On June 2, 2026, at its Build developer conference, Microsoft introduced a family of grounding layers for AI agents under the “IQ” brand. The June 2, 2026 date is stated on the official Microsoft blog. The lineup includes Work IQ, described as a workplace intelligence layer that captures how work happens across Microsoft 365 and exposes it to agents through APIs, with general availability stated for June 16, 2026 (date traces to the primary). It also includes Web IQ, an AI-first web search stack described as model-agnostic and MCP-native that Microsoft says returns results at roughly 2.5x the speed of competitors (claim traces to the primary).
Microsoft also announced Microsoft IQ, generally available across GitHub Copilot, Microsoft Foundry, and Copilot Studio to ground agents in both world and enterprise knowledge, and Microsoft Scout, a personal work agent for Frontier customers that uses Teams and Outlook to handle meeting prep and scheduling (both trace to the primary). These platform pieces are distinct from the seven in-house MAI models Microsoft unveiled at the same event.
For business and technical leaders, the IQ layer matters because it targets the hardest part of deploying useful agents: grounding them in real enterprise context and reliable web data. By making this layer MCP-native and model-agnostic, Microsoft is positioning the context and tooling around agents, rather than any single model, as the durable competitive moat.