There is a problem that most businesses building with AI agents run into fast. The agent can answer questions. It can execute tasks. But it has no idea how your company actually operates. It does not know who is on the project, what was decided in last Tuesday’s call, or that the term “client onboarding” means something specific to your team and something different to your finance system.

Microsoft has been working on this problem for a while. At Build 2026 on June 2 in San Francisco, it shipped the answer: the Work IQ APIs, a developer-facing intelligence layer that gives AI agents real-time, semantic access to how a business works.

General availability launches June 16, 2026. Public preview is available now.

What Work IQ Actually Is

Work IQ is not another productivity dashboard. It is a set of APIs that sits between your AI agent and Microsoft 365, pulling in context from email, calendar, meetings, chats, files, and collaboration patterns — and turning all of that into something an agent can reason over.

The practical outcome: when an agent needs to know who owns a project, what was agreed in a meeting, or what the standard process is for a particular task, it can ask Work IQ instead of guessing.

Microsoft organised the API surface into four domains:

Chat — Conversational access to organisational knowledge, letting agents ask questions in natural language and get answers grounded in real company data.

Context — Semantic indexing of the people, teams, projects, and workflows that make up how your organisation operates day to day.

Tools — Ten generic action verbs (create, update, search, summarise, and so on) that agents can use to interact with Microsoft 365 systems without custom integration work.

Workspaces — Secure sandboxed environments where agents can do intermediate processing without exposing data outside the tenant boundary.

The Numbers That Matter for Builders

Microsoft shared two performance figures that are worth paying attention to if you are building or evaluating enterprise AI:

• 2x faster processing compared to traditional Microsoft 365 APIs
• 80% fewer tokens compared to conventional approaches for the same context retrieval

Token efficiency matters more than people realise when you are running agents at scale. Every call to an LLM costs money. Every extra token in the prompt adds to that cost. If you are running agents that check in on workflows hundreds of times a day across a team of fifty people, the difference between efficient and bloated context retrieval adds up quickly.

The pricing model is consumption-based, billed through Copilot Credits, with a new cost management dashboard in the Microsoft 365 admin centre so IT teams can see what agents are spending.

Why This Matters for Enterprise AI Deployment

The missing piece in most enterprise AI deployments right now is not model quality. The models are good enough. The missing piece is grounding — making sure the agent knows enough about the specific context of your business to be useful rather than generic.

Work IQ is Microsoft’s answer to that problem, built directly into the infrastructure that most large organisations already use. If your business runs on Microsoft 365, your agents will now be able to access a structured representation of how your company actually works: who reports to whom, what the current project status is, what deadlines are coming up, and what decisions have already been made.

Security was clearly a design priority. All data stays within the Microsoft 365 tenant boundary with full audit logging. For regulated industries and enterprises with strict data governance requirements, that matters. The Work IQ APIs do not require data to leave the tenancy, which removes one of the main objections to deploying AI agents in finance, legal, healthcare, and government settings.

The Bigger Picture at Build 2026

Work IQ sits inside a broader framework Microsoft calls Microsoft IQ — an umbrella that also includes Fabric IQ (the data and operations layer we covered in April), Foundry IQ (institutional knowledge and procedures), and Web IQ (real-time internet search for agents).

The architecture is deliberate. Microsoft is positioning IQ as the context operating layer for enterprise agents — the thing that makes agents know what they are working with, who they are working for, and how the business expects things to be done.

That positions Microsoft against every other platform competing for the enterprise agent market: Salesforce Agentforce, ServiceNow, and standalone agent platforms. The argument Microsoft is making is that if your organisation already lives in Microsoft 365, giving your agents access to Work IQ is the lowest-friction path to agents that actually understand your business.

What This Means for Business

For business owners and IT leaders evaluating AI agents, Work IQ changes the calculus on build versus buy. Agents built on top of this infrastructure can skip months of custom context-wiring work because the connection to your organisational data is handled at the API layer.

For teams that have already deployed agents and found them too generic or too slow to trust, Work IQ is worth evaluating as a grounding layer. The efficiency gains — 2x speed, 80% token reduction — translate directly to lower cost per agent interaction and faster response times.

For organisations that have held back from agentic AI because of data governance concerns, the tenant-boundary architecture removes what has been one of the most common blockers.

The real test is whether the context layer delivers on its promise in production. June 16 is the moment where that moves from demo to deployment.

---

Want the practical version of this? The free Working With Claude field guide covers the full Claude ecosystem, Claude Code, and how to roll it out across a real business. Download it here.