AI is changing how we build software. But… the biggest change is not that AI can write code. The real change is that writing code is becoming less of the bottleneck.
This changes where engineers need to focus their efforts!
Architecture before implementation
Small technical teams can now build a lot with AI. But adding more developers does not automatically make development faster.
You still need someone who understands the architecture and knows what the product actually needs.
Give AI agents the right tools
Simple, modular CLI tools can be more useful than adding layers of frameworks and wrappers.
The Unix philosophy becomes interesting again: small tools that do one thing well and can be combined.
Senior engineering judgement matters
AI can make changes very quickly.
The problem is knowing whether those changes actually fit the system.
Senior engineers need to understand how the different parts of the architecture work together and catch problems before they become production issues.
Don’t become just an implementation engineer
If your main value is writing isolated features, AI will increasingly handle more of that work.
You need to understand systems, architecture, product decisions, company’s goal and customer needs.
And there is one more thing! Don’t let AI set your pace. Agents can work incredibly fast. That does not mean you need to work all night to keep up.
AI is not a short-term opportunity. It’s a multi-decade shift.
The goal isn’t to write more code with AI. The real goal to build better software through better judgement, a deeper understanding of the product, and a clear focus on long-term outcomes.
I have spent the last decade building projects on my GitHub. Check them all for inspiration or contribution.
I’ve also got more content coming your way on my LinkedIn.