Microsoft IQ Is Now Generally Available: One Context Layer to Ground All Your Agents
Microsoft IQ — combining Work IQ, Fabric IQ, and the new Web IQ — is now GA across GitHub Copilot, Foundry, and Copilot Studio.
One of the more practically significant announcements out of Microsoft Build 2026 is not a new model or a headline agent. It is a plumbing layer, and that is precisely why it matters.
Microsoft IQ is now generally available across GitHub Copilot, Microsoft Foundry, and Copilot Studio. It combines three distinct intelligence layers into a single context foundation that any agent can draw on: Work IQ (signals from Microsoft 365), Fabric IQ (structured business data), and the newly announced Web IQ (live web grounding). The goal is straightforward: give agents a deep, consistent understanding of how your organisation actually operates, without requiring developers to wire everything together themselves.
What the three layers actually do
Work IQ is the organisational memory layer. It surfaces how work happens across Microsoft 365, connecting people, emails, documents, meetings, and the relationships between them. From June 16, the Work IQ APIs are generally available, giving developers programmatic access to this intelligence. Practically, this means an agent can understand not just what a document says, but who owns it, who is involved in the related project, and what conversations have happened around it. The APIs also reduce token overhead significantly. Rather than exposing hundreds of data-specific tools to a model, Work IQ collapses tool calling into around 10 generic tools with progressive disclosure via MCP, so agents retrieve what they need without burning through context windows.
Fabric IQ provides a shared semantic foundation over structured business data. It includes an Ontology layer (currently in preview) that gives agents shared business semantics, helping them understand the relationships between people, data, workflows, and operations. This matters because structured data in most enterprises is siloed and inconsistently named. A semantic layer that an agent can reference consistently is genuinely useful.
Web IQ is new at Build 2026 and the most eye-catching of the three. It is an AI-first web search stack, model-agnostic and MCP-native, designed to give agents access to live information from the web. Microsoft says it returns relevant passages at nearly 2.5 times the speed of the next best alternative, with sub-200ms grounding latency. For agents that need to reason about current events, recent pricing, or anything that changes faster than a training cutoff, this fills a real gap.
The Foundry IQ piece: no more custom RAG pipelines
For developers building on Microsoft Foundry, the practical benefit lands through Foundry IQ, which is also now generally available. Building an agent that reasons over enterprise data has traditionally meant assembling a retrieval-augmented generation pipeline from scratch: chunking, indexing, retrieval logic, and a separate integration for every data source. Foundry IQ replaces that with a single SLA-backed retrieval endpoint that unifies Work IQ, Fabric IQ, Azure SQL, File Search, and MCP sources.
This connects directly to Toolboxes in Foundry, so agents can tap enterprise data without custom plumbing. A Serverless tier is available in public preview for teams that want to start without a full infrastructure commitment.
What this means for you
If you are building agents, the immediate benefit is that the grounding problem gets significantly simpler. You do not need to decide which data sources your agent needs access to and then build individual connectors for each one. You point your agent at Microsoft IQ and let the retrieval layer sort it out, with governance and permissions inherited from your existing Microsoft stack.
If you are in IT or running Microsoft 365, the Work IQ APIs going GA on June 16 are worth watching. Agents built on this layer will understand your organisation’s context from the start rather than needing to be trained or configured to understand who works with whom, or which document is the current version of a thing. That is a meaningful reduction in the setup cost for new agent deployments.
For governance and admin teams, there is a relevant detail on billing. Consumption-based pricing for Work IQ APIs and agent runs is turned off by default. Enabling it is where you set named billing policies, define which services and agents can spend credits, and configure per-user caps and alerts. Microsoft has made this an active opt-in rather than something that silently accrues costs.
The bigger picture
Microsoft is not presenting IQ as a standalone product. It is the context layer underneath the broader Foundry platform, which also gained Hosted Agents (reaching GA within 30 days), Microsoft Agent Framework 1.0, and Agent Control Specification at Build. Together these are meant to move enterprise AI from experimentation toward production-grade operation: consistent grounding, managed runtime, observable execution via OpenTelemetry, and billing controls from day one.
The IQ Series learning repo on GitHub is the best place to get hands-on. It includes video episodes, Jupyter notebooks, and Azure deployment templates covering Foundry IQ, Work IQ, and Fabric IQ, which makes it a practical starting point if you want to understand how the pieces connect before committing to a build.
The Work IQ APIs go GA on June 16. Foundry IQ and Web IQ are available now.