1. The study in brief
This is the first edition of the AgentMention EdTech AI Visibility Index — a study of how ChatGPT answers Ukrainians’ questions about learning. We asked ChatGPT 1,000 unbranded questions across 10 education categories, from “which Python courses are worth taking” to “where to prepare for the NMT in math”, and analysed which brands appear in the answers, in what context they are mentioned and which sources are used.
| Categories | 10 |
| AI assistant | ChatGPT |
| Prompts | 1,000, 100 per category |
| Answers analysed | 30,000 answers |
| Brands in AI answers | 764 unique brands (one brand could appear in several categories) |
| Data collection period | August 2026 |
| Prompt language | Ukrainian |
| Region | Ukraine |
The full methodology and metric definitions are at the end of the document.
An analysis of Ukraine’s EdTech ecosystem: key market segments, the main players and their visibility in ChatGPT. The study helps build a complete picture of the market structure and identify the most attractive niches for growth.
2. Overview and AI search trends
Key findings
What changed in search over the past year
Education is a category where people don’t buy on impulse. They ask friends, read reviews, compare competitors’ websites and only then decide. More and more often these questions go to AI assistants: instead of a dozen tabs, people get one structured answer at once — and, surprisingly, often trust that choice more.
A few reference points that set the context for the study:
- Audience scale. In July 2026 ChatGPT passed 1 billion weekly active users — an OpenAI spokesperson told The Verge, and the company publicly confirmed the figure on August 6. For comparison, in February 2026 there were about 900 million weekly actives, so the last 100 million took almost five months to add.
- AI traffic in Ukraine is not yet a major channel, but its contribution is already noticeable. According to Similarweb for August 2026, generative assistants bring Ukrainian education brands between one and six percent of their traffic, with a median of about 2%. This traffic is almost monolithic: ChatGPT’s share reaches about 90%.
Why this matters for EdTech in particular
In education queries, assistants almost always explain why they recommend a particular brand. Being called “expensive, but with a job guarantee” and “cheap, but with no practice” are two different market positions, even with the same number of mentions.
How assistants choose whom to recommend
To read the tables that follow correctly, it helps to understand the mechanism. In an assistant’s answer to “which Python courses are worth taking”, a brand appears in one of three ways.
First — the model’s memory. During training the model saw thousands of texts where the brand was mentioned in the context of Python courses. The more often and consistently the brand appeared in that context, the more likely the model is to name it without any search. The main problem: changes reach the model with a delay of months.
Second — live search. ChatGPT with search enabled queries a search engine, reads a few pages and synthesises an answer. Different factors work here: whether the page made it into the top results, whether an answer is easy to extract from it, whether it has facts that can be quoted. This path reacts to changes within days.
Third — synthesis from rankings. The assistant finds a “top 10 coding schools in Ukraine” page on a media site or in a catalog and retells it. One such article can shape the answers to dozens of prompts. That is why in section 5 we look separately at which domains are these “ranking makers”.
In practice all three paths mix, which is why the study measures not only “who was named” but also “what was cited” and “how ready the site is to be quoted”.
3. Category leaders map
For each of the 10 categories we give the type of queries, the list of tracked brands, a table with four key metrics — Brand Visibility, SoV, Position, Sentiment — and a short interpretation.
Brand Visibility and Share of Voice can diverge. A brand mentioned in 60% of answers alongside 10 other brands has a lower SoV than a brand mentioned in 40% of answers where it is the only one. The first metric shows how “top of mind” a brand is, the second how much it dominates.
