Why AI Interviewers get better answers than humans?

What if your AI interviewers are pulling better answers than your human team, just because people aren't worried about being judged?

Filip Schouwenaars has been building in AI since 2013, back when it was neither cool nor easy. He was DataCamp's first hire, rose to VP of Engineering, then moved to ML research at VITO. In January 2026, he joined Conveo, a Y Combinator S24-backed AI research platform that lets companies run qualitative interviews 10 times faster than before. He's also one of us at Syndicate One.

Filip sat down with Robin Wauters to talk about what actually works (and what doesn't) when you bring AI into a startup moving at speed.

• He breaks down why people give answers that are three times as long to an AI interviewer as they do to a human interviewer.
• He shares what we learned at the S1 portfolio AI meetup about where Belgian companies really stand on adoption.
• Filip says Belgium's biggest missed shot isn't the capital. It's deep tech that never makes it out of academia.
• And he gives the most honest answer we've heard yet to the classic founder question: Would you ever start your own company?
• Six weeks at Conveo. Filip hasn't written a single line of code. That's exactly how it should be.

Recorded following Conveo's $5.3M seed round.

Listen. Share it. Building something big from Belgium? You know where to find us.

What is The Syndicate One Podcast?

Conversations with the founders, investors, and operators driving Belgium’s startup scene,
Exploring the ideas, stories, and strategies that accelerate the ecosystem flywheel 🚀.

FILIP: Hi, my name is Filip Schouwenaars, and I'm a software engineer at Conveo, which is an AI-led research platform.

ROBIN: Filip, thank you for joining and taking the time for this Syndicate One podcast interview. We know you as part of Syndicate One, but you already have a very interesting background and career. Maybe walk us through your backstory and how you got involved in the Belgian tech ecosystem in the first place.

FILIP: I would say my intro into the tech ecosystem in Belgium was a bit of a lucky shot. It starts in 2013. I had just graduated from KU Leuven with a degree in electrical engineering and artificial intelligence, and I started an internship at DataCamp, which is now no longer a startup — it's a scaleup. I basically started creating courses for them. They were still in the phase of teaching in R, which is not a very common language now. Now they also teach Python and SQL. So it's basically an online data science school, DataCamp. I started creating courses there, they liked it, I liked it, and I became a full-time employee.

Since then, I took on various roles. In the first four years, I basically went up the career ladder — software engineer, team lead, and, at some point, VP of Engineering — though I didn't really like that role, so I wanted to stay closer to the product. After four years, I did an amazing trip with my now-wife, then-girlfriend, and after that, I went freelance, doing shorter-term projects, always with startups — basically CTO and CPO roles.

Then the former CTO of DataCamp reached out, and I rejoined DataCamp for another three and a half years, where I built a new product. That was basically the entire DataCamp chapter — seven and a half years in total. I experienced it go from the early days, when I was basically the first employee, to being part of a company with more than 150 people.

After that, I took a big turn and went to work at a Flemish research institute — VITO — which is in the woods of the Kempen. A very different scenery. There, I went super deep into machine learning and AI —more specifically, AI modeling for everything related to sustainability— with a lot of surrogate modeling, which is training AI models to replace expensive physical computational models.

But then the gravitational pull of the startup ecosystem was a bit too strong, and at the beginning of this year — January 2026 — I decided to join Conveo, a super exciting AI startup riding the wave like crazy. I'm liking it a lot.

ROBIN: We're going to talk about Conveo a lot more in this interview, but it sounds like you've had a lot of experience in AI — 13 years since you graduated, deep into it already. You've seen a lot of this evolution before most people picked up on the trends. Is it becoming very difficult, even for someone like you, to keep track of what's going on? Or does having that history help?

FILIP: The history helps. When people talk about neural network architectures — transformers, attention-based models, that kind of thing — I can relate it more easily to concepts I already learned. But even for me, it's going so fast that keeping track of everything is quite hard. There are also different levels at which you can keep track. You can go super deep and really look at what's novel about a given architecture and follow the AI arms race between the labs. But then there's also the layer above that — on the user end of these models, keeping track of all the applications AI enables. That's also hard to keep up with, and I'm more in that second camp. I'm not necessarily a fundamental AI researcher closely following every paper; I follow things more from the practical, end-user side.

Something worth adding: I was learning about neural networks before it was cool and before it was really accessible to an average software engineer. If you didn't have a Stanford PhD, training neural networks with GPUs meant a lot of hoops to go through. What I've seen since I started my career is that the entire developer and deployment experience has become much, much smoother. There are way better open-source packages now. That has been a big reason for the wide adoption of these tools.

ROBIN: That means a big democratization — also in the way software is being built and used. Is that always a good thing in your opinion?

