AIVOSITY
A dark field of faint points, like businesses in a market, with four glowing AI-model hubs connecting to one amber node through thinning fog
Preparing the view

AI VISIBILITY REPORTS

Your customers ask AI first.
Does it answer with you?

We ask ChatGPT, Gemini, Claude and Perplexity the 30 questions your buyers actually ask, three times each, and send you the tally: how often you come up, who gets named instead, and what the models get wrong about you.

  • ChatGPT
  • Gemini
  • Claude
  • Perplexity
See how it works Skip to the report

01 — THE QUESTION

Before they search,
they ask.

"Which dental clinic in Málaga is best for implants?" "Recommend a payroll provider for a 40-person company." People ask assistants this way all day, and each answer names two or three businesses. Everyone else is simply absent.

02 — THE FOG

Most businesses have no idea what AI says about them.

How often you are named, who is named ahead of you, the opening hours that are three years out of date. None of it reaches your analytics, and no model publishes it. If you do not go and look, you never find out.

03 — THE BREAK

Aivosity
clears it.

We ask the questions your buyers ask, across every major model, and measure the answers.

04 — YOUR REPORT

One report.
Every model.
Your real position.

Mention rate, competitor rankings, the errors worth fixing, the sources behind them, and a readiness score for your site — written so you can start on it the same week.

Request your report

HOW IT WORKS

We ask what your buyers ask, then measure every answer.

Search rankings tell you where you stand on a results page. They tell you nothing about the single answer an AI assistant gives. Aivosity measures that answer directly, for your business, your industry and your location.

  1. 1

    30 buyer-intent questions

    We generate thirty realistic questions a potential customer would ask about your industry in your location: comparisons, recommendations, "best for", "near me", pricing and trust questions.

  2. 2

    Four models, queried live

    Each question goes to ChatGPT, Gemini, Claude and Perplexity through their APIs. We record every business named, every claim made about you, and every source the model cites.

  3. 3

    A report you can act on

    We benchmark you against the competitors the models actually recommend, flag factual errors, and crawl your website to score how readable it is for AI systems, with concrete fixes.

MENTION RATE & RANKINGS

Know how often you're the answer

See the share of buyer questions where each model names you, and where you sit in the list of businesses it recommends against competitors you may not even track.

ERRORS & SOURCES

Catch what AI gets wrong about you

Wrong prices, closed-on-Monday, a service you stopped offering in 2023. We list every factual error, the model that made it, and the source it learned it from, so you can fix the source.

WHY IT MATTERS NOW

A recommendation engine has replaced the results page. Nobody sent you the scoreboard.

A results page gives the buyer ten options and lets them pick. An assistant reads the web for them and hands back two or three names. That answer is the shortlist now, and until recently there was no sensible way to see it, never mind measure it. That is the gap we fill.

30buyer-intent questions written for your industry and location
4AI models tested side by side, through their official APIs
3runs per question per model, because answers vary
360recorded answers behind a single report
6report sections, from mention rate to a 30-day plan
5working days from order to delivered report

WHAT BUYERS ACTUALLY ASK

Nobody types keywords into an assistant. They ask it what to do.

These are the kind of questions we write for a report: full sentences, with context and a decision at the end. Your name is never in them. If a model brings you up, it did that on its own.

  • CompareWhich is better for a family of five, a hotel in Nerja or an apartment?
  • RecommendRecommend a dentist in Málaga that does same-day crowns.
  • PriceHow much does a bathroom renovation cost in Valencia, and who is trustworthy?
  • Best forBest coworking space in Barcelona for a quiet, focused week?
  • TrustAre there any complaints about solar installers in Murcia I should know about?
  • Near mePhysiotherapist near Ruzafa open on Saturdays?
  • AlternativesAlternatives to the big payroll platforms for a 40-person Spanish company?
  • LanguageEnglish-speaking immigration lawyer in Alicante with fixed fees?
  • How to chooseWhat should I look for when picking an accountant as an autónomo?
  • RecommendWhich boutique hotels in Seville have rooftop pools and quiet rooms?
  • Best forBest Spanish school in Salamanca for a four-week intensive course?
  • CompareCompare the top three estate agents for buying in Alicante as a foreigner.
  • PriceWhat is a fair monthly fee for a small-business accountant in Madrid?
  • TrustIs that clinic in Marbella properly accredited for implants?
  • RecommendWhere can I buy handmade Andalusian ceramics online with EU shipping?
  • Near meVet in Girona that handles exotic pets?

