Last week, I built an entire module — backend, frontend, API routes, database models, tests — in 4 hours. It would have taken me 2 weeks in 2023.
Here's the uncomfortable truth: 90% of what I used to do is now done by AI.
The boilerplate. The CRUD. The CSS. The debugging. The "I've written this exact function 200 times" code. Gone. An AI agent handles it in seconds while I sip coffee.
So am I worried about my job?
Not even a little.
The 10% That Matters
Because the 10% that's left? That's the part that took me 30 years to learn.
Architecture. Knowing WHY a system should be designed a certain way — not just how. Understanding that the elegant solution today becomes tomorrow's technical debt if it doesn't account for scale, team dynamics, and operational reality.
Domain expertise. Understanding the business problem deeply enough to know which technical approach will actually work in production at 3am on a Friday. No amount of generated code can substitute for deeply understanding the problem space.
Judgement. The instinct that says "this looks right but it'll break under load" or "this is technically perfect but the ops team will hate it." This comes from scars. From production outages. From seeing the same mistakes play out across decades.
Taste. Knowing what NOT to build is more valuable than knowing how to build everything. The best architects I know are defined by what they choose to leave out, not what they add.
AI can generate code at superhuman speed. But it can't decide what's worth building. Not yet.
What I Tell Enterprise Leaders
Stop worrying about AI taking jobs. Start thinking about what to automate.
Every hour your engineers spend writing boilerplate is an hour they're NOT spending on architecture, strategy, and innovation. That's the real waste.
The companies that will win aren't the ones that resist AI. They're the ones that free their people to do the work only humans can do — while AI handles the rest.
New Roles Are Already Emerging
The job market isn't shrinking. It's shapeshifting:
- AI orchestration engineers — designing how multiple AI agents work together
- Prompt architects — crafting the instructions that turn AI capabilities into business value
- Human-AI workflow designers — building the processes where humans and AI collaborate
- Domain-specific AI trainers — teaching AI systems the institutional knowledge of specific industries
The Bottom Line
Your move is simple: stay on top of AI, or get left behind.
I chose to stay on top. After 30 years of writing code, the best code I write today is the code I tell an AI to write for me.
The future belongs to those who learn to lead the machines, not compete with them.
This is exactly why we built Astra AI — an AI system that handles the 90% of IT incident investigation that used to require senior engineers, so your team can focus on the 10% that actually needs human judgement.
What's your experience? Has AI changed how you work? Drop a comment — I'd love to hear from other veterans.
Top comments (0)