Ma Réputation En Ligne
FR
// AI visibility · AI search · GEO

AI search: be recommended by ChatGPT, Perplexity, Gemini and Claude.

We run the 50 most likely questions in your market through the 4 major AI engines. You leave with a precise map: when your brand is recommended, alongside whom, and what to fix so that it happens more often.

See the plans
Audit delivered in 48 hNo commitment4 AI models + Google
// The shift

SEO is not going away. But it is no longer enough.

A growing share of search now happens as a conversation with an AI model. Unlike Google, an AI model does not hand back a list of links: it cites, and it recommends, and the traffic it sends is already qualified. If your brand is not recommended in those answers, you are invisible to an audience your competitors are already capturing. That is the bet we reorganised the agency around.

800M

Weekly users on ChatGPT, and still growing.

OpenAI · Oct. 2025
-25%

Of traditional search volume by 2026, moving to AI chatbots.

Gartner · 2024
3/4

Brands absent or badly positioned in AI answers today.

MREL internal audit
// Understanding GEO · LLM 101

Being recommended by an AI model
does not happen on its own.

One stubborn assumption: “our site is maintained, it should come up everywhere naturally”. Here is, simply, why that is not the case — and why GEO is real work, distinct from the maintenance you already pay for.

« We already pay for maintenance. The site should come up on its own in Google and in AI models, shouldn't it? »
The honest answer: no. And it is not a question of money badly spent — it is that maintenance, SEO and GEO are three different crafts. Here is why.
// First — how does an AI model “think”?
Classic engine
Google ranks pages
You type a query, it returns a list of links. It is up to the reader to click, compare, choose. Your job: climb the list.
← you (with SEO done)
Generative engine
The AI writes an answer
It does not list: it synthesises a single answer from everything it has learned about you, across thousands of sources. It recommends you… or it forgets you.
› “best online reputation agency?”
Being first on Google does not guarantee being recommended by ChatGPT. Two mechanisms, two pieces of work.

: the criteria Google and AI models use to judge whether a source is reliable — experience, expertise, authoritativeness, trustworthiness. Those are what we work on.

// Three crafts people confuse all the time
01 · Maintenance
Keeping the site alive
Updates, security, backups, uptime.
  • The site stays online and secure
  • Creates no visibility at all
  • Gets you neither found nor recommended
02 · SEO
Being found on Google
Content, technical work, authority, keywords.
  • Ranks you in Google results
  • A useful foundation for AI models
  • Does not get you recommended by an AI model
03 · GEONew · 2026
Being recommended by AI
LLM-ready structure, multi-source presence, monitoring.
  • Gets you recommended inside AI answers
  • Reaches an audience SEO no longer touches
  • The ground your competitors are not on yet
Maintenance keeps the building standing. SEO puts your sign out on the street. GEO is about making sure that, when someone asks an AI model for advice, you are the one it recommends.
The models change — a new version every month.
The sources they read keep shifting.
Your competitors are optimising too.

GEO is active upkeep. Not something you acquire once.

Without follow-up, your position decays — exactly like search rankings left to themselves. That continuous work is precisely what we take on.

// The AI visibility audit

We do not guess your AI visibility. We measure it.

Every question in your market is put 5 times to each of the 4 AI models, because an AI model never answers twice the same way. What comes out is a real recommendation frequency, not a snapshot taken at random.

01

Scan of the 4 engines

Recommended or not, in what position, against which competitors.

02

Technical score out of 100

Structure, schema, speed: what makes your site readable by AI models.

03

Priority score per query

Google volume, buying intent and competitor overlap, so you act first where it counts.

04

Results tied back to reality

AI traffic and Google clicks through your own data: GA4 and Search Console.

Measured repeatedly, tracked month after month. The difference between “I think we are visible” and “here is our curve, engine by engine”.

// And it is not one AI model, but four. Plus Google.

Every engine has its own rules.

Being recommended by ChatGPT tells you nothing about where you stand in Perplexity, Gemini or Claude. Each reads different sources, weights them differently, and updates its model on its own schedule. And Google is still the foundation.

ChatGPTOpenAI

Draws on its training corpus and on web search. Rewards authority and sources that stay consistent over time.

PerplexityLive search

Cites sources retrieved in real time, and shows them. Freshness and how quotable the content is come first.

GeminiGoogle

Draws on the Google ecosystem. Classic SEO, the knowledge graph and Google's own signals weigh more here.

ClaudeAnthropic

Favours reliable sources and clear, well-structured content. More sensitive to quality than to volume.

GoogleClassic search

Being visible on Google remains essential, and it often feeds your presence inside AI models. SEO is the foundation of GEO.

Optimising for a single AI model is therefore not enough, and a method that works on one can be useless on another. That is exactly the point of being followed by specialists: adapting the work engine by engine, continuously. : the criteria Google and AI models use to judge whether a source is reliable — experience, expertise, authoritativeness, trustworthiness.

// Our method & our tool

Our expertise, turned into an instrument.

We are not an agency that bought a tool. We are search and reputation specialists who built their own, because our method needed an instrument to match it.

On the ground, since 2018

More than 250 clients supported

Our original craft: making brands visible and credible on Google, with a deep specialisation in healthcare. When AI models became the new front door, we were already there.

→
The method

Know-how, codified

What makes an AI model recommend one brand over another, we learned by doing, not by reading. We turned that know-how into a method that repeats.