Summary map of leaders by Brand Visibility
| # | Category | Leader | Second | Third | Competition type |
|---|---|---|---|---|---|
| 1 | IT & digital professions | Mate academy | GoIT | Hillel IT School | clear leaders |
| 2 | Business, marketing & management | Prometheus | Laba | Projector | clear leaders |
| 3 | Creative & design professions | Projector | SKVOT | Creative Practice | fragmented |
| 4 | Foreign languages | Preply | Englishdom | Green Forest | fragmented |
| 5 | Business education & MBA | KMBS | MIM | LvBS (УКУ) | clear leaders |
| 6 | Learning apps & microlearning | Brilliant | Duolingo | Khan Academy | fragmented |
| 7 | Corporate learning | EY Academy Ukraine | LvBS (УКУ) | Prometheus | hyper-fragmented |
| 8 | K-12 online schools | Оптіма | ThinkGlobal | Атмосферна школа | fragmented |
| 9 | NMT exam prep | iLearn | УЦОЯО | Prometheus | concentrated |
| 10 | Tutors & tutoring | BUKI | Preply | Superprof | concentrated |
We define the competition type by the combined SoV of the top three among the category’s 10 tracked brands: ≥80% — concentrated, 65–80% — clear leaders, 55–65% — fragmented, under 55% — hyper-fragmented.
3.1. IT & digital professions
What was asked. Queries from people who want to change careers: “which Python courses for beginners in Ukraine”, “where to learn QA from scratch with job placement”, “Data Analyst courses in Ukrainian”, “which IT school is best for frontend”. A separate cluster is queries about paying after you get a job.
Brands in the answers. Within the “IT & digital professions” topic, AI mentioned 76 Ukrainian and international education brands — from the big IT schools Mate academy, GoIT and Hillel to niche professional schools, universities, corporate academies and global platforms such as Coursera and Codecademy.
| # | Brand | Brand Visibility | SoV | Position | Sentiment |
|---|---|---|---|---|---|
| 1 | Mate academy | 77.4% | 35.0% | 1.7 | 61 |
| 2 | GoIT | 77.2% | 20.2% | 2.4 | 62 |
| 3 | Hillel IT School | 66.9% | 19.3% | 3.0 | 60 |
| 4 | robot_dreams | 23.0% | 3.8% | 4.0 | 60 |
| 5 | Prometheus | 22.1% | 4.4% | 3.3 | 68 |
| 6 | Projector | 21.1% | 5.6% | 2.4 | 59 |
| 7 | Beetroot Academy | 19.9% | 6.0% | 4.0 | 63 |
| 8 | Laba | 14.7% | 2.3% | 3.2 | 59 |
| 9 | IT STEP Academy | 12.4% | 2.0% | 3.5 | 57 |
| 10 | DAN.IT | 9.1% | 1.6% | 4.9 | 59 |
AI Search has already settled on the top three of Ukrainian IT education — Mate academy, GoIT and Hillel — with Mate most often in first position. For ChatGPT the main criteria for choosing a school are practice, price and job-placement help, not brand strength or community. At the same time recommendations are not built only from the schools’ own websites: catalogs, rankings, reviews and testimonials play a significant role.
3.2. Business, marketing & management
What was asked. “Online marketing courses in Ukraine”, “where to train as a product manager”, “SMM courses for beginners”, “online project management training”, “B2B sales courses”.
Brands in the answers. Within the topic, AI mentioned 70 education brands and institutions — from Prometheus, Laba, Projector, Genius.Space and WebPromoExperts to KMBS, KSE, LvBS, Coursera, HubSpot Academy and CXL.
