Microsoft IQ definitions

What Are Work IQ, Fabric IQ, Foundry IQ & Web IQ?

Microsoft announced three intelligence layers at Ignite 2025 and made them generally available under the Microsoft IQ name at Build 2026, where it also added a fourth, Web IQ, for live web grounding. Microsoft now groups all four under Microsoft IQ, its enterprise context layer for AI. Each one gives agents and copilots a different type of context. Here is what each layer does, what it connects to, what it does not replace, and when it matters.

Published: 2026-04-12 Last updated: 2026-11-17

What is Work IQ?

Work IQ is Microsoft's workplace intelligence layer. It continuously builds an understanding of how work happens across Microsoft 365 and connected business systems, with permission-aware governance built in, and gives agents that understanding as context and as tools they can act with.

What Work IQ provides

Microsoft organizes Work IQ into four areas. Chat handles conversational requests and agent-to-agent collaboration. Context assembles grounding from Microsoft 365 data such as email, meetings, chats and files, plus business data from Dynamics 365 and Power Platform: CRM records are generally available and ERP records are in preview. Tools give agents a small set of general actions across mail, calendar, files, people, chat and sites. Workspaces give agents a governed place inside your tenant to keep intermediate work. Memory across sessions is in public preview.

How agents reach it

Microsoft 365 Copilot uses Work IQ directly. Custom agents reach it through the Work IQ APIs, generally available since June 16, 2026, over A2A, a remote MCP server and REST, and through Copilot Studio and Microsoft Foundry. Every request runs as the user, and new permissions separate reading data from taking action.

How you pay

Microsoft 365 Copilot users get Work IQ in Copilot experiences. Custom and third-party agents, and users without a Copilot license, pay through usage-based billing in Copilot Credits, managed in the Microsoft 365 admin center.

What Work IQ does not replace

Work IQ does not provide business data semantics; that is Fabric IQ. It does not provide curated knowledge bases with citations; that is Foundry IQ. It does not know what is happening outside your organization; that is Web IQ. And it does not fix a broken workflow. It gives AI a clearer view of how work happens, including when the process itself is the problem.

When Work IQ matters

Work IQ matters most when Copilot gives generic answers, when agents need to act inside Microsoft 365 on someone's behalf, or when the problem lives in how people coordinate: meetings, handoffs, follow-ups and approvals.

What is Fabric IQ?

Fabric IQ is the part of Microsoft IQ that describes the state of your business. It lifts data in Microsoft Fabric up to the language of the business, so people and agents work from the same definitions of customers, orders, assets and the rules that connect them.

What Fabric IQ provides

Microsoft describes three layers: unified data in OneLake, business intelligence in Power BI semantic models, and operational intelligence in ontologies. The IQ workload contains the ontology (preview), Power BI semantic models, planning, graph (preview), data agents that answer questions for a domain, and operations agents that watch real-time data and recommend actions.

How agents reach it

Power BI semantic models in Copilot are generally available. Data agents can be published to Microsoft 365, Copilot Studio, Foundry and custom apps. Fabric IQ in Excel, connectors for ChatGPT and Claude, and sharing ontologies with partners were announced in preview at Ignite 2026.

Status and cost

Microsoft declared Fabric IQ generally available at Build 2026, but the ontology item is still in preview, and Microsoft Learn still labels the workload preview. There is no separate license; Fabric IQ runs on the Fabric capacity you already pay for.

What Fabric IQ does not replace

Fabric IQ does not replace Power BI; it extends semantic models into AI and operations. It does not replace data engineering or data quality work, because its value depends on clean, well-modeled data. And it does not retrieve unstructured documents; that is Foundry IQ.

When Fabric IQ matters

Fabric IQ matters most for organizations invested in Fabric and Power BI, when different teams define the same terms differently, or when agents need trusted business definitions to give the same answer twice.

What is Foundry IQ?

Foundry IQ is the managed knowledge layer of Microsoft Foundry. It gives agents curated institutional knowledge, such as policies, contracts and authoritative documents, through knowledge bases that several agents can share.

What Foundry IQ provides

A knowledge base defines which sources to query and how hard to reason over them. Sources include SharePoint, Azure Blob Storage, OneLake, Azure AI Search indexes and the web through Web IQ, with Work IQ and Fabric IQ available as sources too. Agentic retrieval breaks a question into sub-queries, runs them in parallel, reranks the results and returns one answer with citations. Knowledge graphs, announced at Ignite 2026, are in preview.