→
The tool

Built in house

Then we turned the method into technology: a measurement and steering tool that queries AI models continuously, on your behalf.

Illustrative preview with fictional data: each question is put to ChatGPT, Perplexity, Gemini and Claude, and we record whether your brand is recommended.
// What we believe

AI measures. People decide.

Nothing here runs on full automation: the technology does what it does better than we do, and our specialists do the rest.

What the AI does

AI agents built for GEO

  • Built in house, trained on our method, not generic tools
  • They query, measure and monitor the 4 AI engines, continuously
  • Including one agent dedicated to healthcare, the ground we have worked since 2018
×
What the people do

Specialists who decide and deliver

  • A dedicated project manager who knows your account, not a support ticket
  • Every recommendation is signed off by a specialist before it reaches you
  • Strategy, judgement calls, the relationship: you do not delegate that to a machine
The experience of 250+ clients since 2018, encoded in our agents, carried by our teams.
// How it works

Four steps, one cycle that repeats every month.

The real questions your customers put to AI models. Not keywords.
ChatGPT, Perplexity, Gemini and Claude, asked several times over, because a single reading proves nothing.
Finding, recommendation, prioritised action. Carried out by our teams.
Every month, we measure the whole thing again.
For you
You finally know where your visibility is decided.
For you
Your real recommendation rate. And who gets recommended instead of you.
For you
A plan that moves, without you having to run it.
For you
You read the numbers, not our promises.

This technology we built in house — it does not exist anywhere else in this form.

No white-label software resold to you: our own measuring instrument, shaped by our method, working for your visibility.

// Our plans

One plan per situation. Find yours.

Healthcare practitioner, company already online, or a business starting from nothing: each case has its plan. Engagements start at €590 a month.

Find my plan in 3 questions
Healthcare practitioner

GEO Healthcare

Practice, clinic, practitioner: medical, dental, aesthetic. A measurement agent dedicated to healthcare, and the questions patients actually ask.

Explore You already have a site

GEO Business

Company, brand, retail, independent, professional services. We measure where you stand, build the foundations, then steer month after month.

Explore You are starting from nothing

GEO Launch

No site yet? We build it, designed from day one to be read by AI models and by Google, then install your brand in their answers.

Explore
// Frequently asked

AI search, in plain words.

The questions we get asked most often, answered without jargon.

All the questions
What is GEO, or AI search optimisation?
GEO — Generative Engine Optimization — means optimising your presence so that AI models recommend you in their answers. It is the equivalent of Google search optimisation (SEO), but for the new generation of engines: ChatGPT, Perplexity, Gemini, Claude.
How is it different from SEO, from classic Google?
SEO gets you into Google's list of links. GEO gets you recommended inside the AI model's answer. The two work together: performing well on Google often feeds what AI models remember about you.
Is GEO an add-on to your SEO service?
No — it is the other way round. We made the bet that AI models will become the first source of qualified traffic: GEO is the cornerstone of every strategy we build, and SEO, content, reviews, the site and social media are levers in its service. Other agencies add GEO to their catalogue; we reorganised ours around it.
How do you know whether an AI model recommends me?
We put the questions your customers actually ask to the main AI models, and we look at what comes back: are you mentioned? How? Alongside whom? That gives a precise picture of where you stand today and of what needs fixing.
What does an engagement actually look like?
We start by measuring where you stand, we build the foundations (site, content, credibility), then we steer month by month according to the results, with a dedicated project manager.
Do we need to have a site already?
No. If you have one, we work on it. If you are starting from nothing, we build it, designed from day one to be read by AI models and Google.
Do you guarantee that we will appear in ChatGPT?
Yes, for visibility: a brand that had no existence at all in ChatGPT's answers exists in them after our work, and we commit to 50% more visibility inside AI models over the first four months, once the strategy is in place. Being recommended is another matter: AI models decide their answers on their own. That is the objective we pursue and measure every month, engine by engine — not a guarantee.
How much does it cost?
Our engagements start at €590 a month depending on your situation (sector, existing site or not). The best way is to talk it through: we will point you to the right one.
We are not based in Paris, Lyon or Barcelona — can you still work with us?
Yes. AI search optimisation is done remotely, across France and beyond. Our offices are in Paris, Lyon and Barcelona, but the work itself is entirely remote.
// Resources

Our latest analysis.

Every article is itself a demonstration of the GEO method.

Décryptage

L'IA répond, on acquiesce. Sommes-nous en train de perdre la remise en question ?

Elle répond à tout, tout de suite, avec assurance. Et nous acceptons ses réponses précisément dans les domaines où nous serions incapables de les contredire. Ce que la recherche mesure, et six réflexes pour vérifier un travail qu'on ne sait pas faire.

9 min read
Décryptage

ChatGPT Ads en France : les opportunités réelles, les risques et les limites

Un canal publicitaire neuf s'est ouvert en France il y a une semaine. Voici ce qu'on peut en dire honnêtement : les opportunités qui tiennent, les limites qu'on ne peut pas contourner, et ce que personne ne sait encore.

8 min read
Méthode

Comment apparaître dans les réponses de ChatGPT, Perplexity, Gemini et Claude

Guide pratique, tous secteurs. Ce qui marche, ce qui ne marche pas, et ce qu'on ne peut honnêtement pas promettre.

9 min read

Our analysis is published in French.

All resources →
// Next step

Do you know whether AI models recommend you?

We put your market's questions to the 4 major AI models and tell you who they recommend. Delivered in 48 hours, no strings attached.