| # | Brand | Brand Visibility | SoV | Position | Sentiment |
|---|---|---|---|---|---|
| 1 | Prometheus | 59.2% | 26.8% | 2.4 | 62 |
| 2 | Laba | 45.6% | 19.3% | 2.7 | 58 |
| 3 | Projector | 41.9% | 18.9% | 2.9 | 58 |
| 4 | Genius.Space | 22.7% | 5.5% | 5.3 | 72 |
| 5 | WebPromoExperts | 14.7% | 5.2% | 4.0 | 60 |
| 6 | SKVOT | 14.5% | 5.6% | 4.2 | 60 |
| 7 | KMBS | 14.0% | 7.3% | 2.6 | 55 |
| 8 | Klekit | 12.9% | 2.4% | 6.0 | 57 |
| 9 | KSE | 11.3% | 6.4% | 2.7 | 54 |
| 10 | EdEra | 7.0% | 2.5% | 6.6 | 59 |
Business education already has its own “big three” — Prometheus, Laba and Projector, which together take 65% SoV. Beyond them the market is scattered. When recommending courses, ChatGPT most often stresses strategic and leadership skills (88.4%), cost (76.3%) and practical learning (74.7%), and in almost half of cases (45.3%) that the program suits entrepreneurs. Study format (12.6%) and networking (6.0%) become a reason to recommend far less often. Sources are more spread out: besides the schools’ own websites, LinkedIn, education program catalogs and specialised marketing media play a noticeable role.
3.3. Creative & design professions
What was asked. “UX/UI design courses in Ukraine”, “where to learn graphic design from scratch”, “motion design courses”, “online illustration school”, “copywriting courses”.
Brands in the answers. Within the topic, AI mentioned 104 education brands, schools and universities — from Projector, SKVOT, KAMA and Creative Practice to Domestika, School of Motion, Parsons, RISD, Royal College of Art and CalArts.
| # | Brand | Brand Visibility | SoV | Position | Sentiment |
|---|---|---|---|---|---|
| 1 | Projector | 47.5% | 32.3% | 1.8 | 63 |
| 2 | SKVOT | 30.8% | 19.0% | 2.9 | 61 |
| 3 | Creative Practice | 22.6% | 12.1% | 3.3 | 75 |
| 4 | KAMA | 19.5% | 12.3% | 2.5 | 67 |
| 5 | Domestika | 11.5% | 7.6% | 3.4 | 61 |
| 6 | Maslo Motion | 8.1% | 4.4% | 3.6 | 63 |
| 7 | School of Visual Communication | 7.9% | 4.3% | 3.8 | 64 |
| 8 | Mate academy | 7.3% | 6.6% | 2.8 | 53 |
| 9 | Дія.Освіта | 3.2% | 0.9% | 4.8 | 70 |
| 10 | Klekit | 2.9% | 0.5% | 5.2 | 58 |
In creative education Projector has a noticeable lead: 32.3% SoV versus 19.0% for SKVOT, followed closely by Creative Practice and KAMA. The key reasons to recommend are practice (84.3%), portfolio (79.7%) and working with professional tools (72.6%). A feature of the category is a strong global context: almost a third of citations (28.49%) lead to international platforms such as Domestika, and another 9.47% to foreign SEO/affiliate reviews. Ukrainian schools here compete not only with each other but with international courses too.
3.4. Learning foreign languages
What was asked. “Best online English schools in Ukraine”, “where to learn English for IT”, “German courses online”, “IELTS preparation in Ukraine”, “an app for learning English”.
Brands in the answers. Within the topic, AI mentioned 57 language schools, platforms and learning apps — from Englishdom, Green Forest, Preply, JustSchool and Promova to British Council, Lingoda, Cambly, Duolingo and Busuu.
| # | Brand | Brand Visibility | SoV | Position | Sentiment |
|---|---|---|---|---|---|
| 1 | Preply | 50.1% | 19.8% | 2.5 | 71 |
| 2 | Englishdom | 47.4% | 21.4% | 1.8 | 62 |
| 3 | Green Forest | 42.0% | 17.3% | 2.0 | 68 |
| 4 | British Council | 27.2% | 10.8% | 2.2 | 64 |
| 5 | italki | 22.4% | 6.5% | 2.7 | 66 |
| 6 | Cambly | 19.7% | 5.4% | 3.2 | 63 |
| 7 | JustSchool | 17.0% | 6.0% | 3.2 | 71 |
| 8 | Project 12 | 15.2% | 4.1% | 3.5 | 64 |
| 9 | Lingoda | 14.1% | 4.1% | 3.5 | 57 |
| 10 | Promova | 9.3% | 4.6% | 2.0 | 75 |
In language learning ChatGPT puts schools, teacher marketplaces and apps in the same competitive line-up. Next to Englishdom, Green Forest and Project 12, Preply, British Council, Cambly, Duolingo and Promova appear regularly. As a result, local schools no longer compete only with each other.
