He designs voicebots and chatbots that handle hundreds of inquiries in just a few minutes, pull data from CRM systems, verify the creditworthiness of prospective tenants, and create requests directly in internal systems. In this interview, he explains what separates successful automation projects from average ones, why “love projects” are mainly about the people on the client side, and how his call-center experience and passion for gardening show up in the way he builds AI solutions today.

At coworkers.ai, you started in marketing back when it was still one of Artin’s projects. How did you get into building bots? Do you still remember your first one?

The project lead, Tom Lysek, asked me if I’d like to give it a try. The work seemed more than interesting to me, so I went for it. Looking back, I consider it the best career decision of my life. My first bot was a demo that enabled clients to showcase the possibilities of artificial intelligence. But at that time, GPT didn’t exist yet. So we trained our own NLP — a system that learned to understand human language. The robot could actively call customers and collect overdue payments; it could also handle changes to electricity and gas advance payments. It was a demanding process. Today it’s much easier.

Would you say your work has changed a lot with the development of technology?

Absolutely. When I started as an IT Consultant for AI solutions, almost no one knew what I actually did. Today, this role — under various titles — exists in practically every large company, from e-commerce to banking and non-banking sectors. Thanks to technology and the constantly increasing demand for AI, my job has changed beyond recognition. Today we build sophisticated solutions that exchange data with clients’ CRMs. For FAQs and standard communication, robots load knowledge bases directly from websites. And thanks to product feeds, they can also help with sales.

Which “aha moment” on the client side makes you happiest?

More than with standard FAQ bots, I experience the biggest aha moments with bots that handle a specific use case — for example, checking an order status. From my student days I remember my part-time jobs in call centers, where I handled up to 80 such emails a day. So it doesn’t surprise me when a client is amazed that a robot can process several hundred emails in just a few minutes. And on top of that, it can send customers a tracking link and detailed information pulled from the system.

In your opinion, what makes an automation deployment successful?

Above all, it’s the client’s vision and the quality of their requirements. If a company wants a bot only because the competition has one — and they don’t want to spend time developing it — it often ends in dissatisfaction. That’s why I always tell clients: “The robot can’t read your mind. It’s a new employee, and it will only be as good as the care you give it. If you invest time in it, it will quickly learn to do the job exactly the way you need.”

So the client’s approach is essential. Do you have a “love project” in this area?

I would definitely call Heimstaden my love project — a company that rents residential and commercial premises across Ostrava. Years ago, together with a client, Ms. Písková (an AI pioneer at Heimstaden), we built the first voicebot that collects housing inquiries. Its level of sophistication surpassed all solutions built up to that point. The robot not only collects all personal data from the caller, but also verifies their creditworthiness via an API, optionally pulls data from the ARES register, and at the end creates the request directly in Heimstaden’s system.

As I said, love projects aren’t only about the solution — they’re mainly about the people on the client side. And here, the success of Heimstaden’s bots is largely thanks to the customer care manager, Mr. Sedlář. His vision for automating call-center processes continues to amaze me even after more than five bots we’ve built together.

While predictive call steering verifies the caller in Heimstaden’s CRM based on the phone number and gathers all the information to pass to an operator, the chatbot in the My Home app (Můj Domov) enables tenants to submit requests without having to call or write to customer service. We’re talking about more than 100 variants of requests — from reporting a complaint about neighbors to damaged exterior doors. And more use cases are on the way.

Work is clearly a big passion of yours — but what do you like to do when you’re not working?

I like going for a casual walk with my wife combined with a special beer, playing PlayStation with friends, or reading good sci-fi. My absolute favorite is the Mass Effect game series — the idea that once every 50,000 years a race of robotic reapers controlled by artificial intelligence arrives and wipes out all advanced life in the galaxy is terrifying and fascinating at the same time. What’s more, the AI does it in order to preserve life, because in its view advanced civilizations inevitably destroy themselves sooner or later — usually through conflict between organic and synthetic beings.

In recent years, I’ve also discovered the magic of gardening, so I’m constantly thinking about what else our garden can handle.

What made you fall in love with gardening?

I love the calm and the feeling that I’m creating my own small ecosystem. I look for ways to arrange plants so they help one another — a bit like a team where everyone has a role. One plant attracts bees, another provides shade, a third repels pests. And together it works much better than each plant going solo. It’s a kind of natural version of automation: you set it up once and then it runs on its own. Except instead of code, you work with soil and compost.

‍