Microsoft Work IQ API Reaches GA via A2A and MCP
Key insights
- Work IQ API goes GA June 16, 2026, accessible via A2A, redesigned remote MCP server, and REST API on a consumption billing model.
- Ten generic tools for fetching, creating, and updating Microsoft 365 data let agents act on organizational context without building retrieval pipelines.
- A Rego-based policy engine and getSchema endpoint enable runtime data discovery within per-user access scopes.
Why this matters
Microsoft decoupling Work IQ billing from M365 Copilot licensing removes the most common enterprise procurement barrier to embedding organizational intelligence in third-party agents. The addition of A2A alongside MCP and REST means agent frameworks built on any protocol can access the same organizational data layer, reducing fragmentation in the agentic middleware stack. Real production deployments at SLB, HP, and Miro validate that mixing domain-specific data with M365 context produces working enterprise agents, giving new adopters concrete reference architectures to follow.
Summary
Microsoft's Work IQ API reaches general availability on June 16, 2026, via three integration paths: Agent-to-Agent (A2A), a redesigned remote MCP server, and a REST API.
Four components power the platform: Chat for conversational intelligence, Context for assembling grounded organizational data without orchestration overhead, ten generic fetch/create/update tools for M365 data, and Workspaces via SharePoint Embedded for persistent agent workflows. Agents discover data schemas dynamically at runtime through a getSchema endpoint; a Rego-based policy engine enforces per-user access boundaries.
Essentially: (Microsoft, SLB, HP, Miro) already running this in production.
- SLB combined subsurface domain data with M365 context through its TELA agentic assistant.
- HP connected multifunction printer capabilities to Microsoft 365 Copilot integration.
- Miro pulled organizational context into its visual canvas to eliminate collaboration silos.
Billing is consumption-based and independent of Microsoft 365 Copilot licensing, with enterprise governance controls available in the M365 Admin Center.
Potential risks and opportunities
Risks
- Enterprises adopting the Rego-based policy engine inherit a dependency on Microsoft's policy update cadence; breaking policy changes could silently restrict agent access to organizational data without customer notice.
- SLB, HP, and Miro's live deployments are now coupled to Work IQ API versioning; undocumented breaking changes post-GA could disrupt production agentic workflows before stable versioning contracts are established.
- Consumption-based billing without disclosed per-call pricing makes enterprise cost forecasting difficult; high-volume agent deployments could generate unexpected charges before IT governance controls are fully configured in the M365 Admin Center.
Opportunities
- ISVs building vertical agentic solutions in energy, manufacturing, and SaaS can now ship M365-grounded agents without requiring customers to hold Copilot licenses, expanding the addressable market.
- Agent framework vendors can target the A2A and MCP endpoints to offer native Work IQ integrations, gaining a differentiation point in enterprise agent tooling.
- The M365 Admin Center governance expansion creates a consulting and implementation opportunity for system integrators helping large enterprises configure per-agent cost caps and policy rules at scale.
What we don't know yet
- Specific consumption pricing per API call or billing unit is not disclosed in the article.
- Whether the Rego-based policy engine supports customer-authored policies or only Microsoft-defined rule sets is not addressed.
- The article does not detail SLA uptime commitments or rate limits for GA production use.
Originally reported by devblogs.microsoft.com
Read the original article →Original headline: Microsoft Work IQ API Reaches General Availability — Semantic M365 Intelligence Layer for Enterprise AI Agents Ships With A2A, MCP Server, and Copilot Credits Billing