Image a Monday administration assembly. The gross sales director reviews quarterly income at 14.2 million. Finance has 13.6. The regional lead has a 3rd quantity from a spreadsheet up to date “final week, I believe”. Somebody asks the corporate’s new AI agent, and it confidently provides a fourth. The primary twenty minutes go to arguing about which quantity is true. No one discusses what the quantity means or what to do about it.
The primary twenty minutes go to arguing about which quantity is true. No one discusses what the quantity means or what to do about it.
Now replay the identical assembly. Everybody, together with the agent, will get the identical quantity from the identical ruled definition of “income” that the finance reviews use. The agent additionally is aware of the contract phrases behind the 2 largest offers, who has been engaged on the stalled ones, and {that a} key competitor introduced a worth reduce yesterday.
The assembly begins with the query that issues: what can we do subsequent?
It’s the identical AI mannequin in each conferences. The distinction is the context.
Enterprise AI discussions have circled round this query:
Which mannequin ought to we use?
That query is shortly shedding its weight. The main fashions can be found to everybody, usually by way of the identical platforms, and a brand new one arrives virtually each month. Selecting a mannequin nonetheless issues, however it’s not a long-lasting benefit.
The query that issues now could be:
What does our AI really learn about our enterprise?
Microsoft states this very straight: your organization’s IQ is the collective intelligence locked in paperwork, emails, conferences, operational knowledge and the reside internet, and that’s the place the true aggressive edge lives. A competitor can use the identical mannequin tomorrow. It may well’t use your buyer historical past, your contracts, your approach of defining a “key account” or the information in your individuals’s heads.
Fashions are in every single place. Your context isn’t.
That’s the thought behind Microsoft IQ.
Microsoft IQ is described because the enterprise intelligence layer of the Microsoft stack. It provides Microsoft 365 Copilot and AI brokers a shared, repeatedly up to date understanding of the group.
In observe, it brings collectively 4 varieties of information that each group already has however hardly ever connects:
- Work IQ – how individuals work, and the way the enterprise runs. Emails, conferences, chats, recordsdata, individuals and collaboration patterns, and now additionally enterprise functions: data, relationships and processes from Dynamics 365 and Energy Apps. Who’s engaged on what, what was determined, what’s prime of thoughts, and the place every buyer, case or order stands.
- Material IQ – how the enterprise is doing. Enterprise knowledge from Microsoft Material and Energy BI, raised from tables and columns into the language of the enterprise: prospects, orders, shipments, property, and the principles and metrics that join them.
- Foundry IQ – what the group is aware of. Insurance policies, contracts, manuals, tips and different authoritative paperwork, organized as reusable information bases that brokers can draw on.
- Net IQ – what is occurring outdoors. Contemporary info from the online: information, pages, photographs and movies, so brokers aren’t caught at their coaching cutoff.
None of those is a single new product to purchase. Collectively they describe how context reaches AI throughout the Microsoft stack. The excellent news is that a lot of the uncooked materials is already there, sitting in Microsoft 365, Energy BI, SharePoint and Material.
Microsoft now presents Microsoft IQ as one of three platform pillars: Copilot for individuals, Microsoft IQ for intelligence, and Agent 365 for belief and governance. In Microsoft’s phrases, it’s the platform to compound your intelligence from inside whereas defending what makes the enterprise totally different.
Context turns into a strategic asset. The standard, construction and possession of enterprise information now straight form the standard of AI outcomes. It’s not simply “info administration”. It’s a aggressive issue. Microsoft’s recent research with Keystone throughout greater than 200 enterprises factors the identical approach: the businesses furthest forward construct a shared “information backbone” that holds the context brokers use and the teachings from their work, so fixing one agent’s mistake can enhance the others. The SharePoint information base, curated by Copilot on prime of Work IQ, is an early instance of such a backbone at staff degree, as described in an earlier put up, SharePoint Knowledge Base – Team Memory, Curated by AI.
Current investments turn out to be AI investments. Energy BI semantic fashions, SharePoint libraries and Microsoft 365 content material all flip into gas for AI. In accordance with Microsoft, Energy BI accounts for over 90% of semantic fashions in manufacturing, and people fashions can be utilized as the start line for a enterprise ontology.
