Developers who spent years building software before AI, how does the job feel now?
it feels like the same pattern you see with food delivery drivers. once the system shows something can be done faster, the expected delivery time just keeps getting shorter.
Exactly. The faster we can build, the more closely we need to watch what’s being built. Someone still has to understand the code, question the decisions, and catch problems before they make it into production.
I think management often just sees the output and how fast it appeared. They don’t see the work behind deciding whether it’s actually safe to ship, or dealing with it when things break later. After a while, trying to explain that over and over feels pointless.
That’s a good way to put it. We learned a lot by making mistakes and fixing them, and AI won’t remove that part. The mistakes may shift from “how do I write this?” to “is this the right approach, and what happens when it reaches production?”
that’s worth thinking about before things change
“Love coding, hate what the industry has become” really resonates with me. I still enjoy building things, but I got tired of the corporate side too. Working on my own products has brought back some of that feeling of ownership, and AI makes it much easier to try an idea without spending a year getting to a prototype. I hope your app goes somewhere.
being able to try a few approaches quickly really changes what feels worth tackling.
Yeah, I find that part fun too. Getting the automation stable takes work, but once it’s reliable, it gives you more time to focus on the things you actually find interesting — or the decisions that matter more.
That’s a great use for it. Clearing out things that have been sitting there for years while still shipping features must feel good. Were those mostly small fixes, or did AI also help you tackle the bigger, riskier refactors?
Yeah, I get that. If the craft itself is what you enjoy, letting AI do more of the hands-on work can take away some of that feeling of getting better every day. Faster doesn’t always feel more satisfying.
Honestly, it already feels pretty close to that for smaller systems. Once a project gets bigger, though, a lot of the work is keeping the agents from drifting — making sure they understand the architecture and that each change still fits with everything else. I spend much more time directing and reviewing now.
That’s an incredible range of projects for one day. I love that you’ve found a way to keep that curiosity going after all these years — with a whole team of agents to explore things with now.
Exactly. We’ve spent years learning whatever a project needed and putting it to use quickly. That habit helps a lot now, whether it’s a new tool, AI, or figuring out a business problem.
40 years — that’s quite a perspective. I agree that experience makes a big difference in knowing when to trust the output and when to question it. I also like your point about junior engineers. They’ll still learn by building and fixing things, even if the way they get there looks very different from ours.
Haha, same here. I’ve spent a lot of my career being the “whatever this project needs” developer. People say a good dev should be able to do a bit of everything, but sometimes it just feels like you’re the team handyman. Looking back, though, being able to adapt to different problems is a skill in itself.
Since you were ten? That’s impressive. Glad building software is still so much fun for you.
I get that. What worries me about vibe coding is how easily the code becomes a black box. It can look fine and ship quickly, but if nobody really understands it, the surprises show up later when requirements change or something breaks in production. That’s where I still want someone who can explain and own the decisions behind it.
Yeah, exactly. The iteration cycle is so much faster now. You get through the straightforward parts quickly, but that just means you reach the harder problems sooner.
Exactly. That experience makes it easier to know where to start, and to spot when an AI suggestion sounds good but wouldn’t work in practice. Having seen how things play out across different companies is a real advantage.
Yeah, spending so much time on tedious stuff you never wanted to deal with is exhausting. I wouldn’t be surprised if more specialized AI tools can handle a lot of the cost and bug hunting in a couple of years. The people part might be harder to fix, though.
This sounds a lot like how I work now. I treat AI like a small team that can help with different parts of the job. Sometimes the most useful thing it does is push back on an idea before I spend time building it.
I get a lot more done this way, but I still spend most of my time making decisions: what’s worth building, which tradeoffs to accept, and whether the result is actually ready to ship.
Sorry, I misunderstood what you meant. Seeing work dry up and layoffs coming is scary.
Honestly, even at good companies, I never felt much of a sense of belonging or security. That’s part of why I decided to build my own thing. It has its own risks, but I feel more grounded working on something that’s mine. I know that doesn’t make what you’re facing any easier.
Yeah, I can see that happening. Honestly, after years of putting so much time into company work, and I got tired of the internal politics too. At some point it felt a bit absurd, so I left to build my own products. AI has made doing that on my own a lot more possible, and I’m enjoying this part of the change.
I feel you. I spent so many nights working late, especially around releases or when something broke in production. AI has made parts of the job easier, but I’m still tired from those years, boss.
it useful for smaller, well-contained modules. but in a larger codebase, I have to watch it closely and check how each change fits the architecture.