answers moved and which still need work.
The sample and the full analysis
The sample is a first reading of your own website, prepared for the consultation and free of charge. It shows the kind of finding we write and where your account may need work. The paid analysis covers every subject and question agreed in the proposal, across the latest AI models, and writes each finding out in full: what the models say, the passage on your site it should match, why the gap matters and the change we propose.
Working with your team or agency
We go through the report with the people who will act on it: your communications, web or product team, or the agency that handles this work for you. Each proposed change names its page, so it can be briefed as written. Your team keeps control of the site.
Questions about AI reputation management
- How do I get AI assistants to speak well about my brand?
Give them accurate facts to repeat, on pages they can read. AI assistants describe you from what they find, so start with your own website: state the facts buyers ask about clearly, with figures, conditions and dates.
Our AI visibility toolbox covers the technical side. Our analysis then shows which of your claims AI already repeats and what to publish next.
- Why isn’t AI saying what I wrote on my website?
Often, AI cannot read the page. A setting in your robots.txt file or firewall, or a page that only shows its text after JavaScript runs, can keep AI out.
If AI can read the page, other sources may say it more clearly or more often, such as reviews, resellers or an old article. Or the claim is hard to repeat, like a figure without its unit. The toolbox covers the technical fixes. Our analysis finds and fixes the rest.
- Why doesn’t AI read my website?
Usually something is blocking it. Many sites block AI by accident, through a robots.txt rule or a setting at their hosting or security provider that was switched on and forgotten. Other sites only show their text after JavaScript runs, and several AI tools cannot run it.
The AI visibility toolbox lists the AI crawlers by name and the checks to run, in order.
- AI assistants already mention us. Why does the wording matter?
Being mentioned is AI visibility: it puts you in the answer. What the answer says about you is your AI reputation, and it decides whether the buyer shortlists you. If a model describes your product by an attribute every rival shares, the work that went into designing it does not reach the buyer. We measure both: where and how often you appear, and whether what is said matches the product you built, claim by claim. If you already track AI answers, we can start from the ones your tool has collected.
- How do you find out what AI says about us?
Through AI narrative analysis. We begin with open questions that give the models no brief, to learn what they already know about you. Then we narrow, question by question, toward the attributes that set you apart: a specification, a credential, a role, a date. A brief runs to tens of thousands of questions across the latest AI models, worded the way your buyers ask. One question in one assistant shows one answer. The full set shows which claims hold across models and which change with the wording.
- Can you compare us with our competitors?
Yes, competitor analysis is one of the primary services we offer. You can name your competitors directly, or we identify your closest peers from the products you sell and the buyers you compete for. The comparison shows where you rank on each requirement your buyers raise, which claims the models repeat for each brand and which sources they cite. Where a rival is described more precisely than you, the report quotes the passage the models drew on.
- What do you read besides the AI answers?
Your website first, because it is the account you control. We also analyze primary data sources that reference your product, brand or person such as FDA, EPA, SEC, financial reports, clinical trials, etc. Then any material you choose to share under confidentiality: new product information, press releases, published research, unpublished internal documents, product specifications, independent third-party reports and other documents. We also record every source the models cite, including pages you do not own. Each finding quotes the passage it relies on, so your team can check our reading line by line.
- How do you tell a wrong answer from a fair one?
Each finding states what the answer says, which product, person or condition it concerns, and how it compares with your website. A factual error, a missing condition, a difference of opinion and an unsupported allegation are reported separately, because each calls for a different response. A criticism your own record supports is reported as supported.
- How soon will we see a difference?
The first report is ready within 3 days of agreeing the brief, and sooner when the matter is urgent. How quickly the answers change depends on the matter and the size of the gap. The fastest change we have measured took less than 24 hours, from a client publishing our proposed wording to the models repeating it. Broader topics move more slowly, so we set the pace to the matter: a sprint of days to weeks, or a monthly cycle of readings and actions.
- Will you tell us if someone is spreading a damaging account of us?
Yes. We flag claims and allegations that have no factual basis, identify the publishers and accounts behind them, and show how many of the models repeat them. In past matters we have identified networks of publishers and automated accounts, run by paid actors, that spread allegations later proved false on X, Reddit and LinkedIn, and in full-length articles. When the pattern points to coordination, the report says so and sets out the evidence, so you and your counsel can decide how to respond.
- Do you remove or suppress negative information?
No. Published material rarely disappears, and we do not try to remove it. Our work is to give models a documented account to find and repeat. We set out the facts with their sources and, with your approval, publish them where models read, keeping each allegation next to its answer: the ruling, the correction, the date it was resolved. An allegation with no support is hard to sustain beside a clear, cited record. Content published unlawfully, obtained without permission or in breach of your rights is a separate matter: the law provides routes to removal, such as a cease-and-desist letter, which your counsel can pursue.
- Who carries out the recommendations?
Whoever you choose. Your in-house team or agency can publish the changes we propose, or we can draw up an action plan and carry it out with your approval. Either way, you decide what is published. The same confidentiality and terms apply whether you work with us directly or through the agency that represents you.
Start with a free consultation
We agree the scope, depth and outcome you want, and show you a sample of our work on your own website.
Request a free consultation