The Courage to Slow Down in the Age of AI

We were building faster than ever. A small team shipping what used to take ten people. And it still wasn't fast enough.

That's the tension nobody warns you about. You adopt AI, you transform your workflow, you ship features in hours that used to take weeks — and the people around you look at it and say "why isn't it faster?" Because in the time it takes you to do things properly — to plan, to test, to make sure the architecture holds — someone who doesn't understand engineering sees only that the output could be higher. And in that pressure to go faster, the first things you cut are the ones that feel slowest: the meetings, the check-ins, the conversations. The human parts.

I know because I just lived through it.

Last July, I joined Aibility as CTO to build Aimee — an AI coach helping people discover what's possible with artificial intelligence. Over the months, we grew a team around it. Tom — co-CEO with Martin — driving product and design, Kačka helping shape the UX, Slava and later Petr joining on development, Žaneta driving sales, Filip as the founder, and everyone else making the company tick. It was a real team effort.

We built with AI from day one. Claude, Cursor, agents — the full stack. And the speed was real. A small dev team delivering what would have taken a traditional team many times our size. Aimee ended up working across multiple countries, used by thousands of people. We helped Aibility adopt GitHub for their knowledge base, set up modern workflows with agents and CLI tools, built Aimee for Slack, created self-service tooling. We did a lot.

And at the end of April, our dev team — me, Petr, and Slava — is leaving. Aibility is refocusing on the Superpowered Professional methodology. Our chapter is closing. We didn't ship everything we dreamed of. But we shipped enough to know it was real, and the company is in good hands moving forward.

It wasn't smooth sailing though. And honestly? The rough parts taught me more than the wins did. That's what I actually want to talk about.

Here's what I got wrong.

We were a small team. Five, six people at the core. I genuinely believed we didn't need heavy process. Groomings, sprint plannings, formal demos — that's for bigger teams, right? We're small, we're fast, we message each other on Slack every day — that counts, right? We had our weeklies. We had bi-weekly office days. That should be enough.

It wasn't.

What happened in practice is that things drifted. Our product person gradually stepped back from the day-to-day. Our key stakeholder wasn't present at most of the meetings we did have — getting information secondhand instead of seeing the work directly. Demos happened for a while, then stopped. The marketing team, the education team, the back office — everyone was doing their part, but nobody was seeing the full picture together. Not regularly. Not in a way that kept us aligned.

And here's the uncomfortable truth: we tried, a few times, to get the rituals going. But we never drove it hard enough or long enough for it to stick. There was no one person relentlessly making sure the right people were in the room, that the stakeholder saw the product evolving week by week. In a small team you assume alignment happens naturally. It doesn't. It has to be built and maintained, deliberately, by someone who cares enough to keep insisting even when it feels like overhead. We didn't do that well enough. I should have insisted harder.

It cost us. In quality — things slipped through that shouldn't have, testing wasn't where it needed to be. In alignment — people working in different directions, not out of malice but because the connective tissue between us had worn too thin. In shared understanding — devs, product, stakeholders, nobody seeing the full picture together.

Those rituals we never properly established? They weren't process overhead. They were social infrastructure — the thing that kept us connected as people. Without them, we were just individuals working on the same thing.

We learned that the hard way. And I believe what happened to us is a preview of what's coming for everyone.

AI is removing the reason people learned to work with other people.

I want you to sit with that for a moment. Not "AI is taking jobs." Not "AI is changing teams." Something deeper and, I think, more unsettling.

A developer who used to pair with a colleague now pairs with an AI. A designer who needed a conversation with a product manager can now generate, iterate, and ship alone. A project manager who ran standups finds that agents already track everything. The need to collaborate — actually need, not choose — is shrinking. Fast.

And here's the thing: the standups, the code reviews, the pair programming, the whiteboard arguments — those weren't just about producing better software. They were the training ground. For communication. For empathy. For the messy, uncomfortable, incredibly important skill of understanding another human being's perspective.

When you don't need to collaborate, you stop learning how.

Think about who this hits hardest. The people who already struggle with social engagement. The introverted developer who never loved standups but who, because of them, learned to articulate their thinking. The junior who grew by sitting next to a senior, absorbing not just code patterns but how to disagree, how to compromise, how to work through conflict. Those people don't get that training anymore. Not because anyone decided to take it away — but because the work no longer requires it.

I worry we're heading toward a world of brilliant solo performers who can build anything — and struggle to work with anyone. And honestly? That worries me more than job replacement.

"But we'll have meetups! Communities! Slack channels!"

Maybe. But here's the crucial difference. Those are opt-in.

