Everyone’s Building the OS. Who’s Writing the Rules?

Sycamore Labs just raised $65 million to build a "trusted agent operating system."

Good. The more infrastructure that ships, the clearer the pattern becomes.

Microsoft is building Agent 365 — observe, govern, secure. AWS has Cedar for policy enforcement. CrowdStrike, Astrix, 1Password, and Google all shipped agent identity frameworks in the same week at RSAC. Now Sycamore wants to be the operating system that ties it all together.

That’s seven enforcement platforms in thirty days. Billions of dollars. Serious engineering teams. And every single one of them assumes the same thing: that the governance policies already exist.

They don’t.

Open Agent 365. The first screen asks you to define guardrails. What guardrails? For which workflows? With what boundaries? Based on what operating model?

Most teams don’t have answers to these questions. Not because they’re behind — because nobody asked them to think about it until the enforcement tools showed up.

This is the empty library problem at scale. The infrastructure is magnificent. The shelves are bare.

Enforcement without intelligence is a security camera watching ungoverned agents do whatever they were last told to do.

The question was never "how do we enforce the rules." The infrastructure teams have that covered. The question is: who helps you write the rules in the first place?

One workflow. One set of boundaries. Rules that enforcement platforms can actually execute.

That’s what’s missing.