Construct as soon as, reuse in every single place. As a substitute of each AI undertaking constructing its personal connection to the identical knowledge, information could be organized as soon as and shared throughout many brokers. Fewer duplicate initiatives, quicker supply and extra constant solutions.
Shared definitions turn out to be a management matter. When AI solutions from “the” definition of income, buyer or lively undertaking, somebody has to resolve what that definition is. That isn’t an IT choice. It’s a enterprise choice.
Governance travels with the context. Permissions, sensitivity labels and entry guidelines have to comply with the information wherever brokers use it. Microsoft is constructing this into every layer, nevertheless it nonetheless must be deliberate and owned.
- Copilot solutions enterprise questions from ruled Energy BI knowledge. Fabric IQ in Microsoft Copilot Chat and Cowork is generally available. Ask for the most recent gross sales forecast, and the reply is grounded in the identical semantic mannequin that powers the group’s reviews. For Material and Energy BI prospects, that is on by default, respects present entry controls and comes with out extra AI token prices. (Typically out there, September 2026. Integration with the brand new Copilot Code expertise is coming by way of the Frontier program.)
- Enterprise knowledge could be described in enterprise phrases. Material IQ itself is usually out there, together with Operations Brokers, which monitor reside situations towards enterprise guidelines and suggest or execute actions, and Graph for reasoning throughout relationships. The ontology, the shared mannequin of enterprise entities, relationships and guidelines, could be generated from present Energy BI semantic fashions in a couple of clicks. (Material IQ, Operations Brokers and Graph typically out there since June 2026. Ontology is in preview, with basic availability deliberate for 2027.)
- Firm information could be organized as soon as for a lot of brokers. Foundry IQ information bases are typically out there, with SLA protection, compliance certifications and an MCP server that makes them usable from virtually any agent platform. (Typically out there since June 2026.)
- Copilot Studio brokers can use the identical information. The Foundry IQ integration with Copilot Studio is generally available: one information base can serve a number of Copilot Studio brokers, with cited solutions and help for personal networking. (Typically out there, October 2026.) Work IQ is also generally available in Copilot Studio, so low-code brokers can draw on emails, conferences, recordsdata and enterprise app context. (Typically out there, November 2026.)
- One information base can draw from the opposite IQs. New information sources let a single Foundry IQ information base attain into Work IQ, Material IQ (together with knowledge brokers and ontologies), Azure SQL, uploaded recordsdata and MCP sources, with no customized integrations. (Preview.)
- Copilot and brokers perceive enterprise functions. Enterprise Purposes in Work IQ deliver data and processes from Dynamics 365 Gross sales, Dynamics 365 Buyer Service and Energy Apps into Copilot and brokers, with studying, creating and updating data below consumer affirmation. Extra on this under. (Public preview since September 30, 2026. Normal availability for Energy Apps model-driven apps deliberate for November 2026, Roadmap 571888.)
- Work context is offered to any agent. The Work IQ API grew to become typically out there on June 16, 2026. It’s licensed independently of Microsoft 365 Copilot and billed by utilization in Copilot Credit, with no separate subscription or per-user licence, so customized and third-party brokers can use the identical work context Copilot makes use of. Information and instruments from enterprise methods already related to Copilot by way of Copilot connectors are additionally changing into out there to brokers by way of the Work IQ APIs, with no separate setup and with present permissions and admin controls in place. (MC1538999, rolling out late November to late December 2026.) Brokers may even be capable to learn a consumer’s personal Copilot reminiscence, akin to customized directions and saved recollections, so their assistance is extra private. Entry is read-only, follows every consumer’s personalization settings, and the admin coverage is on by default. (MC1539004, rolling out December 2026.)
- Brokers can search work content material straight. The brand new Work IQ Retrieve tool lets brokers search e-mail, recordsdata, conferences, Groups messages and related sources, returning cited content material that respects every consumer’s permissions. (Public preview, October 2026.)
- Brokers can comply with conferences as they occur. Work IQ real-time transcripts give apps and brokers entry to Groups assembly transcripts whereas the assembly continues to be operating, for reside help and insights. (Roadmap 570651, preview since September 2026, basic availability deliberate for November 2026.)
