Companies don’t need to know 900,000 AI models. They need to choose the right one. And that’s what we do, says Roman Berglowiec, Everbot

May 15, 2025

Companies, both large and small are beginning to feel lost in the flood of AI models, with new versions and tools popping up from all sides. It’s clear the market will eventually consolidate. Only the best will survive. Our guest is Roman Berglowiec, CEO of Everbot—a company that specializes in AI and offers an app that selects the most suitable model for your task based on your prompt.

Ivana Karhanová: How many AI tools are currently available on the market?

Roman Berglowiec: When you talk to people, it often feels like there’s only one–maybe ChatGPT–but in reality, there are over 900,000 language models out there. These appear across thousands of platforms and tools. Some are better at specific things likewriting, coding, design, and analytics. In Europe and especially here in the Czech Republic, we’re just beginning to explore this world. Most people stick with a few well-known tools, but the ecosystem is much broader and expanding rapidly. It’s already difficult to keep up–new tools are released daily–and that overload often paralyzes people, so they fall back on what’s familiar.

Ivana Karhanová: The tools you mentioned are quite general-purpose. But the 900,000 apps or bots you referenced earlier are much more specialized, right?

Roman Berglowiec: Exactly. Many of these models are trained on top of open-source foundations, like those from Meta. They’re then fine-tuned for very specific use cases: social media content, ad copy, data analysis, even writing high-quality articles. So, while there’s a vast number of tools, each one tends to focus on doing one particular task really well.

Ivana Karhanová: Your app, Everbot, is based on selecting the right model for each prompt. Why build something like that?

Roman Berglowiec: That’s the idea we’re actively developing and preparing to release around May or June. The concept is simple: you describe what you want to generate–whether it’s text, video, image, or code–and Everbot figures out the best model to do it. It takes the burden of knowing which model is best off the user. You don’t need to know whether Mistral or Claude is better for Instagram captions. The app handles that. We’re starting with a focus on marketing tasks, but developers, coders, and others are already experimenting with it.

Ivana Karhanová: You started using AI in your companies back in 2018. That makes you one of the early adopters here. What were you using it for then?

Roman Berglowiec: Back then, it was mostly about processing large amounts of data. We didn’t even call it AI at the time – we just needed automation to keep up with what we were doing. We were collecting data on competitors, tracking pricing strategies, analyzing customer segments, search trends… things like that. Looking back, it was a lot of manual effort, but those early automation techniques were absolutely critical.

Ivana Karhanová: So mostly marketing-related use cases?

Roman Berglowiec: Yes, definitely. We focused heavily on marketing and sales data. For us, those departments are often the quickest and most impactful areas to implement AI. That’s still why we see marketing as a natural starting point for adoption in most companies.

Ivana Karhanová: How did AI look in 2018 compared to today?

Roman Berglowiec: Back then, only the data scientists in our team could use it. Today, anyone can. You don’t need to be a tech expert. Tools are much more user-friendly now. Even someone like me who is not a data specialist can work with AI effectively.

Ivana Karhanová: What were some early lessons you learned from applying AI in business?

Roman Berglowiec: We realized early on that if you can work with data well, you can make much better decisions. Over time, we integrated AI into more and more parts of our business. At first, it was about small automation steps: visualizing data, writing simple scripts, using U.S.-based tools that weren’t available in Czech. But that was already helping us massively. Today, things we spent hours or days on back then can be done in minutes. It’s taught us that when you implement AI, you can grow faster with fewer people – without compromising on quality.

Ivana Karhanová: Let’s move forward to 2020 and your paint-by-numbers business. You used AI there too?

Roman Berglowiec: Yes, that was actually a direct result of using AI for opportunity analysis. During the early pandemic, we were running an e-commerce business and had to decide where to focus. We processed a massive amount of data to find fast-growing markets with high demand and low competition. One of the top hits was paint-by-numbers. AI helped identify that, and we decided to build the business ourselves.

Ivana Karhanová: What exactly did the AI analysis tell you?

