E46 - AI-Native Ops: Making AI Safe for Production with William Collins

What happens when your “coworker” can generate code and changes faster than your team can review them, and production still has to stay up?

William Collins breaks down what AI-Native Ops looks like when you take reliability seriously: where reasoning should stop, where deterministic automation should begin, and how guardrails like compliance checks, version pinning, and controlled workflows keep AI from turning into outage fuel. Cory and William also dig into why context windows and tool sprawl matter in real systems, how protocols like MCP and agent-to-agent communication are shaping day-to-day automation, and why regulated environments can’t adop...