Recommendation

The classic "who should I go to" question. This is where mention rate is won or lost.

Recommend a dentist in Málaga that does same-day crowns.

Comparison

The buyer already has a shortlist and wants it ranked. Position in the answer matters as much as presence.

Compare the top three estate agents for buying in Alicante.

Best for

A specific need attached to the category. Models answer these from whatever pages state that need plainly.

Best coworking space in Barcelona for a quiet week?

Price and value

Buyers ask what things should cost and who is fair. Invented or outdated prices are a common factual error.

How much does a bathroom renovation cost in Valencia?

Trust and safety

Accreditation, complaints, reliability. Models pull these from reviews, registers and press.

Is that clinic in Marbella accredited for implants?

Practical fit

Language, hours, location, niche. Easy to answer if your site says it; impossible if it does not.

English-speaking immigration lawyer in Alicante with fixed fees?

THE PIPELINE

From three inputs to 360 recorded answers.

The process is deliberately dull, because dull is what makes it comparable: same questions, same models, same number of runs, every answer kept. That is the difference between a screenshot and a measurement.

Aivosity pipeline Business, industry and location feed a generator that writes 30 buyer questions. Each question is sent to ChatGPT, Gemini, Claude and Perplexity several times. Every answer flows into one report. Business Industry Location 30 questions buyer intent, no brand name prompt-engineered, reviewed ChatGPT Gemini Claude Perplexity × 3 runs each Your report mention rate competitor ranking errors · sources website readiness
InputsYour business name, the industry as buyers would describe it, and the location or service area. Nothing else is needed and no logins are requested.
Question generationA prompt-engineering layer writes 30 varied questions across recommendation, comparison, price, trust and fit. You can review them before we run.
QueryingEach question goes to ChatGPT, Gemini, Claude and Perplexity three times, within one window, in the models' default consumer-style mode.
ExtractionStructured-data analysis identifies every business named, its position, the claims made about it, and any cited source, across all 360 answers.
VerificationClaims are checked against your facts and your website; the crawler traces stale claims to their origin and scores your site's readability for AI.

FOUR MODELS, FOUR TEMPERAMENTS

Each assistant builds its picture of your market differently.

So we never flatten them into one score. The gap between two models is the useful part — it tells you which sources each one trusts, and which of them has never heard of you.

ChatGPT

OpenAI

Blends what it learned in training with live web browsing. Answers tend to be confident and list-like, naming a handful of businesses with short reasons. Strong on well-known brands; stale directories can persist in its picture of smaller ones.

Gemini

Google

Grounded in Google Search, so Business Profile data, reviews and Maps signals weigh heavily. Often the first model to reflect a corrected listing, and the harshest on businesses whose profile and website disagree.

Claude

Anthropic

Careful and explanatory, more likely to describe how to choose than to hand over a single name, and more likely to qualify claims. When it does name businesses, the framing (premium, specialist, budget) is unusually consistent.

Perplexity

Perplexity AI

An answer engine built on live search with visible citations. Every claim comes with a source, which makes it the clearest window into which pages shape your reputation and the fastest to reward fixes on those pages.

ANATOMY OF AN ANSWER

One response, four things we measure.

Here is a shortened example for a physiotherapy question. Every answer in your report is marked up like this, so you can check our numbers against the words that produced them.

QWhich physiotherapy clinics in Seville are best for sports injuries?

For sports injuries in Seville, Fisio Nervión is frequently recommended for its work with local football clubs, and Clínica Movimiento has strong reviews for running-related rehabilitation and offers weekend appointments. Centro Triana Fisioterapia is a good option on the west side of the river. Both Nervión and Movimiento list sports specialisations on their websites, and reviews on Google and Doctoralia are consistently positive.

  • MentionYou were named, in second position, with a specific reason ("running-related rehabilitation"). Both the position and the reason are recorded, because they tell you what the model believes you are for.
  • CompetitorsTwo other clinics were named. Across 360 answers these mentions add up to the competitor ranking: the businesses AI recommends instead of you, and how often.
  • Factual error"Weekend appointments" is wrong; the clinic closes Saturdays. The report quotes the sentence, names the model, counts the recurrences, and traces the likely source (an old directory listing).
  • Cited sourcesGoogle reviews and Doctoralia shaped this answer. The source map aggregates these across every response so you know which pages to correct or strengthen first.

