Waiting for the Agents

I’m mostly waiting for my AI agents to finish.

That thought keeps looping in my head. And right after it, the uncomfortable follow-up: do I actually want this?

I feel the tension very clearly. Since March I’ve been working on my new product, and I’m honestly moving about 5x faster than before. At times it almost feels like cheating. In roughly nine months I built a full platform largely on my own, with some help from Davy. Something that would normally require a full dev and design team, plus months of coordination and meetings.

That speed is not just because of AI tools like Claude Code. It’s also Rails, and many years of deep experience with this exact stack. Still, the question keeps coming back: what does my future as a software engineer actually look like?

Since Claude Opus 4.5 was released, it feels like another major step change. After a short brainstorm with agents, I can sometimes ship a complex feature in a single shot. And for simpler things like bugs, I just point to the ticket, and minutes later the fix is there.

My role is clearly shifting. Less time typing every line of code. More time directing, validating, and steering agents. From writing code to managing systems that write code.

I personally get a lot of satisfaction from the end product. Shipping something real that creates value. At the same time, I can absolutely see that if you love coding for the sake of coding, this future might feel deeply uncomfortable.

It’s exciting. It’s unsettling. And it’s happening fast.