Last month, I attended four AI events in Copenhagen, each leaving me with something different to think about.
At the Grok Bot meetup at Trustpilot office, I got a glimpse of how agent UX will evolve. Afterwards, I experimented with building a full software engineering team of agents for my personal news digest project, with a group chat of developer, designer, DevOps and a PM agents. It was definitely overkill for such a small project, but I was surprised by how naturally the agents communicated with each other, almost like a team in a Slack group chat. It also made me think about the limitations I’ve encountered when using Cursor to coordinate parallel agents working on more complex features.
At Claude for Fintech community event, the closing fireside chat reinforced systems thinking as a valuable skill: breaking problems into components, identifying where you can add value, and keeping feedback loops small.
One idea particularly stuck with me: the question is increasingly becoming “Should we build this?” rather than “How do we build this?”. Just because AI makes something cheaper to build doesn’t mean it’s worth building. Strong CS fundamentals remain essential, but so do adaptability, product thinking and understanding what customers actually need. Repodo’s great talk covered their development loop where there is no human code writing and only minimal human interventions. The industry is moving even more towards automation.
Builders & Brews introduced me to some interesting possibilities with Nebius and NVIDIA tools, while PineGrove AI’s event offered a look at the current state of local AI.
Furthermore, attended the Bilt.me hackathon which actually surprised me in how well it has progressed when I first tried it couple months ago. The product felt amazing. Easy-to-use for vibe coding, yet also provided some serious mobile development opportunities for prototyping. I tried making an application for learning about investing and pensions in Estonia using native iOS components which worked surprisingly well.
My biggest takeaway from the month: as building software becomes easier, knowing what to build, making good engineering decisions and automating the unnecessary become even more valuable.
