research/the-solo-stack

The solo stack: shipping production software as a team of one

March 22, 2026 · Software engineering · Product · Strategy

The cost of building software collapsed faster than most organizations have noticed. A disciplined solo builder with modern AI tooling now ships what took a funded team a year ago. This changes which architectures make sense.

Choose boring, managed, and observable

Every component a solo operator adopts is a component they’re on call for. Our stack bias: managed platforms over self-hosted anything, boring databases over interesting ones, and static-first delivery wherever the product allows it. Cloudflare’s edge platform, for instance, turns hosting, CDN, and CI into a git push — that’s an entire ops role deleted.

The bottleneck moved

Writing code was never the hard part; now it’s visibly not the hard part. The constraint is judgment: what to build, what to refuse to build, and when a feature’s maintenance cost exceeds its value. AI made implementation cheap and made taste expensive.

What doesn’t change

Production discipline doesn’t scale down. Backups, error tracking, uptime monitoring, and a written runbook matter more for a team of one, not less — there’s nobody else to remember. The solo stack isn’t about doing less engineering. It’s about spending all of it where users can feel it.

← All research