FILIP: I would say yes. I can't immediately think of a bad side to that. I think it's great.

ROBIN: So you joined Syndicate One around the time you were still at VITO working in research. What attracted you to the Syndicate One network and model, and what have you got out of it so far?

FILIP: I was always very involved with startups and trying to keep track of everything. But then I took a turn into a research institute — a different environment, a different work setup, lots of freedom and flexibility, which I loved. But I was missing that one foot inside the startup ecosystem. If I'm being selfish about my reason for joining Syndicate One, it was to basically keep my finger on the pulse of what's happening in the tech startup scene.

I came across Syndicate One through a connection at YC. And I'm loving it. It's exactly bringing that for me. Last week, for example, we had a meetup on how you can leverage AI throughout your organization — not only at a technical level, but also more generally, including Claude, Claude Workspaces, and everything around MCP connectors. This was actually the first time I felt like I was bringing together my experience in just knowing about these things, my experience in teaching, and being able to bring that to the portfolio companies. That felt like a great day where everything came together.

ROBIN: That was actually going to be one of my questions. What did you learn during that AI meetup? We're recording this a few days after we brought together a lot of talent, not only from the portfolio. Any takeaways you want to share?

FILIP: A couple of things. One is that the degree of adoption varies widely across companies. Some people are still in the phase of figuring out what everything is. Others are further along but still have questions and are still inspired by what others have to say. The other takeaway is that the meetup created something of a safe space among portfolio companies where everyone could openly share what they're struggling with. We didn't have to pretend we were massive AI adopters — people who weren't yet could be honest about it, and those further along were super helpful. I don't think it ends there for us. This was a kind of test case, and it's clear there's room for Syndicate One to fill a gap, because a lot of people are still in the "what is all this?" stage. If we can give them some guardrails — to use an AI term — some guidelines and a platform to stand on, I think that would be super helpful.

ROBIN: Do you think the companies that are adopting AI the fastest and experimenting as they go will, by definition, be the companies that win? Or is there still merit to waiting it out and taking a more cautious approach?

FILIP: It's a good question. I can't see into the future, but I do think there's a risk in not following it at all and just brushing it off as a fad. There are clearly use cases where AI is a real enabler and a genuine multiplier on output. At Conveo, for example, Tom Husson is doing great work — and he even acknowledges that it's still AI; it can still make mistakes — you have to watch out. Some of what he's building might be completely outdated in three to six months. But during those three months, we still had an edge. And you can rest assured that a large percentage of what you're building right now to have AI systems do things for you will be recycled. But that's the same as in software engineering: if you build an advanced software application, it's constantly evolving. And with AI, it's even more so, because it's so extremely fast-moving.

ROBIN: For those who don't know Conveo — it's an AI-native company that we've backed as Syndicate One. What are you building, who are you building it for, and why is there such a need for it?

FILIP: Conveo is an AI-led research platform that enables you to conduct qualitative research interviews — and increasingly quantitative ones as well — that are completely AI-led. With the help of AI, which is throughout the whole stack, you can design a study: say you want to learn how people perceive Syndicate One, what emotions it evokes. You start a study, and it helps you draft research objectives and questions. Then you launch it, targeting, for example, Belgian entrepreneurs or people considering becoming entrepreneurs. It finds those people through recruitment panels. You say you want 20 respondents. After the interviews are complete, we produce full transcriptions, conduct a comprehensive AI-assisted analysis of all interviews, and present the results in reports and slide decks, with suggestions for follow-on studies if necessary.

What's extremely important is that we do qualitative research at scale. What might take you ten months — or three months for those interviews — we do ten times faster. And it's all AI-run. It's working really well. Our main users right now are researchers, but we're really trying to go upmarket, positioning ourselves not just as a platform that can run studies for you. Still, as an insights platform you can trust for any research, from brand tracking to customer experience studies like NPS. But rather than a typical NPS survey, you get the score, and then we ask respondents why, and you can pull out the themes.

ROBIN: I'm guessing most of your customers are enterprises or agencies doing research that are already knowledgeable about AI?

FILIP: I wouldn't necessarily say that. There are very big food and beverage companies using Conveo that are not AI-native — real incumbents. For them, it's truly a "what the f***" moment when they see how radically the time to insights is slashed. And they don't have to interact directly with the AI. We are a layer on top of that, so they don't need to know how to prompt it or provide guardrails — we take care of that for them.

And actually, there's research showing that if you conduct a research interview with an AI assistant — people do know it's an AI, but it still sounds slightly robotic — the responses are, on average, three times longer than with a human interviewer. You'd think it would be the opposite, but when you're talking to a person, you're also more conscious of what that person might think of you.

