Why Hire an Engineer Who Won't Use AI?

I'm using Cursor with Opus 4.5 to build new SaaS MVPs, and it's easier than ever before. Previously, the code quality wasn't good enough. But after the launch of Gemini 3, Opus 4.5, and GPT 5.2, we reached the inflection point. I'd rather work with AI to write code now.

The New Calculus

We still need senior engineers to architect and talk to the AI. But why would I hire a junior engineer today? Only if the person is hyperactive with five Claude Code terminal tabs and fully onboard with the latest AI tools. That's the edge juniors have over seniors who feel reluctant.

Boris Cherny and the Multi-Agent Future

Boris Cherny, who created Claude Code at Anthropic, wrote on Threads that he runs 5 to 10 agent sessions in parallel. Five terminal tabs for different tasks: feature development, tests, PR review, debugging, documentation. Plus additional browser sessions. Roger Wong's summary breaks down how he orchestrates it.

It feels like a lot of multitasking for a human. But it feels like the future. Why would you hire an engineer writing code manually when this is possible?

Why Hire Anyone Not Using AI?

The question isn't theoretical. As we've written about what happens when a CTO makes AI non-negotiable, some leaders have already drawn the line. The inflection point changes the calculus for everyone.

You need people who architect. You need people who direct. And you need people who execute at the speed of multiple agents, not one pair of hands.

What This Means for Founders and CTOs

Hiring calculus has shifted. Invest in people who run with AI. Avoid those who won't. When kids can ship revenue in weeks with the same tools (as my own kids did), the bar for who adds value has moved.

The engineers who compound are the ones who multiply themselves with AI. The rest become a bottleneck.