WEBSITE READINESS

Models can only recommend what they can read.

Most sites are written for someone skimming on a phone: a slogan, a photo, a button. A model reading the same page finds nothing that says what you do, for whom, where, or roughly what it costs. Our crawler reads your site the way a model would, and scores what it finds.

Six areas, one score, specific fixes

  • Plain statement of what you do, for whom and where, in the first paragraph of the homepage and in the page title.
  • Structured data (Organization, LocalBusiness or Service, FAQ) present, valid, and matching the visible text.
  • Key facts in real text: hours, prices, address and services not locked inside images, PDFs or scripts.
  • One page per specialism, so "best for X" questions have a page that says X in its title and first line.
  • Consistency with third-party listings: the same name, address, hours and services everywhere the models look.
  • Crawlability and freshness: robots rules, load performance, dated content, and whether the pages that matter are linked from the homepage.

INSIDE THE REPORT

Numbers, names and the fixes that move them.

Every Aivosity report is built from real model responses captured for your business. It is not a generic audit. This is an excerpt of what a report looks like.

  • Overall and per-model mention rate across all 30 questions
  • Ranked list of the competitors each model recommends instead of you
  • Every factual error, with the exact model response and its likely source
  • Sources the models cite, so you know which pages and directories shape your reputation
  • Website readiness score from an automated crawl: structured data, clarity, crawlability
  • A prioritised action list for the next 30 days
Explore the full report
Example: boutique hotel, Valencia · 30 questions · 4 models Report ready
41%MENTION RATE
ChatGPT
47%
Gemini
30%
Claude
57%
Perplexity
33%
Competitor ranking
  1. Hotel Marqués del Mar
  2. Your hotel
  3. Casa Turia Suites
  4. The Botanic House
Flags 3 factual errors Closed on Mondays (wrong) Pool listed — none
Cited sources TripAdvisor · Google Business · your website · a 2024 travel blog

FROM MEASUREMENT TO MOVEMENT

The number moves when the sources move.

You cannot edit a model. You can edit the pages and listings it reads. The chart below is an illustration, not a client result — it shows the shape we look for over six monthly runs. The four levers on the right are where the work usually is.

Illustrative mention-rate trend An example business rises from a 12 percent mention rate to 41 percent over six monthly reports after fixing sources and website content, while a competitor drifts from 38 to 34 percent. 0%15%30%45%60% Report 1Report 3Report 56 You · 41% Competitor · 34% 12%

Illustrative figures, not a client case study. Mention rate is the share of question-runs that name the business. What actually happens depends on your industry, your competition and how quickly the underlying sources get corrected — nobody can promise a number here.

  1. Correct the sources the models citeThe stale facts usually sit in two or three places, not two hundred. The report names them, so you can fix the ones that actually feed the answers.
  2. Say the obvious things in textWho you serve, where, for what, at what price. The readiness section shows exactly which claims your site fails to make in a form a model can quote.
  3. Claim the "best for" gapsIf no model links you to something you genuinely do, it is usually because no page says so in plain words. One clear page and a matching FAQ entry is a small job with an outsized effect.
  4. Keep the question set fixed and re-measureMonitoring plans run the same 30 questions each month so the trend is real, and flag new errors and competitor movement as they appear.

WHO IT IS FOR

Any business a buyer might ask an assistant about.

The question set is generated for your sector and place, so the report reflects how people actually shop for what you sell. These are the sectors we work with most; the method works for any other.

COMMON QUESTIONS

Can you guarantee ChatGPT will recommend us?

No, and be wary of anyone who says they can. What we can do is show you exactly where you stand today and which sources and pages are behind it. Changes tend to appear over one to three monthly runs once the underlying sources are corrected, but that depends on your market, not on us.

AI answers change every time. Is one report meaningful?

That is why each question is asked three times per model and the report gives the aggregate. Mention rate is the share of all 360 answers that name you, not the outcome of one lucky query.

Do you need access to our website or accounts?

No. Everything is built from public information plus the facts you give us in the order form. We never ask for logins, and the questions sent to the models do not contain your business name.

All questions and answers

START HERE

Find out what AI
says about you.

Tell us your business, industry and location. Most reports are delivered within five working days.

Request your report
AIVOSITY

Aivosity shows businesses whether AI assistants recommend them, how they rank against competitors, and what to fix to be cited more often. Founded in 2026 and based in Spain.

PRODUCT

COMPANY

LEGAL