How Super.work builds AI products without PMs

Main Takeaways

• Building AI products means exploring what works, not starting with a fixed plan.
• At Slite, there are no product managers, but the team still builds fast and stays product-led.
• Moving from product to operations changes perspective: clearer decisions come from understanding the whole business.

I’ve worked in product for about ten years now. PM, then Head of Product, and over the last year, COO at Slite. The title still feels a bit weird to be honest as I'm still a Product person but we played around with the title and stuck with COO.

When I joined Slite, I was helping build our knowledge base. Later, I helped kick off our second product, Super.work, which started as an AI search layer and has slowly grown into something more workflow-oriented.

From product specs to business decisions

I still do product work—just from a different angle. I’m not as precious about every detail in the interface anymore. I care more about: how do we position this? Will people understand it? How does this help us grow?

Back when I was a PM, I sometimes overlooked that part. We built features that looked great, but then sales couldn’t pitch them. I didn’t pay enough attention. These days, I think more like a bridge. I know what marketing is trying to do. I know what’s happening in sales. I watch our costs. That context changed how I make decisions.

If I could go back, I think I’d write fewer product specs and more business cases. Ask earlier: how do we explain this to customers? Why would someone actually pay for this?

And maybe even more important—I over-engineer a lot less now. Not sure if it’s because I’ve grown, or because I just don’t have the time. But I’ve started making faster, clearer decisions. And I probably have more impact because of that.

Building AI products is different

When we started working on Super.work, it became clear fast that the way we’d built before wouldn’t work.

In traditional product, you usually start with a problem. You define the solution, then build toward it. With AI, it’s often the opposite. We explore what’s possible, then figure out if it’s useful. One of our engineers, Antoine, was developing our AI Assistants and realized that we could automate and schedule them. That wasn’t planned—it just emerged.

I sometimes write a press release or a short concept doc to give direction. Then someone prototypes something. Then we talk. Then we ship. There’s no long discovery phase. The devs lead with exploration. Designers jump in once there’s something real to shape. And I help frame it all so customers understand it.

We’re not solving problems—we’re exploring opportunities. And that’s a big mindset shift.

You also can’t design AI features in Figma and expect them to work. You need to see the real thing in action. Without a prototype, it’s impossible to understand what the model is actually doing.

And you can’t really QA an AI product in the traditional sense either. You don’t know what summaries your customers are getting. The quality depends on the inputs—often messy docs and unpredictable tools. That’s why we’ve been working on a kind of query planner, so answers become a little more predictable. Still, you can’t control everything.

All our paying users are in Slack with us, which helps a lot. We see what people try, where things break, and what surprises them. That feedback loop is gold.

No PMs, but still product-led

We don’t have any product managers at Slite or Super. But we’re still very much product-led. That says a lot.

Chris and I split product responsibilities between us. And honestly, I don’t think I’d hire a PM right now unless they were technical and could build themselves. But then that’s what a good product engineer does anyway.

That might sound harsh, but I think the product role is changing fast. I didn’t move into operations to leave product behind. If anything, I want to get back to it. But I also saw where things were heading and wanted to broaden my perspective.

Smaller, faster, more technical teams

Looking ahead, I think teams will get smaller and more technical. One person might do the job of what used to be three. We move quickly. Antoine ships a feature in a few days. We announce it. Then we move on to the next thing.

I don’t know if we’ll still need big product teams, or agile coaches, or even PMs in the traditional sense.

If you’re working in product, you’ll need to be closer to the tech. You’ll need to understand how models behave, what’s possible, and what’s not. The bar is going up.

And for companies building in this space—especially in Europe—I hope we see more visibility around who’s doing what. There’s a lot happening, but not many places to actually see it.

What won’t change

Even with AI, some things stay the same. You still need people you trust. You still need clear communication. You still need a good culture. Otherwise, what’s the point?

I don’t think we’ll all become robots—or want to. Tools change. Workflows shift. But the human side matters just as much.

The idea of going on an offsite with ten AI agents and no humans sounds… deeply boring.

Final thoughts

I’m still learning. I don’t have it all figured out. But working across the business gave me a different kind of clarity.

I used to spend a lot of time making beautiful product specs. Now, I spend more time making things happen. I see the bigger picture. And I probably ship faster because of that.

For now, it feels like the right place to be.

Perspectives: Honest conversations on crafting great products over a cup of coffee.

I sit down with friends across design, data, engineering, ops, and more—people who work closely with product leaders, from PMs to CPOs. We talk about how teams really work, where things break down, and how AI and new ways of working are reshaping the future of product.

Coffee of the day

Playground Coffee · King Kongo · Hamburg