I recently read a McKinsey deck about the 24-hour sprint, and it's been stuck in my head ever since.
The idea is two shifts. Overnight, for sixteen hours, a factory of agents does the work: turning requests into requirements, checking the architecture, building and testing a first cut, then leaving a report behind for the morning. The next eight hours are ours, for the parts that still need a person. You review what got built, pick apart the risky code, fix what's weak, and get the agents ready before they run again that night.
I think this might genuinely be where the best companies end up.
But they're honest that it only works under certain conditions. You need to actually know what you're building. Everyone has to be on the same tech stack, not fifteen different ones. There's a clean, repeatable path from a requirement to working code. And the people who understand the business stay close to it the whole way, instead of checking out after the kickoff.
I read that list and agreed with every line. Which is exactly why I think only a small number of companies will ever pull it off.
I've never worked at a startup. But I have been on teams that had total ownership of what they built, where the product decisions lived inside the team. That was the only time any of this ever felt possible to me.
The trouble is how rare that is. The roadmap goes a bit fictional. Two teams end up on different stacks for reasons nobody remembers. The requirements scatter across Jira, Slack, three docs and someone's memory, and the person who could settle it is in another meeting. None of those were badly run places. It's just what size does to you.
So here's the thought I can't shake. Maybe the move is to stay small. Small enough that the team still owns the whole thing. As the agents get better, all the friction we've learned to live with starts to look really expensive, and a tiny team that can still see the whole business at once might just outrun a big one that can't.
It's a gut feeling more than anything. But I think the next few years will reward the teams small enough to still own what they build.