Everyone Wants to Code, Nobody Wants to Operate
With all the hype around AI coding agents, a question keeps coming up: will we even need developers anymore? Yes. More than ever.
But not for the same reasons as before.
Writing code was always the easy part of the job. The hard part? Keeping that code running over time. Software engineering is programming over time. It's about how systems change.
The No-Code Lesson
Let's take no-code as an example of what's being sold as the future — custom tools, throwaway, built by non-experts to solve specific problems.
The classic story:
Joe from accounting spends 10 hours a week on a repetitive task. He can't get dev resources — they're busy on the product. No worries, Joe is smart. With some Googling, a few no-code tools, and Excel macros, he builds a tool.
His 10-hour task now takes 1 hour. 🎉
Except... time passes. Business changes. Accounting rules evolve. Joe is now shackled to his system. He can't take vacation. Nobody else can run it. And it never really works right.
This is what Feynman called "the computer disease": automating things is fun, you forget you don't always need to.
What's NOT Fun
The not-fun part? Running a service. Reliably. At scale. For years.
People don't buy software, they hire a service.
You don't care how iCloud works — you just want your photos to magically show up across devices. You don't care about Word, Notion, or gDocs — you just want to write what's on your mind and share it.
Good software is invisible.
The Real Engineering Questions
When code becomes easy to produce, here are the questions that matter:
- What's your uptime?
- Your defect rate?
- How quickly do you recover from incidents?
- Do you detect issues before users or after their complaints?
- When a vendor misbehaves, do you notice or wait for tickets?
- Can you scale without breaking things?
- Can you build systems bigger than one brain?
- At 3am in a different timezone, does someone respond?
- Will you sign a guarantee that your software works when I need it?
SRE Becomes King
SRE = Site Reliability Engineering.
It's the art of running services reliably. Monitoring, alerting, incident response, capacity planning, chaos engineering...
With AI agents that can generate code, the 90% to get a working demo becomes easy. It's the remaining 190% that makes the difference:
- Edge case handling
- Error recovery
- Continuous security
- Performance at scale
- Living documentation
- Zero-downtime migrations
What This Means for You
If you're a developer: learn ops. Kubernetes, observability, incident management. That's where value is created.
If you lead a team: hire people who know how to run systems, not just code features.
If you're automating with AI: don't underestimate operations. Generated code is 10% of the work. Running it in production is the rest.
Want help putting AI systems into production that actually hold up? Let's talk.