Hey folks,
You might remember back in September 2025 I dropped a post sharing my initial thoughts on AI in coding. Most of what I said still holds up for me, but after actually using it day-to-day for the last couple weeks, I’ve got some real-world updates and a few tweaks to my take.I finally jumped in and started tinkering—installed Cursor IDE (which comes loaded with those built-in AI agents) and let it loose on my projects. Bottom line: it’s a mixed bag, same as before. There are legit pros, some clear cons, and I’m still 100% not on board with the idea that AI is about to replace developers anytime soon.
Here’s what I’ve seen in practice: - New / greenfield projects — this is where it shines. Give it a clean slate and tell it the tech stack you want? It can spin up a solid skeleton stupidly fast—faster than any of us typing from scratch. Huge time-saver there.
- Styling / UI polish — eh, not so much. It’ll throw something together that looks generically “fine,” but if you have a specific vibe, brand guidelines, or just want things to feel right, you’re usually back to tweaking by hand. It’s like asking a robot to match your personal taste—it gets close, but rarely nails it.
- Existing / legacy codebases — way more hit-or-miss. Simple refactors or small changes can somehow take 5× longer than if I’d just done them myself (super frustrating). But then it’ll surprise you and handle a chunky piece of business logic flawlessly on the first try. It’s unpredictable, which means you can’t just lean on it blindly.
So yeah—no magic bullet. It speeds some stuff up and drags on others. Overall, it makes certain parts of the day a little easier, but it’s not solving every problem or making anyone obsolete.
Important part (and I’m dead serious here):
You still have to understand your codebase. You still have to know what the AI just spat out, why it did it that way, and where it probably screwed up the edge cases or missed the nuance. There will always be manual adjustments—sometimes small, sometimes big. Developers aren’t going anywhere. If you’re already in the field, you’re still very much needed. If you’re thinking about getting into development, don’t let the hype scare you off—you’re still needed too.
The skillset might shift over time (more reviewing, prompting, architecture thinking; less rote boilerplate), but the core—actually understanding code, systems, and trade-offs— isn’t going obsolete. That’s non-negotiable, even as AI usage ramps up.
Curious what y’all are seeing out there. Anyone else deep in the Cursor / AI trenches lately? Hit me with your wins, headaches, or hot takes.