
AI gives you an answer before you’ve had time to think. That’s useful. But it can make you lazy. I try to stay away from the screen sometimes: puzzles, chess, anything that makes me wrestle with a problem myself because being faster isn’t worth becoming less capable.
Was thinking about code ownership today. Primary/secondary owner models for vertical slices used to work well. But they just get create blockers. Instead, we’ve added architecture drift detection to our PR evaluations. If it fails, the PR doesn’t merge.
A question I get quite a bit is “what is the future of software engineering?”
Not only that, but what is the future of working with engineering teams?
Big release for us today!
We are live on Product Hunt 🏆🐱
v7 marks our biggest release
- 🤝 cal.com/events
- 🧭 cal.com/discover
- 🤖 cal.ai
- 🎥 cal.video
producthunt.com/products/cal…
Join us!
2
12
424
If AI writes most of the code, what exactly are we reviewing in a performance review? The question is: did you make good decisions? Did you grow? Did you use AI wisely? Did you help the team move in the right direction? Yes… that’s harder to measure. But it matters more.
Global teams give you access to incredible people and diverse ways of thinking. They also create frustration, burnout, and isolation when someone is always taking the late call or waking up to a day’s worth of decisions.
We’ve had a new intern cohort start within the last 3 weeks. I don’t care how many PRs they ship in four months. I care whether they leave with a deeper understanding of how systems work. If I ask you how web caching works, do you want to ask Claude… or do you want to know how yourself?
I’m truly concerned about societal impacts of AI and seems not enough people are talking about it.
Juniors learning from experienced mentors? way less
Professor office hours? Way less
People deeply thinking about hard problems? Way less
People just talk to their AI. 👎🏼
AI has caused a Dunning-Kruger epidemic.
We’ve basically let car painters build engines easily but they have no clue about engines. I estimate this will come to a head sooner or later in terms of burnout and frustration for senior engineers.
9
332
The software engineer role is splitting into two. One path is product: you own the what, and AI handles most of the how. The other is foundation: systems, scale, security, the deep stuff. Some people won't want to just prompt all day. That's fine. But you'll have to choose.
Was thinking about bluntness today. You can be blunt without being rude.
No mind-reading games with people like that. And on a team spanning 15 countries, the trick is knowing who wants it straight and who needs a different approach.
A lot of teams ask, “Which PRs are low risk?” Better question: how do we add enough tests, checks, and safety measures so that today’s medium-risk change becomes low risk? That’s how we get faster without gambling with production.
Everyone assumes AI means engineers work less. Our activity data says the opposite. You prompt, it runs, it comes back. So what do people do? Open ten more threads... This is why we didn't get our time back. We filled it with more context switching.
"Move fast and break things" was meant to encourage experimentation, not excuse poor execution.
In the age of AI, there's even less reason to choose between speed and quality. Automated testing and AI-assisted development make it possible to move quickly, learn fast, and build things right.

