There’s a lot of noise about automation replacing developers, and the timing looks damning until you actually line the graphs up.
The steep drop in software engineering roles doesn’t match the release of ChatGPT, Claude Code, Gemini, or the DeepSeek moment. It matches interest rates going from 0.25% to 5.5% between March 2022 and August 2023. Far more boring, and far more human.
Zoom into the last twelve months and engineering roles are climbing again, up 15% since the start of the year, while AI has only got better at writing code. The layoffs are real, and AI keeps getting named as the cause, but the hiring data tells a different story.
It isn’t evenly distributed, though. Front-end and mobile roles are down, junior roles have taken a hit across the board, and it’s harder than ever to break into the industry. This is a boom for engineers who adapt, not a boom for everyone.
Treat AI as a power tool rather than a threat. A power tool doesn’t replace the person holding it. It changes what one person can get done in a day.
Think like an engineering manager. Orchestrating a team of agents is a different job to writing every line yourself. I spent six years as an engineering manager before coming back to engineering, and that turns out to be far more relevant now than it was when I left.
When all you have is a hammer. Stop reaching for AI when deterministic code or human taste would do the job better.
You are the bottleneck. Once the code writes itself, you’re the constraint. Spin up too many agents at once and you’ll drown in slop.
Slow down to go faster. Slow is smooth and smooth is fast.
0:00 AI didn’t kill software engineering0:25 Six years as an EM, back to engineering1:15 The layoffs are real, but is AI to blame?1:33 Was it Claude Code, DeepSeek or ChatGPT?2:19 The real cause3:13 Staying relevant as a software engineer3:22 AI is a power tool, learn to use it3:46 Start thinking like an engineering manager4:16 When all you have is a hammer4:52 You are the bottleneck5:25 Slow down to go faster