This episode starts with a spacecraft lost to a unit of measure. Two competent teams, two different standards, and a seam between them that nobody owned.
This episode starts with a spacecraft lost to a unit of measure. Two competent teams, two different standards, and a seam between them that nobody owned.
A few episodes ago, I promised an episode on governing a fleet of agents. Today our Lab runs about a dozen of them in production, with coding agents behind them drafting fixes the same day a problem is reported. So the question every leader eventually asks becomes real: when agents do the work, and other agents change how they do it, who owns the standard?
I start somewhere uncomfortable. For the first six months, the biggest bottleneck in our program was me — not because anything was going wrong, but because almost everything ran through me. The lessons from more than two thousand changes lived on my laptop. Even the AI coding assistant on my machine was better at our system than the one on anyone else's.
I share how we got that knowledge out of one person and into the system, the loops that feed a single nightly gate, why we moved agent definitions out of a point-and-click screen, and how the developer's job is moving up. And I'm candid about where the gate has a limit.
If you take one thing away: if the knowledge lives on one laptop, you don't have a standard. You have a person.
This continues the Lean AI arc, following Respect for People.
I'm Chris Bradley, Chief AI Transformation Officer at Veritiv and head of the AI Transformation Lab. Subscribe for new episodes biweekly.