3.5. Business education & MBA
What was asked. “MBA programs in Ukraine”, “best business schools in Ukraine”, “Executive MBA Kyiv”, “courses for business owners”, “online MBA in Ukrainian”.
Brands in the answers. Within the topic, AI mentioned 53 business schools, universities and executive-education brands — from KMBS, MIM, KSE Graduate Business School and LvBS to INSEAD, London Business School, Harvard Business School, IMD, HEC Paris and Wharton.
| # | Brand | Brand Visibility | SoV | Position | Sentiment |
|---|---|---|---|---|---|
| 1 | KMBS | 64.0% | 28.6% | 1.9 | 54 |
| 2 | MIM | 52.4% | 18.6% | 3.2 | 56 |
| 3 | LvBS (УКУ) | 51.5% | 19.7% | 4.0 | 58 |
| 4 | KSE | 50.6% | 20.5% | 3.2 | 55 |
| 5 | INSEAD | 21.8% | 6.9% | 2.1 | 66 |
| 6 | London Business School | 14.7% | 1.4% | 5.3 | 70 |
| 7 | Harvard HBS | 12.6% | 1.5% | 4.5 | 67 |
| 8 | МІБ | 9.7% | 1.5% | 6.0 | 57 |
| 9 | Львівська школа керівників бізнесу | 5.5% | 0.6% | 5.5 | 64 |
| 10 | КРОК | 5.0% | 0.7% | 6.5 | 64 |
There is no single universal leader in MBA — ChatGPT separates schools by audience type and career goal. KMBS is more often linked to business owners and C-level, KSE to an international context and analytics, Бізнес-школа УКУ to entrepreneurship and leadership, and MIM to the classic MBA. Here the assistant recommends not just a “strong brand” but the program most relevant to a specific user profile.
3.6. Learning apps & microlearning
What was asked. “Learning apps in Ukrainian”, “best self-development apps”, “an app for learning to code on your phone”, “short 10-minute-a-day courses”, “an app for NMT prep”.
Brands in the answers. Within the topic, AI mentioned 103 education brands and apps — from Brilliant, Duolingo, Khan Academy, Blinkist and Headway to Anki, Quizlet, Coursera, DataCamp, Mimo and TED-Ed.
| # | Brand | Brand Visibility | SoV | Position | Sentiment |
|---|---|---|---|---|---|
| 1 | Brilliant | 62.9% | 22.9% | 2.4 | 81 |
| 2 | Duolingo | 51.0% | 17.7% | 2.2 | 63 |
| 3 | Khan Academy | 48.1% | 15.5% | 2.9 | 68 |
| 4 | Blinkist | 41.5% | 14.4% | 3.2 | 61 |
| 5 | Headway | 29.7% | 10.1% | 3.1 | 59 |
| 6 | Kinnu | 17.9% | 4.8% | 3.8 | 67 |
| 7 | Anki | 17.3% | 6.1% | 3.9 | 71 |
| 8 | Shortform | 12.4% | 5.3% | 2.6 | 58 |
| 9 | Readwise Reader | 8.0% | 2.4% | 2.2 | 83 |
| 10 | Scroll | 2.9% | 0.6% | 3.9 | 69 |
Competition in this category is global from the start: next to Headway, recommendations feature Brilliant, Blinkist, Duolingo, Shortform and other international services. The choice depends heavily on the scenario — a daily habit, book summaries or building a specific skill. Comparison reviews, app rankings and the App Store/Google Play play a noticeable role in shaping recommendations.