Another relevant example: some brands doing fast-moving consumer goods — cleaning products, that kind of thing — really want to know what someone's home actually looks like. In the past, they had to visit in person, which was expensive, and people cleaned up before anyone arrived. On Conveo, respondents open their phones and film their rooms — sometimes they're dirty, and that's exactly what the brand wants to see. So it's not only faster but, in many cases, more honest.

ROBIN: This is a market ripe for disruption, but that also means the space is getting crowded. How do you deal with that?

FILIP: The race is definitely on. There are some big players out there. If you dig deeper, there are sometimes genuinely PhD-level problems to solve — for example, doing live emotional analysis of someone watching a trailer video: recognizing 50 different emotions in real time. That's pretty deep technical work. And you see these different platforms all moving in that direction.

The angle that we're really proud of is being extremely customer-obsessed. A lot of times, customers choose Conveo because they choose us — the team — not just the product. What that translates into: we put an engineer in the room on almost every client feedback call. We try to avoid adding a layer of product management between the customer and the fix. So when there's a small but annoying bug — and if someone's spending five hours a day on the platform, that bug matters — it might be 30 minutes to fix with today's agentic coding tools. We go out of our way, and sometimes during the call itself, we do the fix and push it live. And for clients, that's amazing. They say: "You're on my side. I know that whenever I have a problem, you're going to listen and try to make my life easier." That sets us apart a lot.

ROBIN: Do you also use AI on a personal level?

FILIP: I have almost a schizophrenic personality when it comes to that. On the job, it's wild. A big part of my role is still writing software — figuring out what to build, talking to users, writing code. I've been with the company six weeks now, and I haven't written a single line of code myself. I've very much taken the red pill when it comes to using AI at work. But personally, I don't have an extremely busy personal agenda or a crazy number of side projects, so I don't feel like there's much to optimize in my personal life yet. What I do use it for is the common stuff — consumer-grade ChatGPT usage. I don't have any MCP connectors set up to my personal ChatGPT or Claude license whatsoever. Very different from how I use it at work.

ROBIN: Did you actually join Conveo through Syndicate One?

FILIP: Nuanced answer. One of the co-founders of Conveo is Dieter De Mesmaeker, who is the company's CEO and was formerly CTO at DataCamp. I worked a lot with Dieter over the years. He's amazing. And many former DataCamp colleagues took different detours and ended up at Conveo. So I already knew the team was great, and they were also nudging me a bit — "you should join, you should join." What helped on the Syndicate One side is that from the quarterly updates, I could see they were killing it. That was actually one big reason: this is a rocket ship, and I need to make sure I'm aboard before it exits the stratosphere.

ROBIN: Taking more of a big-picture view — have you seen any positive evolution in the last 10 to 12 years in the Belgian tech ecosystem?

FILIP: That's something I say quite often. DataCamp was founded in 2013, and I still remember how hard it was to raise capital at favorable terms. If you went with Belgian investors, it would cost you an arm and a leg in equity, even for small amounts. And if you went to the US, it was: "You're Belgian, not London-based, not US-based — why would we bother?" What I see right now is that there's way more capital in Belgium itself, which is great — and Syndicate One is definitely helping there. But Europe, as a place where tech startups can thrive, has also helped open up the pockets of US investors.

I remember we were seeing double-digit month-over-month growth at DataCamp, and back then, it was pretty hard to find the right investors on the right terms. Today, those same numbers would get a "shut up and take my money" reaction.

From my time at VITO, I feel that the model of deep tech spinning out into great startups is happening way too rarely relative to the amazing research that Belgium produces. For all the amazing research coming out of places like KU Leuven, Ghent, and others, there are examples, but way too few. I hope we can figure that out somehow, because there's so much value to unlock.

ROBIN: That's true at the European level as well. It's almost as if being in academia means you're expected to stay in academia. In the US, it's way more common to do a PhD at MIT and then say: "Now I want to become a tech entrepreneur." We still have to live that a bit more here.

FILIP: Big gap between Europe and the US in that respect.

ROBIN: Would you ever consider starting your own company, with the knowledge and network you've built up?

FILIP: It's a great question. I sometimes get asked this because I've been in different leadership positions, and people tend to say I do a good job in those roles. What I've seen in all the successful companies I've joined is a healthy mix of co-founders: people who are big-vision thinkers and people who can take that vision and turn it into a reality — something you can actually charge for. I've seen that at DataCamp. I'm seeing it again at Conveo. I think it's pretty rare to see founders who have both, and those are the unicorns. I'm definitely one of the second kind.

I'm not necessarily a big-vision thinker. So rather than me saying I want to be the one founding a company, it would take someone coming to me with a big, hairy idea and saying, "I want to start this, will you join?" I wouldn't say no. But I think I won't be the one to take the very first step, because it's just not really my personality.