- Brokers can work with SharePoint, not simply learn it. SharePoint tools in Work IQ give brokers greater than 60 SharePoint operations, from discovering and organizing content material to lists and approvals, with separate permissions for studying and for making modifications. They’re additionally rolling out to Copilot Studio. (Public preview, November 2026.)
- Brokers can attain the surface world. Net IQ brings recent internet content material to brokers whereas honouring writer preferences. It’s now open to all paid Azure accounts, with no waitlist and usage-based pricing, and it plugs into Foundry IQ as a information supply. (Accessible, November 2026.)
- Enterprise context could be shared with companions. IQ Sharing in Material lets organizations share knowledge, and the enterprise that means behind it, with prospects and companions whereas maintaining governance of the underlying property. (Preview.)
So there may be already so much to work with. And that’s not a full checklist – not by all means! That’s solely overlaying what has come out just lately.
Material IQ isn’t the one approach enterprise knowledge reaches AI. With Enterprise Purposes in Work IQ, knowledge and processes from Dynamics 365 Gross sales, Dynamics 365 Buyer Service and Energy Apps come into Copilot and brokers, with Dynamics 365 Finance and Operations rolling out as properly. The enterprise functions keep the methods of file. What modifications is that Copilot and brokers can now perceive them, along with emails, conferences and recordsdata.
Microsoft describes the idea in three parts:
- Mannequin the enterprise as soon as. A shared semantic mannequin tells each agent what the information means, how data relate and what the group’s personal phrases imply. It builds on what already exists within the apps (views, types and relationships) plus a enterprise glossary, so context isn’t rebuilt agent by agent.
- Flip the playbook into enterprise abilities. How the group works, akin to which standards matter and when approval is required, is captured as soon as as a reusable ability that any agent related to Work IQ can comply with, together with brokers in-built Copilot Studio.
- Act with governance in-built. Copilot and brokers present the proposed modifications, comply with any required approval and replace the data inside the consumer’s personal permissions, with Agent 365 offering oversight.
Microsoft’s personal instance is a buyer renewal. The chance seems to be wholesome in Dynamics 365 Gross sales. Dynamics 365 Buyer Service, nonetheless, exhibits an unresolved tools situation, and up to date buyer emails and a Groups assembly clarify that fixing it’s a situation of renewal. No single app exhibits that image. Work IQ connects it, and a renewal-readiness enterprise ability turns it into concrete subsequent steps.
A sensible solution to see the distinction between the 2 routes for enterprise knowledge:
- Material IQ provides the agreed analytical and operational image: the numbers, definitions, traits and relationships throughout the information property.
- Enterprise Purposes in Work IQ deliver the reside data and processes themselves: this buyer, this case, this order, the steps round them and the flexibility to replace them.
Most organizations will want each. The primary solutions “how is the enterprise doing?”, the second “what is occurring with this buyer, and what ought to occur subsequent?”
The motion aspect of this, and why permission to know isn’t permission to vary, was coated in an earlier put up, From Answers to Actions with Work IQ. Within the Microsoft IQ image, the identical functionality is what connects individuals’s work with the methods the place the enterprise really runs.
Till now, many AI undertaking has rebuilt the identical basis: hook up with SharePoint or Azure AI Search (or another central container with index), put together the paperwork, arrange search, add some enterprise knowledge from one other system and possibly add internet search on prime. Each staff solved the identical drawback once more, and each answer dealt with permissions barely in another way.
Foundry IQ modifications that sample. A corporation creates a information base that holds the information related to a site, akin to customer support, procurement or HR. That information base can draw from paperwork and likewise from Work IQ, Material IQ and Net IQ. Any variety of brokers can then use the identical information base, whether or not they’re in-built Copilot Studio, Microsoft Foundry or one other platform.
Behind the scenes, Foundry IQ works out how to reply. It breaks a fancy query into smaller ones, searches the related sources in parallel and returns a grounded reply with citations. Microsoft reviews that this strategy improved recall by as much as 54% in comparison with easy single-shot retrieval.
Three issues make this particularly fascinating from a enterprise angle:
- Reuse. A well-built procurement information base can serve a procurement assistant, a contract overview agent and a provider danger agent on the identical time.
