Over the past year, I’ve been diving deeper into the world of AI—both as a developer and as someone who simply enjoys exploring new tools. I’m still relatively new to building with AI, and I’ve definitely felt like I’m playing catch‑up at times. But the more I experiment, the more I’m realizing how much potential there is, not just for productivity, but for creativity and problem‑solving. It’s been energizing to see how quickly things are evolving and how accessible these tools have become.
Working with Devin
One of the tools I’ve spent a lot of time with recently is Devin. What stands out to me is how capable it is across a wide range of tasks—everything from small, repetitive chores to more complex engineering work. Interestingly, it has become my go‑to for the simpler tasks. Whether it’s something mundane or something that just needs a quick, reliable solution, Devin handles it well.
As Devin continues to mature—especially as it’s equipped with more “knowledge” files that give it the baseline understanding of a junior developer—I can see it becoming even more versatile. It’s already able to work across multiple repositories for a single problem, which is incredibly useful.
That said, it’s not perfect. I’ve seen it hallucinate, miss files that are clearly present, or propose solutions that feel a bit off. But even with those quirks, the benefits often outweigh the drawbacks. The ability to offload certain tasks and let Devin explore multiple codebases at once has saved me time and opened up new ways of approaching problems.
Diving into Copilot CLI, Custom Agents, and Skills
Lately, my focus has shifted toward Copilot CLI and the broader ecosystem around building custom agents and skills. This has been one of the most exciting parts of my AI journey so far. There’s something incredibly satisfying about crafting tools that fit the way I work—automations, workflows, and helpers that feel like extensions of my own thinking.
Exploring what’s possible with custom agents has pushed me to rethink how I approach development. Instead of asking, “How do I solve this?” I’m increasingly asking, “How can I design an AI‑powered workflow that solves this with me?” It’s a different mindset, and it’s opening up new and interesting solutions to problems I’ve solved the same way for years.
What I’m Learning About the AI Space
The more time I spend in the AI world, the more I see a consistent message: AI is a tool—powerful, flexible, and evolving—but still a tool. And like any tool, the right workflow for one person might not be the right workflow for someone else. There’s no single “correct” way to integrate AI into your day‑to‑day work.
What matters is finding the agents, skills, and patterns that make you more productive. Use what helps. Ignore what doesn’t. Build what’s missing. The real value comes from shaping AI to fit your workflow, not the other way around.
If you want to leverage AI to be more productive, let’s talk. I’d love to help.