3.7. Corporate learning
What was asked. Queries on behalf of HR and L&D: “corporate learning platforms in Ukraine”, “an LMS for a 200-person company”, “a sales training provider for a team”, “corporate English for employees”.
Brands in the answers. Within the topic, AI mentioned 162 corporate learning providers, business schools and learning platforms — from EY, Prometheus, Golden Staff, Sigma Software University and Laba to Korn Ferry, CCL, INSEAD, Pluralsight, Coursera for Business and Udemy Business.
| # | Brand | Brand Visibility | SoV | Position | Sentiment |
|---|---|---|---|---|---|
| 1 | EY Academy Ukraine | 22.5% | 20.2% | 2.4 | 58 |
| 2 | LvBS (УКУ) | 14.0% | 15.4% | 3.1 | 58 |
| 3 | Prometheus | 13.8% | 12.8% | 2.4 | 66 |
| 4 | Golden Staff | 13.4% | 8.1% | 2.7 | 84 |
| 5 | KMBS | 11.6% | 10.0% | 2.8 | 57 |
| 6 | Sigma Software University | 11.6% | 6.2% | 2.2 | 72 |
| 7 | Laba | 11.3% | 10.5% | 2.8 | 64 |
| 8 | CCL | 8.7% | 6.9% | 2.5 | 68 |
| 9 | KSE | 8.0% | 6.5% | 3.1 | 55 |
| 10 | Prog Academy | 6.4% | 3.5% | 3.5 | 59 |
EY leads in queries about systematic development of managers and teams — soft skills, leadership and building corporate learning programs. Prometheus and Laba appear more often when the topic is large-scale employee training, KMBS and LvBS when it is executive development. In this category what decides is not only a brand’s visibility but how clearly ChatGPT links it to a specific business task.
3.8. K-12 online schools
What was asked. Parents’ queries: “an online school licensed by the Ministry of Education”, “a distance school for a child abroad”, “best private online schools in Ukraine”, “online school grade 5 reviews”, “how much does a distance school cost”.
Brands in the answers. Within the topic, AI mentioned 105 online schools and education platforms — from «Оптіма», ThinkGlobal, «Атмосферна школа», Liko Education and Alterra School to Stanford Online High School, Pearson Online Academy, K12, Crimson Global Academy and King’s InterHigh.
| # | Brand | Brand Visibility | SoV | Position | Sentiment |
|---|---|---|---|---|---|
| 1 | Оптіма | 67.6% | 28.3% | 1.8 | 71 |
| 2 | ThinkGlobal | 54.4% | 23.2% | 2.8 | 60 |
| 3 | Атмосферна школа | 38.9% | 10.2% | 3.1 | 65 |
| 4 | Liko Education | 29.1% | 16.3% | 3.8 | 61 |
| 5 | Alterra School | 16.5% | 3.4% | 3.3 | 58 |
| 6 | Main School | 15.0% | 4.4% | 3.8 | 62 |
| 7 | Міжнародна українська школа | 14.4% | 4.4% | 2.6 | 59 |
| 8 | Українська дистанційна школа | 13.0% | 4.2% | 3.1 | 56 |
| 9 | Unicorn School | 13.0% | 3.5% | 3.7 | 65 |
| 10 | Академія «А+» | 8.0% | 1.9% | 4.5 | 63 |
Government sources carry unusually high weight. In its answers ChatGPT regularly turns to the Ministry of Education, and uses Education.ua rankings to compare schools, including by NMT results. A school’s own website is not enough here — what matters is how the brand is represented in external education sources.
3.9. NMT exam prep
What was asked. “NMT math prep courses”, “free online NMT prep”, “a tutor or courses for the NMT”, “an app for NMT prep in Ukrainian”, “NMT history prep in 3 months”.