Roman Berglowiec: It showed us that this market was exploding. We looked at product search trends, competitor growth across Europe, profitability potential… We even developed a framework to identify business opportunities. At first, we wanted to sell those insights to other companies – but we’re in the Czech Republic, where few are willing to pay for just ideas. So, we picked one idea and launched it ourselves.

Ivana Karhanová: So, the entire business idea, even the brand, came from AI?

Roman Berglowiec: Exactly. The name, the logo, the strategy–it all came from a mix of AI and human creativity. The project grew rapidly reachingnearly 100 million CZK in the first year. We expanded to 12 countries, partly thanks to Adastra, who helped us develop an AI tool that generates and adapts images into formats like paint-by-numbers, diamond painting, and dot art. That tool gave us a huge edge. We offered thousands of original designs, while others had maybe hundreds.

Ivana Karhanová: Today, converting images like that is quick, right?

Roman Berglowiec: Not quite. It’s still complex. The tool has to convert the image to our specific palette of 240–250 acrylic colors, preserve facial details, map the paint areas precisely… Even today, AI can’t fully automate it. We still need a graphic designer involved, but the tool sped up production enormously.

Ivana Karhanová: What did it take to scale to other countries?

Roman Berglowiec: We used local e-shops, but localization was tricky. At that time, good translations were hard to come by, and we made a few funny mistakes. Today, AI helps even here–not just with translation, but with cultural context and tone.

Ivana Karhanová: Let’s fast forward again. In 2024, you launched Everbot. Was it a response to the explosion of models?

Roman Berglowiec: Yes, for three reasons. First, internally. We had people using dozens of tools and paying for too many accounts. Our CFO kept asking, “What’s this invoice?” So, we needed to unify that. Second, we saw major international investment in AI aggregators, especially in the US and Asia. And third, we saw a gap in the Czech market. We wanted to build the local leader in this space.

Ivana Karhanová: So, this wasn’t like with the painting business – this idea didn’t come from AI?

Roman Berglowiec: It was based more on real-world experience, but we still validated it using AI–analyzing market trends, user behavior, funding activity… The need for simplification and smart routing of prompts is real.

Ivana Karhanová: The idea is to provide one app with multiple models, under one login?

Roman Berglowiec: Exactly. One login, one place. Users stay in control of their data–we don’t see their content–but we can observe which features are used. That helps us iterate and provide consistency, especially in teams. When everyone uses different tools, the outputs are inconsistent. Everbot solves that by centralizing and standardizing AI access.

Ivana Karhanová: And now you’re offering it to other companies?

Roman Berglowiec: Yes. We started as our own first customer. Our company group has around 3 billion CZK in revenue, across many divisions and Everbot helps teams across marketing, HR, sales, and leadership. Marketing is still the most natural place to start–content creation, campaigns, analysis. We’ve used AI here since 2018, so this was a natural progression.

Ivana Karhanová: How is Everbot being received on the Czech market? You offer it to individuals as well as companies, right?

Roman Berglowiec: Yes, we do. But the Czech market is still in its early stages when it comes to AI adoption. We often say: AI technology is already in orbit, but many users are still riding bicycles. And that’s okay–we’re at the very beginning. Right now, around 40% of Czech companies have had some exposure to AI. But in many cases, it’s not a decision from management–it’s employees experimenting on their own to improve how they work, make it more efficient or higher quality. So, AI is often entering companies from the bottom up.

We’re mainly targeting small and medium-sized enterprises – they’re incredibly important. In Europe, there are about 25 million of them, and they generate over 50% of the continent’s GDP. Plus, decision-making processes are usually much quicker in SMEs than in big corporations.

Ivana Karhanová: So even though you say, “AI is for everyone,” there are 900,000 tools out there. Your goal is to create a kind of digest, so users don’t have to choose between ChatGPT, Gemini, or another obscure tool?

Roman Berglowiec: Exactly. Everbot has only been on the market since November, but over 17,500 people have tried it so far, which we think is great. In the first version, users had to manually switch between models–try Mistral, then GPT-4.5, then Claude. We’re now launching a new version that will handle that selection for you. You just say, “Create a post for Instagram,” and Everbot automatically picks the model that’s best suited for that task–maybe something obscure but specialized and better than the popular models. You shouldn’t have to think about the technical side. You just enjoy the best possible output.

