AI Probably Won’t Take Your Job. But Don’t Get Too Comfortable.

Talk to any developer right now and you’ll hear it. The way software gets built has shifted. Not gradually… dramatically, and in the space of about a year.

I’ve spent over 15 years building software and working with AI. I’ve seen hype cycles come and go. This one’s different.

Six months ago, exploring a new feature idea meant scoping it, writing the logic, building a rough implementation, testing it, refining it. Hours. Sometimes days. Now I can prototype the core logic, produce tests, and pressure-test the whole approach in under an hour. Not perfectly. But well enough to know if it’s worth pursuing, and that changes everything about how I think and experiment and move.

That’s not a small thing. That’s a fundamental shift in what’s possible between breakfast and lunch.

And here’s what I want to be clear about: I’m not talking about AI helping you write an email or autocompleting a formula in Excel. Those are nice. This is something else. I’m talking about AI agents doing meaningful chunks of work, work that would previously have taken an expert hours or days, in minutes.

Any role that involves generating, analysing, producing, or making decisions on a computer is going to feel this. Not eventually. Now.

AI is a multiplier. Which sounds great until you realise what that means for someone whose value is purely execution. If you’re faster than the person next to you, that used to be worth something. Now the gap between fast and slow can be closed by anyone with access to the same tools. The leverage is shifting toward judgment, toward context, toward the person who knows which question to ask rather than the person who can answer it fastest.

There’s a bigger shift underneath all of this, though. One that doesn’t get talked about enough.

For decades, we’ve built dashboards and interfaces because that was the only practical way to get humans and computers talking to each other. We built software to translate intent into structured inputs. Click this. Filter that. Export. Repeat.

Now we can just… ask.

That sounds simple. It isn’t. It changes how systems get designed, how data gets connected, and who gets access to insight. When AI can move across systems and interpret data in context, the gap between the person who operates a tool and the person who analyses its output starts to blur. A manager shouldn’t have to wait three days for a report that answers a question they had this morning. They should be able to explore it directly.

The organisations that are genuinely grasping this aren’t just bolting AI onto their existing stack. They’re restructuring their data so AI can actually reason across it. That’s a different project entirely.

Yes, there’s hype. Yes, models hallucinate and make embarrassing mistakes and confidently tell you things that are wrong. None of that changes the direction.

The practical question, the only one worth spending time on, is how you adapt to it.

The people who experiment now, who actually understand where the limits are, who adjust how they work based on what these tools can and can’t do… they’re going to move faster than everyone else. Not because they’re smarter. Because they’re not waiting to see how it plays out.

I’m already building differently because of it. The shift has already started for me. The question is whether it’s started for you yet.