Brands in the answers. Within the topic, AI mentioned 88 education brands and NMT prep platforms — from iLearn, ZNOHUB, Prometheus, «ЯвКурсі» and ZNO.UA to BUKI School, Optima Academy, Mathema, JustSmart and GoITeens.
| # | Brand | Brand Visibility | SoV | Position | Sentiment |
|---|---|---|---|---|---|
| 1 | iLearn | 69.6% | 38.8% | 1.6 | 67 |
| 2 | УЦОЯО | 65.8% | 28.4% | 2.5 | 57 |
| 3 | Prometheus | 32.3% | 14.3% | 2.6 | 63 |
| 4 | ZNO.ua | 12.2% | 4.2% | 2.2 | 60 |
| 5 | ZNO.org.ua | 9.8% | 3.3% | 2.2 | 56 |
| 6 | BUKI | 8.7% | 3.7% | 2.8 | 59 |
| 7 | NMT Academy | 6.5% | 2.5% | 2.3 | 57 |
| 8 | Оптіма | 5.8% | 1.8% | 2.9 | 67 |
| 9 | KEVIN | 5.1% | 2.1% | 2.0 | 60 |
| 10 | ZNOHUB | 3.4% | 0.8% | 1.8 | 50 |
Unlike many other EdTech categories, authority here is built around officialness, up-to-date materials and practice specifically in the NMT format. Paid courses win not on content but on their ability to turn preparation into a guided path — with a teacher, progress tracking and personal support.
3.10. Tutors & tutoring
What was asked. “Where to find an online English tutor”, “a platform for finding tutors in Ukraine”, “a math tutor for grade 8”, “how much does a physics tutor cost”, “a service that matches a teacher to a child”.
Brands in the answers. Within the topic, AI mentioned 78 platforms, schools and services for tutoring — from BUKI, Preply and Superprof to Mathema, Novakid, Outschool, Wyzant, JustSchool and ThinkGlobal.
| # | Brand | Brand Visibility | SoV | Position | Sentiment |
|---|---|---|---|---|---|
| 1 | BUKI | 72.8% | 46.6% | 1.2 | 60 |
| 2 | Preply | 64.6% | 32.3% | 2.0 | 72 |
| 3 | Superprof | 26.2% | 9.4% | 2.8 | 68 |
| 4 | italki | 9.0% | 4.0% | 2.7 | 73 |
| 5 | Kabanchik | 6.9% | 3.2% | 3.3 | 66 |
| 6 | Learning.ua | 4.1% | 1.1% | 2.2 | 52 |
| 7 | Mathema | 3.3% | 1.3% | 2.8 | 65 |
| 8 | Wyzant | 3.3% | 1.0% | 2.6 | 81 |
| 9 | Optima Academy | 3.1% | 0.8% | 1.8 | 59 |
| 10 | Tutor.ua | 1.1% | 0.3% | 3.4 | 50 |
The market is effectively split between two scenarios. BUKI is stronger in school subjects, the NMT and local Ukrainian demand, while Preply more often comes out ahead in languages and international online learning. So the competition between them is not entirely head-on — each brand has already “claimed” its own share of user intent.
4. Positions in Sentiment & Attribute
An assistant almost never names an education brand “just like that”. It adds a characteristic: “suitable for beginners”, “expensive, but with a job guarantee”, “there are complaints about support”. These characteristics are the brand’s position in the eyes of AI — and they decide whether a mention turns into a click.
We classified every brand mention along two axes: sentiment (positive, neutral, negative) and attribute — the specific characteristic the assistant ascribes to the brand.
