One thing I've never understood about software engineering interviews:
Companies want developers who can build production-ready systems.
But most interviews still revolve around solving abstract puzzles under a 45-minute timer.
Are we testing engineering skills or just preparation skills?
The AI race is starting to look surprisingly one-sided.
Claude keeps delivering where it actually matters: model performance and real-world usability.
GPT needs to catch up.
Gemini has the marketing and distribution, but the hype often feels bigger than the actual experience.
Grok is losing some of its initial momentum.
Kimi hasn't managed to stay consistently relevant.
Meta feels out of the race, and Apple is barely in it.
Right now, Claude seems to be setting the pace while everyone else is trying to keep up.
AI coding tools didn’t make me faster at first.
They made me faster at generating code I then had to fix.
The shift happened when I started treating the agent like a junior dev:
- Clear task
- Constraints
- Test requirements
- Review the output
Now it’s actually useful.
AI models are getting smarter.
Cloud infrastructure is getting more expensive.
The companies that figure out how to run powerful agents cheaply and reliably will dominate the next decade.
Simple as that.
2026 is the year many enterprises stop thinking “cloud-first” and start thinking “AI-first.”
It’s no longer about lifting and shifting apps.
It’s about designing infrastructure, data platforms, and operations around AI workloads from day one.
Are you still cloud-native, or are you becoming AI-native?