The genius of the old way — the thing we didn't appreciate until it was gone — was that collaboration was opt-out. You had to actively resist working with people. The default was together.

AI flips that default. Alone becomes the path of least resistance. And opting into collaboration requires exactly the social skills that are atrophying because you don't practice them.

It's a loop. And it's already spinning.

I see it already. And here's the uncomfortable part: the solo path genuinely works in some ways. One person with AI, no communication overhead, context shared seamlessly between them and the model — they can be remarkably effective. You can't just dismiss that.

But here's what they lose. The challenge. The collision of different brains with different experiences. The moment when a colleague says something you'd never think of — not because they're smarter, but because they've lived a different life and see the problem from a place you can't reach. AI can simulate expert panels and devil's advocates, but it's extrapolating from patterns, not thinking from lived experience. Our messy, contradictory, emotionally-driven human brains still do something that AI doesn't — and that something is exactly what gets lost when you optimize for one person doing it all.

And then there's the quality question. We're moving so fast that things break in ways they didn't before. GitHub — the platform most of us build on — had to publish a special blog post addressing their "recent availability issues" after six incidents in February alone, with some services hitting 90% failure rates. Anthropic accidentally leaked Claude Code's source code through an npm package just this week. Studies are finding that AI-generated code has 1.7 times more bugs and over half of developers have shipped AI code with security flaws they didn't catch.

This isn't because people are careless. It's because the speed is intoxicating and the checks that used to catch these things — the code review where a colleague actually reads your work, the demo where a stakeholder spots the gap, the conversation where someone says "wait, have we thought about..." — those are the same rituals we're skipping. The same social infrastructure we're letting erode.

We trade robustness for velocity and tell ourselves it's fine. Sometimes it is. Sometimes it's 2 AM and something breaks and there's one person who knows how it works — because they built it alone, with an AI, and never talked anyone through it.

And here's the thing that keeps me up at night. Communication isn't just a workplace skill. It's how we stay human with each other. Listening, disagreeing and still respecting each other, holding space for someone's perspective — that's not just what teams need. It's what families need. Friendships. Communities.

The workplace was, quietly, one of the last places where adults were forced to practice being with other adults who think differently. If that goes away, where do people learn it? Social media? We've seen how well that's going.

I don't have a framework for this. I don't have a five-step guide. I'm not selling anything.

But I've built products. I've led teams. I've done coaching work with people on how they lead and how they talk to each other. And I keep getting pulled toward the question of what happens — to people, to teams, to the systems we build around them — when everything changes faster than we can adapt.

About a year and a half ago, I wrote a post called "Jump into the Unknown." I was leaving corporate life, stepping into coaching, wanting to help people on their transformation journeys and, through that, slowly change the systems around us. I had no idea what I was getting into.

I still don't. But I've been inside the thing now. Not reading about AI transformation — living it, with a team, with all the mess and the speed and the drift. And what I came out with isn't a methodology or a playbook. It's a question: how do we keep teams human when the work no longer requires them to be? That's what I want to figure out next. Not as a product. As a purpose.

To Petr and Slava — you are my friends, not just colleagues. Petr, we've been building things together across companies, across years. I know I can always rely on you, and I'm grateful — for you and for your family. Slava, this was our second time working side by side, and I'd do it again without hesitation.

To everyone at Aibility — Tom and Martin, co-CEOs who believed in this from the start. Filip, the founder. Kačka, who shaped how Aimee looked and felt. Žaneta, the queen. Petra, Aneta, Verča, Helča, Blanka, Jakub, Hanka. And those who walked part of the way with us — Verča, Pavel, Honza, Petr, Čeňka, Jakub, Alča, Jakub. Every one of you made this what it was. Thank you.

And that gratitude? That's the point. Those names aren't just a list. They're the people I grew with, argued with, figured things out alongside. No AI gave me that.

We're entering an era where working alone is easier than ever. The tools are incredible. The speed is real. But your AI won't genuinely disagree with you. It won't bring a perspective shaped by a life you haven't lived. It won't push back because it cares about the outcome in a way that's different from how you care. It will be helpful, agreeable, and fast — and none of that will teach you how to work with someone who sees the world differently than you do.

That's the trade-off nobody's talking about. And I think it's the one that matters most.

If any of this resonates, don't just like this and scroll on. Talk to someone about it — a colleague, a friend, the person next to you who's been unusually quiet lately. And if you want, talk to me. I mean that. I'm not the most naturally social person, and reaching out still costs me something every time. But I think that's exactly why it matters — choosing to connect when you don't have to is the whole point.