“At first, we just wanted to see how many customers ChatGPT could bring in.” Ukrainian startup agentMention is developing an AI search product: how it works

— What is agentMention?

AgentMention is a Ukrainian AI Search analytics platform that helps companies track their presence in ChatGPT and other LLM responses.

I have been working on digital products for over ten years. Mostly e-commerce and SaaS, launching both new products and developing existing services.

Now, together with the CTO, I am developing agentMention. We work from Ukraine, but from the beginning we focused on the international market. The AI ​​Search niche is still quite young, so we did not want to limit ourselves to Ukrainian clients.

— How did the idea to create AgentMention come about? What inspired it?

We discussed what everyday tasks can already be solved via ChatGPT and how this can change behavior on the Internet.

At that time, there was almost no talk about agentic commerce or shopping through AI assistants. But it was already interesting what would happen if people started not searching for products on Google, but immediately asking ChatGPT to recommend something.

I thought it would be interesting to experiment with this channel in practice. How much traffic can I get from ChatGPT? How will it convert? I wrote the idea down in my notes. As often happens, current tasks turned out to be more important, and it fell out of my mind for about three months.

I returned to it when I came across a16z’s research on the formation of a new category of tools — GEO (Generative Engine Optimization). Then it became clear that a separate market was already starting to form around this.

The most interesting thing is that the market practically did not exist yet. Some companies that have already attracted over $150 million in investments today did not even have a full-fledged product at that time. Others managed to launch and close.

That was probably what struck me the most. It was clear that everyone was just trying to figure out what a product in this category should be like. And that there was still an opportunity to build something big and new.

We eventually ended up in the Mission Possible startup program, which helped us better articulate what exactly we were building, for whom, and what problem we were solving.

— How many people are in the team, what are their specialties and technological competencies?

So far, we are developing the product together. I am responsible for the product: I research the market, communicate with customers, determine what we will develop next, and I deal with sales and marketing.

My co-founder is our CTO, responsible for all the technical aspects. This includes integrations with AI models, the backend, data storage and processing, analytical algorithms, the platform interface, and its infrastructure.
Although in practice, when there are two of you, the division of responsibilities is quite arbitrary. Today you can discuss your client’s marketing strategy, tomorrow you can design a new dashboard, and the day after tomorrow you can figure out why the invoice payment was not received correctly.

— How much was invested in creating the product?

We are developing agentMention with our own funds. We have not attracted any external investments yet.

The main costs are customer acquisition, development, infrastructure, and working with AI models. This is a fairly significant item for our product, as we regularly send a large number of requests to various models, receive responses, and analyze them.

Therefore, one of the constant technical challenges is to reduce the cost of data collection. We experiment with different approaches to processing information and using models so that costs do not increase proportionally with each new client.

Thanks to Mission Possible and USF, we were also able to obtain cloud credits from AWS. This is a tangible help for us, as it allows us to direct some of our resources towards product development.

— How does AgentMention work?

Let’s imagine that a person wants to buy a laptop, they can ask ChatGPT: "Where is the best place to buy a laptop in Kyiv?" and in response will receive several specific stores. And for businesses, the question arises: why does AI recommend these companies and not others?

Actually, that’s what we’re working on.

agentMention now allows you to conduct research in ChatGPT, Gemini, Claude, Perplexity, and Microsoft Copilot. A separate direction that was recently launched is Google AI Overviews.

The first stage is a client brand analysis. The client adds a website, company name, and selects the market they want to analyze. We research what the company does and what queries potential buyers might be looking for.

The second stage is collecting responses. We run the generated prompts in selected AI systems, save the responses, and analyze which brands appear there. We do this daily to see the dynamics.

The third stage is analytics. The client receives several key indicators:

• Brand Visibility — how often AI mentions or recommends a brand in responses to relevant queries.
• Share of Voice — what share of mentions the company receives compared to competitors.
• Average Position — in which positions the brand usually appears in recommendation lists.
• Sentiment — in what context does AI describe a company, what it says about its strengths and weaknesses.

We also collect sources that the AI ​​cites in its responses. These can be company websites, media, reviews, ratings, or industry resources.

The fourth stage is recommendations for improvement. Based on the information collected, we show what needs to be worked on: what content to create or update, which pages to refine, where to strengthen the brand presence, and whether there are technical issues that prevent AI systems from receiving information from the site.

— How many users, clients and customers do you already have?

As of October 2026, we have 20 paying customers.

We work mainly with the B2B segment. Our main users are CMOs, marketing managers, SEO specialists, and marketing agencies.

We now have experience working with quite different industries: e-commerce, logistics, EdTech, SaaS, Robotics.

We also conduct our own industry research. For example, we recently prepared the AgentMention EdTech AI Visibility Index. For it, we analyzed responses to 1,000 unbranded queries about the Ukrainian education market.

We were interested to see which educational brands ChatGPT recommends most often, who leads in particular categories, and what sources are used to form these recommendations.

— What is the startup’s monetization model?

We have a classic SaaS model with a monthly or annual subscription. The cost depends primarily on the amount of data the client needs to analyze: the number of prompts, AI models, brands.

We currently have three main tariffs.

Starter — $49 per month. This is a basic plan for businesses looking to start tracking their presence on ChatGPT. It includes 50 prompts and up to 1,500 responses per month.

Growth — $99 per month. Designed for marketing teams that need more data. The plan includes 100 prompts, up to 9,000 responses per month, and work with ChatGPT, Gemini, and Perplexity.

Enterprise — from $399 per month. Here we individually select the configuration to meet the company’s needs: the number of markets, brands, models, and volume of requests.

We work separately with agencies and companies that need help not only with data collection, but also with analyzing the results and implementing recommendations.

To get acquainted with the platform, we offer the opportunity to test 50 prompts for free.

— What are the payback periods and financial model?

For now, we are moving according to the bootstrapping model: we finance development ourselves and gradually increase regular revenue.

We see good customer retention. If a company goes through a pilot and starts using data in their work, they tend to stay with us for the long term.

Now our main task is to learn how to consistently attract new clients and make this process predictable.

Ukrainian EdTech startup Promova returns to Boosters. The company’s CEO is leaving his position

On the topic

Ukrainian EdTech startup Promova returns to Boosters. The company’s CEO is leaving his position

Read the country’s main IT news in our Telegram

On the topic

Read the country’s main IT news in our Telegram