Base attribute taxonomy
| Attribute | What it means in an answer | Example wording |
|---|---|---|
| Job placement | job guarantee or job-search help | “pay after you get a job” |
| Hands-on practice | projects, real cases | “learning on real projects” |
| Price | expensive / affordable / free | “one of the most expensive schools” |
| Mentors | expertise of the teaching staff | “taught by industry practitioners” |
| Beginner-friendly | low entry barrier | “suitable for those starting from zero” |
| Flexibility | format, pace, schedule | “you can study at your own pace” |
| Certificate | recognition of the result | “a state-recognised document” |
| License | official status | “licensed by the Ministry of Education”, “AMBA accreditation” |
| Community | community, networking | “a strong alumni community” |
| Support | service, curators | “there is a curator who answers questions” |
| Reputation | references to ratings | “highly rated on DOU” |
| Scale | size and age of the school | “one of the oldest schools in Ukraine” |
This is the base taxonomy shared across the market. For each category we adapted it to the 10 attributes that best describe queries in that segment.
Sentiment by category
| Category | Positive | Neutral | Negative | Main attribute | Main criticism |
|---|---|---|---|---|---|
| IT & digital professions | 42.6% | 53.7% | 3.7% | Hands-on practice | High study intensity |
| Business, marketing & management | 31.3% | 68.7% | 0.0% | Strategy & leadership | — |
| Creative & design professions | 26.4% | 69.4% | 4.2% | Practice & projects | High tuition cost |
| Foreign languages | 44.3% | 47.6% | 8.1% | Price | Uneven teacher quality |
| Business education & MBA | 33.2% | 66.8% | 0.0% | For executives & owners | — |
| Learning apps & microlearning | 31.9% | 66.1% | 2.0% | Short format | Shallow condensed content |
| Corporate learning | 28.9% | 69.5% | 1.6% | Practice & business cases | High cost and hard-to-customize programs |
| K-12 online schools | 28.8% | 70.4% | 0.8% | Flexibility & format | Requires self-discipline |
| NMT exam prep | 31.5% | 65.9% | 2.6% | Tests & drills | Requires self-discipline |
| Tutors & tutoring | 36.3% | 60.7% | 3.0% | Price & budget | Uneven teacher quality |
Leaders’ attribute map
For each category leader — the three attributes assistants ascribe to it most often. This is the brand’s “AI positioning”: how it is described when the brand itself is not in the room.
| Category | Leader | Attribute 1 | Attribute 2 | Attribute 3 |
|---|---|---|---|---|
| IT & digital professions | Mate academy | Job placement & career | Practice & projects | Start from zero |
| Business, marketing & management | Prometheus | Accessible & free | Solid foundation & wide choice | Practical assignments |
| Creative & design professions | Projector | Practice & real projects | Portfolio | Career & job placement |
| Foreign languages | Preply | Speaking practice | One-on-one lessons | Wide choice of teachers |
| Business education & MBA | KMBS | For C-level & owners | Leadership & management | Strategic thinking |
| Learning apps & microlearning | Brilliant | Math, STEM & logic | Interactive problems | Short daily lessons |
| Corporate learning | EY Academy Ukraine | Systematic L&D | Leadership & management | Tailored to the business |
| K-12 online schools | Оптіма | Flexible distance format | License & documents | Academic quality |
| NMT exam prep | iLearn | Tests & drills | Free access | Webinars & video lessons |
| Tutors & tutoring | BUKI | Wide choice of tutors | Reviews & ratings | Easy matching & filters |
Three patterns worth looking for in the data
In 17.7% of answers assistants name a specific tuition price. This creates a separate risk for brands: price is one of the few attributes where even a small inaccuracy can directly affect the user’s decision. An understated price causes disappointment at first contact, an overstated one can screen out a potential customer while they are still in the chat.
5. Sources of influence
ChatGPT can form an answer in several ways: rely on knowledge acquired during model training, use web search, or combine both. So a brand can appear in an answer even without any external sources being consulted.
When an answer is formed using web search, ChatGPT generates search queries, finds relevant pages, analyses several sources and synthesises a final answer from them. It is in this scenario that a brand’s presence in authoritative external sources directly affects whether it makes it into the answer.
