Get recommended by AI. Correctly.

It is the work of getting your company named, and named accurately, in the answers AI assistants give. A search engine returns ten links and lets the buyer choose. An assistant returns one answer naming a handful of companies, and most buyers never go past it. Answer engine optimisation is how you get into that answer.

SEO competes for a position in a list of links. An assistant does not return a list, so there is no position to win in the old sense. What decides the answer is whether the facts a model compares on are stated plainly, as text, on pages it is allowed to crawl. Backlink profiles and keyword density barely move it. Specifications, prices, coverage, certifications and third-party mentions move it a great deal.

We run the buying questions your customers actually ask, across every major assistant, and record the answer, the ranking and every source it cited. Then an AI agent acts on what it found: it gets you listed in the sources those answers cite, writes the missing facts into your pages, and re-runs the questions so you see the answer move. You approve each step.

It works through the recommendations from your check. For each source the answers cite, it creates or completes your listing, pitches inclusion where a list is editorial, and comes back to keep it current. On your own site it writes the missing facts as plain text. Every action is logged, and anything that touches your public profile waits for your approval.

ChatGPT, Perplexity, Google AI Overviews, Claude and Gemini. Your check runs the questions your buyers ask on all of them, in the language of the market you sell in.

We run each question the way a buyer would, with live web search enabled, and capture the answer as it was returned. We keep the full response, the ordered list of companies it named, and the source URLs the model actually read, so every finding traces back to what an assistant said and on which date.

Monthly across your full question set, daily on the questions you choose to track. The lists are recomputed on every request rather than earned once: tracking the same three queries on consecutive days in September 2026, one brand went from third place to absent within a day. A one-off fix decays, which is why the agent keeps at it.

That is most of the value. Every answer we capture is a ranked list, so you see who is named above you, how often, and which sources put them there. Where a competitor is cited and you are not, we show you the page the model read and what it states that yours does not.

Seven languages across Europe, plus the UK and US. A buyer in Amsterdam and a buyer in Milan get different answers to the same question, so we run each question in the language and market it is actually asked in rather than translating an English result.

Because it can only compare what it can read. In the vendor shortlist we captured on 11 September, the reasons given were capability, accreditation and inspection history far more than price or how well known the company was. Those are facts a page either states plainly or it does not.

Four things. Be on the third-party pages the models already cite in your category: that is where most answers are decided, and where the agent starts. State the facts a buyer compares on as text, not inside images or PDFs. Let the crawlers in, since GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot are blocked by default on a surprising number of sites. Then keep it current, because the lists move.

Not to check it. Everything the check reads is public, which is exactly the point: it is what a model can see. Listings on third-party sites need nothing from you but approval. Changes to your own pages are the one place we ask for access, or hand them to your developer.

Yes. Agencies run the check and the agent across a client portfolio from one account, with separate reporting and approvals per client.

The check is free and nothing is held back. The paid work is the agent: every question in your category, across every engine, the listings and page changes actually made, and repeated as the answers move. We price per company once we have seen your check, and tell you in the reply rather than after three calls.

The check reads public pages, so there is nothing confidential in it. Anything you do send us stays in the EU, is used only for your own listings and pages, and is deleted on request. We are GDPR compliant and will sign a data processing agreement.

Because answers are generated per request, not looked up. Your location, your language, your account history and the moment you ask all move the result, and two people asking the same question minutes apart can get different companies named. That is exactly why one test proves nothing and a tracked question over time proves everything.