Ivana Karhanová: Where do you see the market heading? You used the analogy of space–and early space was pure chaos.

Roman Berglowiec: The market today is chaos. Even for someone like me, who follows it closely, LinkedIn feels overwhelming with a new model, new tool, new trend every day. It’s easy to feel like you’re missing something. But I believe the market will consolidate. Tools like Everbot, which not only aggregate models but also help you choose the right one for the task, will become key.

If you look further ahead, it’s not just about writing a single post. Soon, you’ll give your assistant a complex task: “Create a social media campaign,” and it will design the strategy, write the content, generate the visuals, even prepare ads and budgets. Tools like Manus have shown glimpses of this. It’s early, but it’s coming fast. At Everbot, we’re working toward that too – making it possible to define multi-step tasks that the AI agent can complete on its own.

Ivana Karhanová: What you’re describing–uploading brand voice, performing marketing strategy tasks–is already technically possible.

Roman Berglowiec: True, but right now you have to feed the AI pieces one by one. What’s coming is that you’ll input everything at once, and it’ll work through it all. Maybe the AI agent takes 15–45 minutes, but it replaces a team that would need days or weeks.

Ivana Karhanová: So, we’re heading toward working with AI like with a colleague?

Roman Berglowiec: Exactly. You’ll sit down, turn on your microphone, and talk to your assistant like a smart colleague. “Here’s what I need,” and it starts working. Then you give feedback–“change this, fix that”–and it improves. The great thing is that it works with data, not emotions or opinions. Just knowledge and experience.

Ivana Karhanová: I’m picturing open-space offices full of people talking over each other to their AI agents.

Roman Berglowiec: Yeah! One thing I love is that this empowers senior roles–a CFO, for example, can now work faster, smarter, and answer much more complex questions if the company’s data is well-prepared. I don’t think AI will take people’s jobs. But those who don’t adopt it–or refuse to work with it–will fall behind.

Ivana Karhanová: And implementation depends on how good your data is.

Roman Berglowiec: Right. The bigger the company, the more complex the data, and the more security, compliance, and access control you need. Smaller firms handle data more flexibly. But the foundation of all this is a well-structured data warehouse. If you don’t have one, your AI won’t perform well.

Ivana Karhanová: When you say, “data warehouse,” do you mean the company’s complete dataset, or just a SharePoint with marketing materials?

Roman Berglowiec: Both. It depends on what the company wants from AI. In big corporations, the goal might be to centralize everything and then give each team access to the parts they need. But that’s hard to manage. More realistically, most teams will build their own mini data hubs–like a dedicated space with brand guidelines, strategies, past campaigns, etc. That’s a good place to start.

Ivana Karhanová: You mentioned that people who don’t adapt to AI may struggle. Do you really expect there will be professionals who simply refuse?

Roman Berglowiec: That’s a hard question, but yes. I remember as a kid, my grandfather once came in, put a mobile phone with a SIM card on the table, and said, “I’ll never use this devil’s machine.” And walked away. Of course, he eventually changed his mind. Same with AI. Some people will reject it at first, but AI will become embedded in everything–TVs, elevators, microwaves, even fast food ordering. You’ll say, “Take me to the third floor,” and the elevator will take you. AI will be all around us.

Ivana Karhanová: Behind that is just voice recognition and automation.

Roman Berglowiec: Exactly. And the real opportunity is in scaling teams. In our company, we once had two people creating social media content–they spent 320 hours. Now, with AI, we do the same work in 40 hours. That’s 9x time savings. We didn’t fire those people–we reallocated their time. That single change saved us over a million CZK and allowed us to grow.

Ivana Karhanová: So, you’re scaling humans, not replacing them.

Roman Berglowiec: Yes. That’s the mindset shift. AI isn’t about layoffs–it’s about multiplying what your team can do. And if you look at it that way, AI doesn’t just grow your company–it can grow GDP, boost productivity, and raise the economy as a whole. That’s the broader impact.

Ivana Karhanová: Roman Berglowiec, CEO of Everbot – thank you for sharing your view on AI today.

Roman Berglowiec: Thank you. It was a pleasure.