“Top 20 external domains by number of citations”
| # | Domain | Share | Categories | Type | Cited most by |
|---|---|---|---|---|---|
| 1 | mon.gov.ua | 2.64% | 10/10 | Government body | K-12 online schools |
| 2 | education.ua | 2.52% | 9/10 | Catalog / ranking | K-12 online schools |
| 3 | ithub.ua | 0.69% | 6/10 | Course catalog | Creative & design |
| 4 | vidhuk.ua | 0.58% | 8/10 | Reviews | Foreign languages |
| 5 | diia.gov.ua | 0.51% | 9/10 | Government platform | Business / marketing |
| 6 | linkedin.com | 0.50% | 9/10 | Social network | Business / marketing |
| 7 | dou.ua | 0.44% | 7/10 | Media / community | IT & digital professions |
| 8 | apps.apple.com | 0.36% | 3/10 | App Store | Apps & microlearning |
| 9 | play.google.com | 0.32% | 6/10 | Google Play | Apps & microlearning |
| 10 | trustpilot.com | 0.27% | 5/10 | Reviews | Tutors |
| 11 | osvita.ua | 0.24% | 7/10 | Media / portal | NMT prep |
| 12 | learnenglish.life | 0.24% | 3/10 | Reviews / comparisons | Foreign languages |
| 13 | kmu.gov.ua | 0.24% | 8/10 | Government body | K-12 online schools |
| 14 | skillpublic.com | 0.23% | 3/10 | Catalog / ranking | Creative & design |
| 15 | t.me | 0.22% | 5/10 | Telegram | Business / marketing |
| 16 | rada.gov.ua | 0.19% | 6/10 | Government body | K-12 online schools |
| 17 | unian.ua | 0.16% | 2/10 | Media | K-12 online schools |
| 18 | top20.ua | 0.15% | 8/10 | Reviews / ranking | Creative & design |
| 19 | forbes.ua | 0.08% | 5/10 | Media | Tutors |
| 20 | it-kursy.com.ua | 0.07% | 2/10 | Catalog | IT & digital professions |
How ChatGPT works with sources
| Metric | ChatGPT |
|---|---|
| Average citations per answer | 23.5 |
| Share of answers with at least one citation | 99.7% |
| Share of brand websites among citations | 70% |
| Share of media | 4.2% |
| Share of UGC | 2.7% |
| Share of review platforms | 2.4% |
Takeaway. The top 20 domains list is a direct pointer to where your PR and content teams should look. If your brand is missing from a review on a domain cited in your category, a publication there can give your AI visibility a boost.
6. AEO Readiness
Answer Engine Optimization is preparing a website so that an assistant can quickly find, correctly understand and retell information about the brand without distortion.
Basic AEO steps
Typical mistakes on education websites
7. Methodology
Prompts. For each category we built a set of unbranded queries reflecting real search intents: looking for options, comparisons, clarifications, queries with constraints. The prompts were written in Ukrainian on behalf of a typical user of the category. In total — 1,000 prompts across all 10 categories.
Assistant. ChatGPT — the GPT-5.6 Luna model. Each prompt was run 30 times to smooth out answer variability.
Period. August 2026.
The study’s core metrics
Limitations. AI assistants’ answers are non-deterministic: the same query can produce somewhat different answers. To reduce the effect of this variability we used repeated runs. The list of tracked brands was built from ChatGPT’s answers: the main list includes the brands mentioned most often in answers within each category. The study reflects the state of AI answers on the date of data collection and is not a forecast.
Sentiment. An index from 0 to 100 showing the overall tone of a brand’s mentions in AI answers. The higher the score, the more positively the brand is described.
Data. Collected and processed on the AgentMention platform.
8. About AgentMention
AgentMention is a Ukrainian platform for monitoring brand visibility in AI assistants. We track how AI assistants mention brands, which sources they cite and how this picture changes over time, and turn that data into concrete actions for marketing teams.