- Safety on the knowledge layer. Doc-level permissions are revered, and new preview capabilities prolong this to Microsoft Purview sensitivity labels and SharePoint permission sync. Brokers ought to solely see what the individual utilizing them is allowed to see.
- Openness. By way of the Foundry IQ MCP server, the identical information can serve brokers constructed on many frameworks, not solely Microsoft’s. That protects the funding if the agent panorama modifications, which it’ll.
One information base, many brokers. That’s the place the “construct as soon as, reuse in every single place” promise turns into actual.

Similar query, 4 layers
Here’s a situation that exhibits the distinction.
A key provider is delivering late. A procurement supervisor asks an agent: “Which provider delays ought to we escalate this week, and why?”
- Material IQ is aware of the enterprise state of affairs: which deliveries are late, which buyer orders they have an effect on and the way a lot income is in danger. Due to the ontology, the agent can comply with the chain from provider to element to order to buyer.
- Foundry IQ retrieves the provider contracts and finds the supply phrases and penalty clauses that apply.
- Work IQ exhibits who has already been involved with the provider, what was agreed in final week’s assembly and which account managers are getting questions from prospects. By way of enterprise functions, it additionally brings within the open provider and buyer circumstances from the CRM.
- Net IQ provides that the provider’s important manufacturing facility area has been hit by a port strike, in line with this morning’s information.
The reply isn’t only a checklist of late deliveries. It’s a advice: escalate these two, right here is the contract clause to check with, these are the individuals already concerned, and that is prone to worsen earlier than it will get higher.
With out the shared context, the identical reply would take a couple of individuals half a day of emails, spreadsheets and cellphone calls.
This isn’t solely a thought experiment. Microsoft shares customer examples that time the identical approach (outcomes as reported by Microsoft):
- Rockwell Automation constructed a upkeep assistant on Microsoft Foundry that brings collectively veteran engineers’ experience, manufacturing knowledge and machine manuals. At its Singapore manufacturing facility, machine downtime fell by 33%, and the time for brand new employees to be taught troubleshooting dropped from 9 months to a few.
- BNY runs greater than 300 options on its in-house AI platform, constructed with Azure, Microsoft Foundry, Copilot and GitHub Copilot, and plans so as to add Work IQ. Consumer transaction inquiries are resolved 80% quicker.
In each circumstances, the benefit comes from the group’s personal information, not from the mannequin.
The route is promising, however a couple of issues deserve consideration earlier than the context layer is switched on broadly.
Context high quality is the brand new knowledge high quality. AI constructed on outdated paperwork, duplicate variations and unclear definitions will give outdated and unclear solutions, simply quicker and with extra confidence. Content material possession and life cycle aren’t glamorous matters, however they now straight have an effect on AI outcomes.Your knowledge must be to ensure that high quality context!
Ontologies are a enterprise modeling train. Producing an ontology from Energy BI fashions ought to be executed. Agreeing on what “buyer”, “lively contract” or “at-risk order” means throughout enterprise models isn’t. That dialog wants enterprise homeowners, not simply knowledge engineers.
Enterprise knowledge comes with layered controls. Admin controls for enterprise functions in Work IQ (MC1518036, preview from mid-November) cut up the choice throughout three roles. AI admins change it on, platform admins select which environments participate, and app makers choose the person apps and knowledge. Every of these roles wants an proprietor earlier than Copilot and brokers begin reasoning over CRM and enterprise app knowledge.
Utilization-based prices want watching. A number of of those capabilities, such because the Work IQ APIs for customized brokers and Foundry IQ Serverless, are billed by utilization. Billing for Foundry IQ Serverless started on September 13, 2026. Brokers that run usually and extensively can add up, so price visibility belongs within the plan from day one. The Microsoft 365 admin heart now previews central Copilot Credit cost management, together with the Work IQ APIs, with spending limits and alerts. See additionally AI FinOps for Microsoft Copilot – a guide to Cost Management and the Cost Management API
Inner and exterior context are various things. Net IQ brings the surface world in, and that’s precious. It additionally means being clear about which solutions depend on inner, ruled information and which depend on the general public internet. The identical care applies contained in the Microsoft stack: connecting Work IQ to a Foundry IQ information base can add prices, and Microsoft notes that knowledge could also be processed outdoors the Azure compliance boundary, so compliance groups ought to overview it first.
