situation
Sound familiar?
- people no longer ask Google about you but ChatGPT — and you don’t know what it answers
- AI confuses you with a namesake, quotes outdated negative content or confidently makes up facts
- partners and journalists increasingly “check through AI” — and no one controls this field
- your Google search results are in order, but AI answers are assembled from other sources — and the picture is different
Search is changing before our eyes: more and more checks begin and end with an AI answer.
complexity
Where AI gets information about you
AI assistants build an answer from two layers: training data (what the model has “memorised” — including materials removed long ago) and live search (what it finds at the moment of the question).
The first layer cannot be managed directly; the second can, and this is a new discipline: GEO, generative engine optimization. Its mechanics differ from classic SEO: models cite not the “top positions” but sources with clear facts that are consistent with one another.
GEO (generative engine optimization) is the optimisation of presence in the answers of AI systems: working with the sources that ChatGPT, Gemini and Perplexity use when forming answers. Unlike SEO, GEO focuses on how often sources are cited and how consistent their facts are, not on positions in search results.
honestly
What is possible in this field — and what is not
There is no direct “fix ChatGPT button”, and anyone selling one is selling thin air.
The real lever is the sources from which models assemble their answers: they can be corrected, strengthened and created. This does not work instantly or absolutely — but it is the only honest mechanism, and the earlier you start, the better your position will be when AI checks become the norm.
process
How we work
01
AI audit
We ask the main models about you — in different languages and with different wording. We record what they answer, which sources they use, and where the errors and fabrications are.
02
Source map
We determine where the models get facts about you — and which sources need to be corrected, strengthened or created.
03
Field correction
We bring the sources into consistency: the same facts everywhere, clear wording, authoritative platforms. It is the same systematic work with the information field — under the new rules of the reader.
04
Control
We repeat the audit regularly: models are updated, answers change — the picture has to be maintained.
formats
Both formats are customized. The cost depends on the complexity of the search results, the volume of negative content, and the target outcome.
Project
A defined project with a specific outcome: changing search results and resolving a particular issue. Timeline: from a few months.
Ongoing support
Reputation as a managed asset: monitoring, responding, and developing your information landscape. For those whose name works for them every day.
agency
Why SERM.UA
We have been building information fields since 2011 — AI has simply become one more of their readers, the most demanding when it comes to facts.
Our own website is written according to GEO rules: we apply to ourselves what we offer you. More than 500 projects.
question
Can information be removed from ChatGPT?
From training data — practically not. From answers — indirectly: by correcting the sources the model uses in search. Some providers have procedures for personal data — we use those too.
AI made up a fact about me. What should I do?
Record it (screenshots, the wording of the queries), check the sources of the answer and work with them. Hallucinations occur more often where credible sources are scarce — AI fills an empty field on its own.
How is GEO different from SEO?
How is GEO different from SEO?
Is this relevant already or “for the future”?
Already: the share of checks done through AI grows every month. The advantage is that the field is still empty — whoever builds a presence now will become the default answer.
