If you've read my previous posts, you'll know I'm always curious about new tech... but also a little cautious; especially when it comes to big shifts like AI. Lately, it feels like the whole industry is in a rush to figure out what "useful" actually means for AI. No one seems to have a clear answer yet, and that uncertainty makes me pause; not just for technical reasons, but for environmental, financial, and social ones too.

From where I'm sitting, it looks like companies are throwing a lot at the wall to see what sticks. There's a sense of "first to market wins," and a lot of experimentation, but not always a lot of clarity about what problems are actually being solved, or if we're just adding more noise.

Copilot Agent in My Workflow 🤖

I've been using GitHub Copilot Agent for a while now, mostly to see where it fits into my day-to-day. It's not magic, and it's definitely not a replacement for thoughtful engineering, but it does have its moments. The most useful thing for me has been the ability to delegate some of the repetitive or boilerplate tasks like opening a draft PR for a feature I've been thinking about, or running a first-pass code review before I dig in myself.

I've found it genuinely helpful for fixing small bugs, especially when there's already a well-defined issue to work from. At work, if I want to quickly proof out an idea or check if something's doable but don't have the time or headspace, I delegate it to the agent and let it work while I focus on my day-to-day tasks. And despite my initial resistance to this kind of change (it's a shift to oversee implementation rather than do the implementing myself), I'm starting to see Copilot Agent more and more as an assistant. Right now, it feels a bit like working with a junior intern (sometimes it needs too much guidance, but it's making progress), and I'm learning to adapt my workflow around it.

Still Figuring It Out 🤔

I don't have a grand conclusion about AI in software development. I'm still figuring out what "useful" means, just like everyone else. I have concerns about the pace and direction of adoption, and I think it's important to keep asking questions about impact, not just productivity.

For now, Copilot Agent is a small part of my toolkit. I'm watching, experimenting, and reflecting. More thoughts on the state of AI, hype cycles, and what "useful" really means (to me), coming soon.

Until then… ✌🏽