Preview is preview. Among the most fascinating items, akin to ontologies, multi-source information bases and IQ Sharing, are nonetheless in preview. The ontology is deliberate to achieve basic availability in 2027. Till then, these are good for pilots and studying, however manufacturing plans ought to comply with the discharge standing.
The simplest approach in isn’t a expertise undertaking. It’s a enterprise query.
Decide one query that management repeatedly argues about. Income, forecast, buyer churn, supply efficiency, no matter causes the “whose quantity is true?” dialogue. Be sure the Energy BI semantic mannequin behind it’s the agreed one, and examine how Copilot solutions it now that Material IQ is offered in Copilot Chat and Cowork.
Decide one area the place information is scattered. Procurement, HR insurance policies, customer support or product documentation. Arrange it as one Foundry IQ information base and let two or three brokers reuse it.
Title homeowners for context. Somebody must personal the definitions, the information bases and the content material behind them, simply as somebody owns the processes.
Join it to motion fastidiously. Context is what makes brokers helpful, and helpful brokers will quickly be requested to behave. The query of what AI is allowed to vary follows naturally from what AI is aware of. These two discussions belong collectively.
Throughout latest years, enterprise AI has been a race for the very best mannequin. That race isn’t over, nevertheless it not decides the winner by itself.
What decides it now could be how properly AI understands this enterprise: its individuals, its numbers, its information and the world round it. Microsoft IQ is Microsoft’s reply to how that understanding will get constructed, shared and ruled, and Foundry IQ is the place a lot of it comes collectively.
A lot of the inspiration is already out there, and extra is on the best way. Microsoft’s roadmap already lists a “Work IQ: Microsoft IQ” merchandise, described as a unified intelligence layer that helps brokers uncover, purpose over and use organizational information, with a preview deliberate for November 2026. Microsoft has additionally stated that Ignite (17–20 November 2026) will broaden the story additional, so it is a good second to get the fundamentals so as earlier than the subsequent wave arrives.
This put up continues a collection on how Microsoft’s context layer is altering work. From Answers to Actions with Work IQ checked out what AI is allowed to vary, and SharePoint Knowledge Base – Team Memory, Curated by AI at how AI writes down what a staff is aware of. Microsoft IQ is the larger body round each.
The place does your group stand? Is your AI working from your context but, or nonetheless from generic information?
Standing as of November 5, 2026. Preview options, roadmap dates and pricing can change.
Microsoft IQ and the 4 IQs
Latest bulletins
- Introducing SharePoint tools in Work IQ – Microsoft SharePoint Weblog, November 3, 2026
- New and improved: Agents that are persistent, agents for customer conversations, and improved maker experience in Microsoft Copilot Studio – Microsoft Copilot Weblog, November 4, 2026
- The platform approach to becoming Frontier – Official Microsoft Weblog, October 27, 2026
- The AI advantage your company can keep – Microsoft Copilot Weblog, October 27, 2026
- AI that understands your business: Dataverse and Work IQ at PPCC 2026 – Microsoft Energy Platform Weblog, October 23, 2026
- Work IQ: the new retrieve tool – Microsoft 365 Developer Weblog, October 20, 2026
- Foundry IQ in Microsoft Copilot Studio is now generally available – Microsoft Foundry Weblog, October 1, 2026
- New Microsoft data innovations unlock what only your business knows – Official Microsoft Weblog, September 28, 2026
- Work IQ: Business and workplace intelligence in the flow of work – Microsoft Dynamics 365 Weblog, September 25, 2026
Message Heart and roadmap
- MC1539004 – Work IQ expands reminiscence capabilities to brokers (revealed November 5, 2026)
- MC1538999 – Copilot connectors in Work IQ APIs (revealed November 5, 2026)
- MC1518036 – Admin controls to handle Enterprise Purposes in Work IQ (revealed October 20, 2026)
- Roadmap 570870 – Work IQ: Microsoft IQ
- Roadmap 570651 – Work IQ: Actual Time Transcripts
- Roadmap 571888 – Energy Apps model-driven functions in Enterprise Apps in Work IQ
- Microsoft Fabric roadmap – IQ – ontology basic availability and associated objects
Earlier posts on this collection





