Microsoft to Make Work IQ APIs Generally Available on June 16
On June 2, 2026 (local time), Microsoft officially announced that the "Work IQ APIs"—which let external apps and agents tap into Microsoft 365's workplace intelligence layer, Work IQ—will reach general availability (GA) on June 16. The same day, the official @Microsoft365 account announced "Work IQ APIs are here" alongside developer resources, signaling the move toward production use. As of the post, the APIs were in public preview, with GA enabling production deployment.
Work IQ is a foundational Microsoft 365 layer that builds semantic context understanding from data including email, calendar, meetings, Teams messages, OneDrive/SharePoint files, people, organizational patterns, and line-of-business (LOB) systems. This allows Copilot and custom agents to draw directly on "work context, intent, and organizational signals" rather than raw data. According to the official blog, the APIs span four domains: "Chat," which provides programmatic access to Microsoft 365 Copilot intelligence including cited responses; "Context," which retrieves context and source data optimized for agents; "Tools," which consolidates actions such as sending email, scheduling meetings, and uploading documents into roughly 10 general-purpose tools (verbs); and "Workspaces," which retains intermediate state, data, and memory within tenant boundaries in SharePoint Embedded for long-running agents. Supported protocols include Agent-to-Agent (A2A), Model Context Protocol (MCP), and REST.
The backdrop is that the traditional Microsoft Graph API centered on human-oriented data retrieval, requiring custom pipelines—building vector DBs, syncing data, managing permissions—for agents to assemble context. Work IQ absorbs that work at the platform level, offering an intelligence layer agents can use "out of the box." It also opens up to third-party and custom agents the layer already operating behind Microsoft 365 Copilot. Compared with rivals such as OpenAI's Custom GPTs and Anthropic's tools, Microsoft positions enterprise-grade governance—delegated authentication via Microsoft Entra ID that automatically applies permissions, sensitivity labels, and compliance policies—as a built-in differentiator. It is also framed as part of the broader Microsoft IQ intelligence strategy.
Billing is consumption-based via Copilot Credits, with Chat/Context variable and Tools a fixed component. It uses a common currency shared with offerings like Copilot Studio, with no dedicated Work IQ SKU. It is said to be usable regardless of whether a user has a Copilot license, aiming to lower the barrier to entry for developers and IT admins. The tool surface consolidates hundreds of operations into 10 general-purpose tools (verbs such as fetch/create/update plus resource path), and trims file record strings, message IDs, and app IDs to reduce token consumption. On governance, the Microsoft 365 admin center gains a new cost-management dashboard for monitoring credit usage, setting spending caps, and managing at the tenant/group/user level. Samples and CLI are available in preview on GitHub via work-iq-samples, and the Work IQ CLI (preview) enables IDE integration as a local MCP server. Technical blogs also highlight partner cases such as SLB's TELA agent, HP's printer integration, and Miro's canvas integration.
Initial reactions centered on official accounts and MVPs spreading the news, with detailed end-user feedback still limited given the preview stage. @MoeSbaiti, who covers AI signals, praised Work IQ for "solving the custom pipeline problem," reducing token consumption by shifting context processing to the runtime side, simplifying the agent surface with 10 general-purpose tools, and eliminating the need for vector DBs and ingestion. At the same time, he noted that the move to Copilot Credits billing could increase visibility and management burden, and that dashboard monitoring could be a challenge especially for small and medium businesses (SMBs). Overall, expectations around efficiency gains and inherited security dominate, but concrete verification on actual token reduction, cost outcomes, and limits such as rate limits and supported data scope is expected to grow after GA.