Activation
We fix the pages the model cites. Your site — ourselves. Pharmacies, clinics, media, trade journals — through negotiation with editors.
Result: published edits and a repeat measurement using the same questions.
Your brand in AI answers
We measure what ChatGPT, Claude, Gemini, Alice and GigaChat answer. We find the pages they took it from. Yours we rewrite, with third-party sites we negotiate.
Scenario: which toothpaste to buy for a child with sensitive gums?
“For a child’s sensitive gums, the most often recommended are Aquafresh Big Teeth and Oral-B Junior. Sensodyne Pronamel Kids is not sold in the UK…”
brand in the answer
factual error
Finding: two competitors lead the scenario and the distribution fact is outdated. Action: the answer’s source map is ready in the dashboard — fixes start with the two highest-weight platforms.
Scenario: which family SUV under $35,000?
“Most often recommended are Toyota RAV4 and Honda CR-V. The Mazda CX-5 is praised for handling, while the Subaru Forester is said to lack all-wheel drive…”
brand in the answer
factual error
Finding: two brands dominate the scenario’s Share of Voice and one claim is factually wrong. Action: the report lists cited reviews by priority; updating the two highest-weight ones closes the factual error.
Scenario: which car insurance to choose for a new car?
“Consistently high ratings go to Aviva and Direct Line. Admiral is praised for multi-car discounts, while some insurers are said to be slow with payouts…”
brand in the answer
claim without a source
Finding: the negative claim is reproduced in 4 of 5 runs — with no source. Action: the platform traced the chain to the original article; a PR task is set, the next wave will verify the effect.
Scenario: which Moscow residential complex for a two-room flat under 20 million rubles?
“Frequent suggestions are ZILART and Lyublinsky Park. Symbol has good transport access, but prices from 14 million are already outdated…”
brand in the answer
factual error
Finding: answers concentrate on three developers’ projects, prices lag the market. Action: Semantic Match shows the models don’t see your price pages — they are first in the fix plan.
Scenario: which cashback card to choose for travel?
“Popular options are the Citi Premier card and Chase Sapphire Preferred. Capital One Venture X offers good miles on large spends…”
brand in the answer
outdated brand name
Finding: the model still uses the pre-rebrand name: the source corpus is outdated. Action: a register of outdated mentions is exported from the dashboard; a control measurement follows the clean-up.
Scenario: which robot vacuum for a home with pets?
“The leaders are Roborock and Dreame — both handle pet hair well. Xiaomi is cheaper but some units are said to ship without an official warranty…”
brand in the answer
needs verification
Finding: two brands hold the top, the distribution claim needs local verification. Action: the hypothesis is scheduled into the wave — twenty RU scenarios will confirm or lift the limitation.
A new channel
58% of consumers have already replaced traditional search with generative AI (Capgemini, 2025). Purchase decisions increasingly start with a question to a neural network, and its answer becomes the customer’s first meeting with a brand. The same happened with search results in the 2000s and targeted ads in the 2010s: the best positions went to those who arrived on the new territory first.
A report changes nothing. Edits on pages do
Measurement
We put buyer questions to five AI systems and repeat each one several times. We save the answers and work out which pages the model cited.
Result: a map of sources — which sites shape the answer about your category and what exactly they say.
Your site is a source like any other. The only difference is that it needs nobody’s permission. What we change and how — on the activation page.
AI tools for SEO
We do not sell an add-on to your SEO. We do the SEO analysis itself — built on what AI systems cite when they answer buyers — and we take the edits through to publication ourselves.
- A classic SEO audit. Query volumes, the technical state of the site, structure, markup.
- An AI layer on top of it. Which pages an AI cited when answering a buyer’s question, and what exactly it took from them.
- Implementation. We prepare the edits and take them through to publication: on your site ourselves, on third-party sites through negotiation.
If you already have an SEO agency, we work in the same loop: the analysis and the priorities are ours, the hands can be theirs or ours. What exactly we change — on the activation page.
What each team gets
- CMO — AI Share of Voice — a metric none of your board colleagues have yet. Brand share in the new channel, quarterly dynamics, position against competitors: a ready slide about leadership where the market is only starting to look.
- PR and communications — Every notable publication now works twice — for the audience and for the models that will read it. aidentme shows which materials neural networks take into their answers, turning a media plan into an instrument of influence.
- Research and insights — Raw model answers are a new corpus of category perception data. How queries are worded, which brand attributes surface first, who you are grouped with — all saved, labeled and verified by an analyst with an evidence status.
- SEO and content — Search optimization has a sequel. The sources models rely on are identifiable — which means content can influence them. The platform delivers the list for your category ready-made, with priorities.
Metrics
- Share of Voice — brand share of mentions among competitors. Calculated over repeated runs: models answer differently each time, a single measurement means nothing.
- Rank Position — place in the recommendation. Observation: recorded for every saved answer.
- Sentiment — tone of mentions. Automatic labeling makes mistakes, so the final call is made by an analyst.
- Alice and GigaChat coverage — a separate measurement axis: Western platforms do not see these models.
All platform metrics — on a dedicated page.
How we measure
Purchase scenarios → five neural networks with repeated runs → text and source parsing → an analyst reads raw answers and labels every finding: observation, calculation or hypothesis. Hypotheses never make it into recommendations. Details — methodology.
How we work
Diagnostic study
One-off project: category and competitive field, several markets and AI systems, raw answers as evidence, an executive presentation.
Pilot
Working configuration: 120-180 scenarios, 6-10 brands, 4-6 AI systems, two markets or languages, repeated measurements. Ranges adapt to the task.
Subscription tracking
Regular measurement waves, alerts on factual errors and brand disappearance, an updated category benchmark, a quarterly report.
Enterprise program
Multiple brands, categories and countries, custom taxonomy, audience research integration, SSO, roles and managed export.
We measure the markets where your brand competes: the US, UK and Europe via ChatGPT, Claude and Gemini; Russia and Kazakhstan additionally via Alice and GigaChat — all in the same dashboard.
FAQ
What is aidentme?
aidentme measures how generative AI assistants — ChatGPT, Claude, Gemini, Yandex Alice and GigaChat — present your brand: whether they mention it, what they say about it and which brands they recommend in your category.
Which AI systems do you check?
ChatGPT, Claude, Gemini, Yandex Alice and GigaChat — across buyer choice scenarios, with repeated runs and expert review of the answers.
How is this different from SEO?
SEO is about positions in search results; aidentme is about what the AI assistant says about the brand in its answer. If you already have an SEO team, we work in the same loop.
What do I get?
Measurement — a map of sources: which pages shape the AI answer about your category and what exactly they say. Implementation — edits on your own and third-party sites, and a re-measurement with the same questions.
Do I need permission for the brand to appear in AI answers?
On your own site — no: it is a source for the AI assistant just like an aggregator or a media article. On third-party sites — through outreach.
How the demo works
Before the meeting we collect data on your category — the demo runs in the platform’s working dashboard, on real AI answers about your market, not on slides. In an hour you see the full picture: brand shares, sentiment and the sources models build their opinion from. After the demo you receive a pilot plan with a clear budget range and the date of the first report — you decide on numbers, not promises.
Request a demo
Category, 6-10 brands, five AI systems. We will come back with a demo plan and a sample report.
Request received. We will come back with a study plan and a sample report.
Data controller — Disconto LLC (Moscow).