Human infrastructure for the AI era

You know what happens when you inherit bad architecture. You spend years debugging it. Mismatched dependencies, unclear ownership, no documentation, no tests. The system does what it was designed to do. You just weren't the one who designed it, and now it's your problem to maintain it. Or worse yet - build on it.

Most engineering teams have the same issue with their human infrastructure. They inherited it. They never designed it. And it's compounding like tech debt.

As AI accelerates and raw coding skill gets commoditized, the moat is shifting. The organizations that win aren't necessarily the ones with the best models. They're the ones where engineers can actually work together — with each other and with designers, strategists, and clients. To know when there's enough trust to challenge a bad idea, enough clarity to move fast without creating wreckage between people, and enough accountability to keep delivery honest.

This talk names three failure patterns that quietly take down high-performing engineering teams — the kind that look fi ne on the surface until they're not. It also offers three specifi c, repeatable practices that any engineer or tech leader can begin this week. Tiny, survivable interventions threaded into everyday work.

You've been thinking about this as a people problem. We're going to argue it's an infrastructure problem. And infrastructure problems have engineering solutions.

Key takeaways

• Why culture is infrastructure — and why inheriting it is just as dangerous as inheriting bad code
• The three failure patterns most common in engineering organizations, and how to spot them before they crash production
• Three specifi c behavioral practices that cost nothing but attention — and compound over time
• A systems lens for diagnosing where your team's human architecture is load-bearing, and where it's quietly failing