We ask the AIs what your customers ask. Then we fix what they say.
No prompts to write, no dashboard to learn. You give us a web address and we do the rest, in plain language.
01
You give us a URL
That is the whole setup. We read your site, work out what you sell and where, and you confirm it in three fields.
02
We ask what your customers ask
Real buying questions put to ChatGPT, Claude, Gemini and Perplexity with live web search. Not keywords: the sentences people type when they are ready to spend.
03
You get the plan, not a chart
Every gap arrives with the page, file or listing already prepared. You publish it, tick it done, and the next check proves whether it worked.
The way people find businesses changed.
800 million people ask ChatGPT things every week, and 82% of what AI says about a business comes from sites the business doesn't own (Muck Rack). Your website alone can't fix that.
800M
weekly users
ChatGPT alone passed 800 million weekly users.
OpenAI, Oct 2025
1.5B
people a month
Google now shows AI-written answers to over 1.5 billion people a month.
Google, 2025
−25%
search volume
Traditional search volume falling 25% by 2026.
Gartner projection
An assistant doesn't return ten links. It names two or three businesses, in a sentence, with confidence. There is no page two.
800M+
people use ChatGPT every week, and a growing share ask it what to buy and who to trust.
Source: OpenAI, Oct 2025
A handful
of businesses fit in a typical AI recommendation. There is no page two.
Source: Observed across answer engines
133 vs 11
Google reviews: the average for businesses AI recommends, against the average for the ones it ignores.
Source: Insites study of 10,000 businesses
82%
of AI citations point to third-party sites, not to the business's own website.
Source: Muck Rack citation analysis
-25%
forecast drop in traditional search volume as customers move to AI assistants.
Source: Gartner, 2024 forecast
~11%
of cited domains overlap between ChatGPT and Perplexity. Each assistant reads its own corner of the web.
Source: Profound, 100,000-prompt study
Every number here is published research, with its source named. We didn't run any of it, and we don't invent statistics.
One engine for how AI answers about you.
AEO, GEO, AI SEO: three names for one job. We find the questions you lose, build the evidence that changes the answer, and re-check every Monday. You publish what we hand you.
01 · Measure
40
questions, four assistants
Real buying questions put live to ChatGPT, Perplexity, Claude and Gemini, web search on. Every answer verbatim, every cited source recorded.
02 · Engineer
11
findings, each one written
The evidence those answers are built from, produced for you: pages, structured data, listings, citations. Ranked, ready to publish.
03 · Prove
Monday
re-checked weekly
Every change re-checked against the same questions. Before and after, side by side. Moved is moved; not detected yet is said plainly.
A single check is a photograph. The program is the film: the same questions, re-measured every Monday, with each change you publish tracked to its result. Priced like a tool, not like the agency that charges four figures a month for the same disciplines.
Technical SEO, structured data and schema markup: the part assistants actually read.
An assistant reads a page in two layers: the words a person sees, and a structured layer written only for software. Schema markup and JSON-LD are that second layer, and crawlability decides whether either is reachable at all. It is unglamorous work, it is most of the result, and it still works in Google.
AEO, GEO and AI Overviews: the same work under three names.
Answer Engine Optimization, Generative Engine Optimization and AI Overviews describe overlapping work: getting quoted in an answer rather than ranked in a list. We do not sell you the acronym. We show you which questions you lose, hand you the files, and check every week whether it moved.
Evidence first, then the plan.
Your report opens with proof: the real answers, word for word, and the businesses named instead of you. Then it shows every way to change that.
01
The verbatim answers
Not a chart. The actual sentences ChatGPT, Claude, Gemini and Perplexity produced when asked your customers' questions, with the sources they leaned on.
02
Who got named instead
A leaderboard of the businesses the assistants recommended, counted across every answer, with your own position in it.
03
Your report
Every improvement worth making, grouped by area, ranked by evidence. Reviews outrank code changes for local businesses, and the map says so.
04
The materials, prepared
The corrected files, the written pages, the listing descriptions and a note to forward to whoever manages your website.
Honest by construction.
The models are not deterministic, so we built around that instead of hiding it.
- Every question is asked with live web search enabled, so answers reflect today rather than stale training data.
- Questions are sampled more than once, and your score is shown as a range instead of one confident number.
- An assistant that fails to answer is excluded from your score rather than counted against you.
- If we cannot read your site, we say so. We never show a result we could not measure.
Find out if AI recommends you.
Paste your website. See the actual answers, and who gets named instead of you. Free, no signup.
Show me who AI recommends