Paweł Chrzan is no stranger to building things from scratch. A former head of R&D at Booksy, he left the corporate world of Facebook with one goal in mind: to build something of his own. That opportunity arrived in 2020, when he co-founded Wellbee alongside a colleague, riding the wave of demand for psychotherapy that the COVID-19 pandemic had unleashed. Wellbee’s platform, which serves both individual clients and businesses, went on to connect users with a network of more than 300 verified professionals spanning psychotherapy, psychiatry and coaching.
In 2024, the company was acquired by Benefit Systems, the leading player in the EU employee benefits market, with above 1 billion euros of revenue. Now, Paweł has set his sights on a new, more ambitious venture: MOS Health.
You are the co-founder of Wellbee, a genuine Polish success story. What set you on the path to building something entirely new?
The urge to build something of my own had been with me for a long time. When I left Facebook and moved back to Poland, that was already the plan. Mental health became the focus for a combination of reasons. The pandemic was the obvious catalyst — people were struggling, and the entire business model for therapists had been turned on its head overnight, since almost none of them had ever worked online before. That disruption was, in effect, an opportunity. Together with a pre-seed investor, we decided it was worth exploring as a genuine business direction. None of it would have happened without that shake-up caused by lockdown; it’s what allowed us to get the business off the ground so cheaply in the first place. On a more personal level, psychology has been a lifelong interest of mine, so it wasn’t a huge leap.
At what point did you realise Wellbee was going to make it? Was there a specific moment?
It clicked once both our B2B and B2C arms started showing real promise — and this was still early days, around year one. Even then, I knew we were in for plenty of problems; we ran short of cash more times than I’d like to admit. It was something of an uphill battle throughout, which I suspect is true of most companies at that stage. But the fundamentals of the business were genuinely strong, and we managed to land contracts with some remarkably large corporations. Those early signals gave us real confidence. From there, it became a balancing act of scaling sensibly within a tight budget. We were fortunate to secure funding from day one, but we still had to spend heavily on marketing — and a fair chunk of that, if I’m honest, was wasted.
Looking back, what would you say were the biggest mistakes in the early days? What would not you repeat?
Honestly, that advice doesn’t fully hold up any more — agentic coding revolutionised the economics of building MVPs. I was once full of opinions on how extremely small the initial product should be, but I wouldn’t follow my own advice today. The truth is that you can build a genuinely decent product in a week now. Back then, the idea was to start with something radically simple and just run with it, which is exactly what we did — but you have to accept, by design, that you’ll end up scrapping that first version entirely. Its purpose isn’t to deliver value to the customer; it’s to deliver feedback to you. It exists purely to test your hypothesis about how the whole thing will actually work in the real world, before you throw it away and build the version that does. With the agentic coding revolution now under way, there may well be a faster path — expanding on and refactoring that first build rather than binning it — but the underlying principle hasn’t changed.
Given your current position, is it easy to step back and hand over something you built from nothing?
It’s very much a process rather than a single moment. I’m a big believer in the earn-out model, where your company gradually links up with the acquiring business, and step by step, the emotional attachment loosens. It’s a bit like watching a child go off to university — they drift further away, but you’re proud of what they go on to achieve. That gradual handover between the acquirer and the acquired company is, I think, genuinely good for one’s state of mind. Of course, every acquirer is different. In our case, we were bought by one of the largest companies in Poland, and we’ve since been integrated into their systems, both organisationally and commercially.
So now you’re on to the next challenge. How does your approach with MOS Health differ?
There’s plenty of overlap, but also a distinctly different philosophy at play. MOS is, I’d say, far more ambitious. We’re building on the newest possible technology stack, and the product itself is a much more digital proposition — really a digital ecosystem, rather than a service. Wellbee, at its core, was about therapy. MOS is built around an agentic reasoning layer that makes proactive recommendations, and that sits right at the heart of the product. That’s a considerably harder problem to solve, because you have to deliver real value purely through intelligence and user experience — there’s no human therapist to fall back on. But the upside is that it’s infinitely scalable. If personality and intelligence can be properly engineered, the growth potential is enormous.
Who is MOS Health actually built for?
