The Senior Leadership Team that ran Microsoft for decades—powerful executives each owning product divisions—has reportedly been dissolved. In its place: a governance group of five, an engineering group of roughly thirty-five working in close coordination, and a three-person Copilot team with a weekly standup directly with the CEO. Satya Nadella has called Microsoft's size "a massive disadvantage" in the AI race. Most coverage read this as executive shuffling, but I think this points towards something else.
Conway's Law says organizations often end up designing systems that mirror their own structures. That's how Microsoft ended up structured around software products: Windows, Office, Azure, Dynamics, Xbox—each division owning a clearly bounded domain.
In 1943, referring to the need to rebuild the House of Commons after the wartime bombing, Churchill had a prophetic observation: "We shape our buildings, thereafter they shape us." I believe agentic AI calls for equally deep changes—not just in the tools organizations use, but in the structures through which they coordinate, decide, and create. AI agents don't respect product boundaries. They move across applications, data, workflows, business functions.
For decades we built ERP systems, CRM platforms, collaboration tools. Those tools shaped our processes, hierarchies, and reporting structures in return. Today we're building agents and multi-agent systems. What kinds of organizations will those ultimately shape?
Most companies still consider AI as a tooling question—which model, which vendor, which budget. The Microsoft example points toward something far more radical—though whether an organization of that size can negotiate such a turn remains to be seen. Nadella wrote it into the announcement: "Our org boundaries will simply reflect system architecture and product shape."
That's Conway's Law applied to an entire company.
