What have you been experimenting with and learning from it, SW Developer & AI Explorer, A-Wave Oy, Teemu Piirainen?
# The Shift I Didn’t See Coming
A week ago, I gave an AI agent a task and went back to my own work. A while later, I noticed myself glancing over to check how it was doing.
Like you’d check on a colleague working at the next desk. That moment caught me off guard.
I didn’t plan for it. I had spent months experimenting with AI in my daily work as a software developer. Trying different approaches, giving agents more responsibility, and seeing where things break. Most of those experiments ended with me stepping back in and fixing something. Some failed completely. A few worked surprisingly well in ways I hadn’t expected.
But that moment, checking on an agent the way you’d check on a teammate, was different. Something had shifted without me noticing.
I started thinking back. When did it change?
There was no single turning point. It happened gradually, through dozens of small experiments. Early on, I supervised everything closely. One task at a time. Full control. I treated AI as a faster extension of my own hands.
Over time, I started trying bolder things. Running several agents at once. Giving them overlapping responsibilities. I didn’t know what would happen. I tried it anyway. And sometimes things emerged that I hadn’t designed. Agents noticing things I hadn’t asked them to look for, their work shaping each other’s focus. That’s when it got interesting. Not because the technology impressed me, but because I started losing control in ways that turned out to be productive.
Then came the moment that stayed with me. I launched an agent on a larger task, turned my attention elsewhere, and kept working. When it finished, it reported back with a summary. I reviewed the output, gave feedback, and assigned the next thing.
I stopped and realised: I’m not working with a tool anymore. I’m coordinating work.
That felt strange. A little exciting. And honestly, slightly uncomfortable. For over twenty years, I’ve been the one writing the code, solving problems with my own hands. Now my work was quietly becoming something else. Deciding what to delegate, reviewing what comes back, thinking about what to try next. I hadn’t planned for that. It just happened.
No article or tutorial prepared me for that feeling. And I think that’s the point.
A year ago, I cancelled my online course subscriptions. Not because I stopped wanting to learn, but because I realised the real learning was already happening at my desk, through every failed experiment and every unexpected result. Every mistake taught me something about where the boundaries are. Every surprise showed me where the real opportunities hide. None of that was in anyone’s curriculum.
The learning is deeply personal. My way of working with AI looks nothing like anyone else’s, because it grew out of my own experiments, my own mistakes, my own discoveries. There’s no shortcut to that.
What surprises me most is that this isn’t really a story about technology. It’s a story about how work can quietly reshape your role, if you keep experimenting and stay honest about what you notice.
I didn’t set out to become someone who coordinates AI teams. I just kept trying things and paying attention to what happened. Now, when I glance over to check how an agent is doing, it feels normal. That might be the biggest shift of all.
Thank you, Teemu, for all your expertise and for sharing!
Dev.to: https://dev.to/teppana88
Small experiments and inspiration 