It works both ways — for a company or directly for an individual. The system takes in whatever data you’re willing to give it, and it copes well even when that data is patchy. From there, it tries to genuinely understand you: who you are, what your goals are, and it will actually negotiate those goals with you. Over time, it builds an extremely detailed profile of you and your day, to the point where it might flag something like an unusually high heart rate and suggest it could be down to too much coffee — perhaps a single espresso instead of a double. It looks at how your meetings are going, which moments in your week are most stressful, and gathers that intelligence to tailor how it communicates with you. It learns the kind of language that resonates with you or sets you off, and it works out which channel you actually want to be reached on. Prefer Slack? It’ll message you there. Prefer iMessage or email? Fine too. It learns your communication style and adapts to it, guiding you gradually towards your goals. In many ways, it’s both an insights platform and a tool for introducing incremental change into your life. We pull data from any wearable, so we’re completely agnostic on that front, and we sync calendars alongside wearable data to build a genuinely rounded picture of who you are. We’re also conscious that people are drifting away from apps generally, so we want to meet people where they already are, using whatever channel suits them best.
What does the US market offer that Europe doesn’t, in this space?
The US, particularly over the past year, has introduced regulatory changes that matter enormously to a business like ours. The category is defined as lifestyle rather than medical, which means you can deploy AI systems provided you steer clear of medical language, diagnoses and direct recommendations. That gives you room to analyse data that would likely be classed as medical in Europe. Sleep quality, for instance, and its correlation with other factors — in the US that’s framed as healthy lifestyle guidance, whereas in Europe it risks being seen as practising medicine. Europe remains far more tied to older regulatory frameworks, which matters a great deal for specific, serious health conditions. But preventive care and the promotion of healthy lifestyles — habits that keep people well in the first place — are arguably the most underserved part of the traditional medical system. Our aim is to keep people away from needing a medical treatment for as long as possible.
Do you think preventive health will eventually overtake reactive healthcare?
The budgets allocated to reactive health are on an entirely different scale to preventive care. We’re realistic — we know we’re a drop in the ocean. But we’re hopeful that, over time, we can help shift some of that reactive spending towards prevention instead.
Health is deeply personal. How do we build trust in AI when it comes to something so sensitive?
Every advanced diagnostic tool produces false positives and false negatives — it will get things wrong, and there’s no getting around that. What we can control is being deterministic. We work from a very strong foundation of data, which makes hallucinated answers far less likely. We never claim to prevent disease. Think of us more as an assistant, while the user remains firmly in the driver’s seat — a bit like coding tools. They’re excellent, but the person behind the wheel still needs to be steering for genuinely great things to happen. We’re fully aware that users arrive with very different levels of knowledge about preventive health, so our role is to help them make informed decisions. We’re never prescriptive — we’re there to inspire, not instruct.
What is the long-term goal for MOS Health? Are you building towards an exit, or something more lasting?
This is a company I could genuinely see myself involved with indefinitely. I think there’s a huge amount of work still to come, especially as broader healthcare systems evolve, and real potential for this to grow into a proper medical company. Our ambition is to move beyond prevention and into caring for users in medical settings too — systems that monitor patients continuously and help them build lasting habits. We do believe the future of medical intervention will look something like this: a blend where medical specialists lean on systems like ours to support their patients.
Was there a single moment that made you focus on workplace habits and unhealthy lifestyles, or was it more gradual?
There wasn’t really a eureka moment. It’s something I’ve believed in for years — that this was where things were heading. I think we’ve built up certain unhelpful expectations about office life. Movement alongside colleagues should be encouraged far more — walking meetings, ideally outdoors. Instead, there’s a culture of running yourself into an energy debt and powering through on coffee, which simply isn’t sustainable long-term. You can’t operate in sprint mode indefinitely, and burnout has become almost universal as a result. I’m hopeful our technology can flag, with real precision, exactly when someone is pushing too hard — a gentle nudge to take a break, have lunch, look at the trees. Food culture in offices is another area where the US lags considerably behind Europe — sweets and colourful snacks everywhere. Making the healthy option the default choice, rather than the unhealthy one, matters enormously. Even something as simple as the weather plays a role: you’re far more likely to go for a walk on a beautiful day than when it’s pouring with rain and the traffic’s backed up — even if you’re exactly the same person both days.
What’s your own background, and what shaped the way you think about all this?
I genuinely believe having ADHD has helped me build these companies. I was diagnosed with it before it was even a formally recognised diagnosis in Poland. But human behaviour has fascinated me for as long as I can remember, and I tend to approach it from a mathematical and symbolic angle — which traces back to my studies in quantitative economics and theoretical mathematics, and it’s that same grounding that now underpins how I think about AI. I don’t think you need to be a deep expert in one single field — being broadly knowledgeable across many, and understanding how they connect, matters more. Jack of all trades, master of none is, in my view, actually a rather good trait for an entrepreneur to have. You need a working knowledge of a wide range of areas. For me, that’s AI, health and performance, which is where the company’s focus sits — but naturally you also need to juggle marketing, leading a team, legal matters and a great deal more besides.