How permissions work

Foundry IQ can synchronize access control lists for supported sources, honor Microsoft Purview sensitivity labels, and run queries under the caller's Microsoft Entra identity, so an agent only returns what that person is allowed to see.

Status and cost

Knowledge bases have been generally available, with an SLA, since June 2026, and are reachable over REST, SDKs and an MCP server. Some features depend on the API version you use and remain in preview. Costs follow Azure AI Search and Foundry usage; Foundry IQ Serverless is due to reach general availability in December 2026.

What Foundry IQ does not replace

Foundry IQ does not replace structured analytics; that is Fabric IQ. It does not replace collaboration context; that is Work IQ. And specialized retrieval needs, unusual formats or non-Microsoft sources may still call for custom retrieval alongside it.

When Foundry IQ matters

Foundry IQ matters most when agents must answer from approved documents, cite their sources, and respect permissions, especially in regulated work where a confident wrong answer is expensive.

What is Web IQ?

Web IQ is the newest Microsoft IQ layer. It is a set of AI-native APIs that give agents fresh, real-world information from the public web, returned as ranked, citation-ready context that drops straight into a model's context window.

What Web IQ provides

Results from web pages, news, images and video, with titles, links, snippets, timestamps and provenance. Microsoft reports a 164-millisecond 95th-percentile response time and coverage in more than 100 languages and markets, and says Web IQ already grounds Microsoft Copilot and ChatGPT.

How agents reach it

Web IQ is model-agnostic. Agents call it over REST, MCP or an SDK, and Foundry IQ can use it as a knowledge source. Microsoft positions it for agents and multi-step workflows, separate from Grounding with Bing, which serves traditional search experiences.

Status and cost

Web IQ arrived at Build 2026 and, since Ignite 2026, is available to every Azure customer. It needs a paid Azure subscription (Azure credits do not work), is deployed as an Azure resource, and is metered per call, from $1 per 1,000 calls.

What Web IQ does not replace

Web IQ does not know your business. It cannot tell an agent your contract terms, your metrics or who owns a decision; that is Foundry IQ, Fabric IQ and Work IQ. And web results are evidence, not truth: agents still need to show their sources.

When Web IQ matters

Web IQ matters when an agent's answer depends on something that changes outside your walls: a regulation, a supplier's news, a market price, a product recall.

A practical example: how all four layers work together

Consider a supply chain organization trying to understand delivery delays.

Fabric IQ sees the data

Fabric IQ detects that certain suppliers are delivering late more frequently and that regional on-time delivery rates have dropped. Because it understands the business ontology, it can identify which materials, products, and customer segments are affected.

Foundry IQ provides the knowledge

Foundry IQ retrieves supplier contracts, service-level agreements, and penalty clauses from the organization's document repositories. It understands the contractual obligations around delivery timelines and can surface relevant terms with citations.

Web IQ adds the outside world

Web IQ picks up news of a port closure and severe weather in the region where two of the late suppliers ship from, with sources the agent can cite.

Work IQ reveals the human pattern

Work IQ observes that operations teams are spending excessive time in email chains and delay review meetings, manually chasing supplier updates. It detects that the human workaround has become the process.

The result: an AI agent that understands where and why delays occur (Fabric IQ), what the contractual obligations are (Foundry IQ), what is happening outside that explains it (Web IQ), and how people are currently handling the problem (Work IQ) can recommend escalations, draft communications quoting relevant SLA terms, and flag process improvements. That is the combined intelligence layer at work.

Part of a larger system

Why these questions matter beyond Microsoft

Understanding Microsoft's intelligence layers is the fourth stage of the AI Advantage Framework. The first three stages address which work to fund, whether the information is usable, and whether the workflows can execute. Platform decisions make more sense when those foundations are in place.

AI Advantage Framework

AI Fit & Governance → AI-Ready Data → Operational AI → Microsoft Intelligence

Need help deciding where these layers fit?

Marquee Insights helps Microsoft-heavy teams separate what is ready now from what requires patience, and decide where to invest first.

Treb Gatte

Founder & CEO, Marquee Insights

Dual Microsoft MVP: Microsoft Fabric & Microsoft Foundry

One of four people worldwide with dual Microsoft MVP designation across data and AI platforms. 24 years of enterprise experience at Microsoft, Starbucks, Wachovia, and Inmar.

Learn more about Marquee Insights →