Project Copley: Building an Open Knowledge Layer for Work IQ and Cross-Platform Agents | The Synozur Alliance
2026-08-20 · Teams Webinar
Work IQ made Copilot context-aware by default across Microsoft 365, and brokered retrieval is now largely a solved problem. That success exposed a harder one, and it isn't Microsoft's alone to solve: an agent can retrieve a number, a policy, or a "process" with no idea whether it's current, correct, or safe to act on.
A group of us has been sketching an open, early-stage response — a thin, governed schema that would extend Work IQ into a consistent knowledge framework for Microsoft systems and cross-platform agents alike. It leans on work the community has already done rather than inventing another ontology: Google's Open Knowledge Format, JSON Schema and SHACL, MCP and Agent Skills, and open-governance patterns from the Open Compute Project.
Hosted by Synozur, this session is a working conversation, not a presentation. We'll spend the first fifteen minutes putting a rough draft on the table — a three-layer model of source pointers, curated knowledge with real owners, and a schema that keeps both honest — and the rest of the time on your reactions to it. The open questions are genuinely open: Is a new profile even warranted, or should this be folded into something existing? Where does it break in non-Microsoft estates? What have you already built that solves part of this?
Nothing here is settled, and it shouldn't be decided by any one company. If you've wrestled with knowledge governance for agents, you belong in this conversation — whether you want to help build it, poke holes in it, or tell us it's the wrong idea.