Dev.to AI 🤖 Ai 👁 0 📖 2 min read

AI gave me spare time, and sprints don't know what to do with it

Ever since I started parallelizing a lot more of my work with AI, I've felt the need to go looking for side projects. Of course, I try to make sure those side projects are directly valuable to the company. And that's le

Ever since I started parallelizing a lot more of my work with AI, I've felt the need to go looking for side projects.

Of course, I try to make sure those side projects are directly valuable to the company. And that's led me to things like exploring potential refactors in some of our classes, speeding up certain scripts, investigating slowdowns in our APM, and looking into our testing infrastructure to see if I can help there.

That's just me as a developer, sitting there and thinking, "How can I better help this company?" And I can definitely find plenty of valuable things to do, between what I see in the code and my general understanding of the company's mission, its direction, and what matters most to our clients right now.

But even with that, my view is never the complete picture. A manager or a stakeholder sees things from a different angle and knows about plans and priorities I might not. So what I pick might not be the best use of my time.

I could be refactoring an area that's about to change for reasons nobody has announced yet. Or I could be speeding up a feature while the company's real focus is somewhere else entirely. So I do welcome a little bit of steering. I'd still like to keep some time for free exploration for sure, but it makes sense to let the people with the bigger picture point some of it.

So the side projects aren't wasted time, and the system isn't completely broken. But I think there's a better way to handle this. For years, a lot of managers have prepared tickets in an almost sequential way. You work on one thing at a time and don't pick up the next task until the first one is done or about to be done. The only time you'd start something different is when you're waiting on something async, like a PR review or an answer from someone.

With AI, though, a lot of the time you hand the agent a task for one ticket, and all of a sudden you need to find something else to do. That gives you plenty of time to go look at PRs or share some updates with the team, which is great. But sometimes that's not enough. You finish the PRs, you send your messages, and you're still waiting for the agent to come back.

I think sprints make this especially awkward. Every time I wrap up my sprint tasks early, I have to go looking for the next ones to pick up, and that ends up messing with the numbers people are tracking behind the scenes. Which is fair, they plan around those numbers. But it shows how poorly the sprint model fits this new way of working.

I'm not trying to come out strong against sprints or managers, and I'm not here to sell a new process either. I just think the way we plan work needs to be rethought. It needs to be a bit more flexible, and it needs to make room for the fact that I want to, and very reasonably can, work on three separate things at the same time now.

📰 Read the original article on Dev.to AI

Originally published by Dev.to AI. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.