Work IQ is the intelligence layer that personalizes Microsoft 365 Copilot to you and your organization. It is the “brain” behind Copilot that understands context, relationships, and work patterns, so Copilot and agents can be faster, more accurate, and more secure than model companies that are built on connectors alone.
Work IQ is comprised of three tightly integrated layers – data, context, and skills & tools. We’ll examine the components in each of these layers, highlight some of what’s coming to make Work IQ even better in the coming months, and share example prompts you can try to see the power of Work IQ in action.
Multi-Model
Before we discuss the components within the layers of Work IQ it’s important to note our approach to the use of foundation models. Microsoft 365 Copilot brings leading models from multiple providers directly into Copilot experiences. We offer Copilot chat users a choice of foundation models from Open AI and Anthropic and will be bringing advanced models from other providers as well. This means users can access – and Work IQ will compliment – these models’ advanced reasoning and multi-step capabilities in their Copilot experiences, not just through specialized tools. Copilot applies the right model for the task and gives users the option to choose the model that best meets their needs.
Data
The foundation of Work IQ is its secure access to and understanding of both structured and unstructured data from Microsoft 365, Dynamics 365, Power Apps, and Power BI (coming soon), and other connected business systems that represent work happening across your organization.
A customer’s Microsoft 365 tenant data provides Copilot with a foundational understanding of individual and collective work. Centered around the users and groups in an organization, it includes the permission-based, information-protected content stored in SharePoint and OneDrive, including Word, Excel, PowerPoint and other file types, as well as Outlook emails, and Teams meetings and chats. The data contained in your Microsoft 365 tenant also includes rich metadata and signals that further describe patterns of action, activity, collaboration, and communication between users and groups over time.
Customers can ingest business data from other systems and line-of-business applications into their tenant using Copilot Connectors, enabling Copilot to reason over data that may reside in non-Microsoft systems. Customers can choose from hundreds of pre-built connectors or build their own custom connectors.
In addition to Microsoft 365 data, we are integrating Dynamics 365 and Power Apps data into Work IQ. Dataverse is the container for the structured datasets that power customers’ Power Apps and Dynamics 365 applications. Later this month, customers will see Microsoft 365 Copilot embedded within Dynamics 365 Sales, and Dynamics 365 Customer Service as an in-app experience like Copilot in Word, PowerPoint and Excel. With this access, Copilot will have the ability to reason across both Microsoft 365 productivity data and the business data generated by their systems of record, making it possible for Copilot to answer complex questions like “Help me evaluate how issues raised by my parts supplier in our Teams call last week might impact my inventory and sales in the coming months” and to provide very detailed, specific answer connecting business communications with business data. For more information on Work IQ in Dynamics 365, check out this blog with additional details. We anticipate broad access to Dataverse via Work IQ across all Microsoft 365 Copilot experiences in Summer 2026.
Context
Implicit grounding in Microsoft 365 data – and soon Dynamics 365 and Power Apps data - serves as the baseline for understanding work context. Work IQ expands this with an additional, always-evolving layer of insights that enhance the speed and accuracy of Copilot’s response to queries. Work IQ helps Copilot learn how people and businesses work – the skills they have, the projects that are important, the frequency of collaboration, who they work with and for, critical workflows, the velocity of communication and much more.
Copilot’s memory is designed to further enhance Copilot’s ability to tailor its experience to Copilot users in Copilot Chat and Copilot in the M365 apps. Memory is constructed from a combination of persistent, explicit memory and query-dependent, implicit memory.
Explicit memory is provided to Copilot by the user. A user can personalize Copilot by creating “Custom Instructions” – for example a user may manually add an instruction to “Only provide responses to prompts in the active tense”. Alternatively, a user can also create “saved memories”. For example, prompting Copilot to “remember that I do not like responses in the passive tense” will result in Copilot creating a saved memory “Prefers responses written in active voice; dislikes passive tense”. In each case, Memory is explicitly created by a user action. Copilot also creates implicit memory. To do this, it uses chat history to infer a durable body of insights. As the body of insights grows, Copilot can provide increasingly personal responses and actions.
Beyond chat history, we are also working on incorporating activity – like workflows - that help increase the fidelity of Copilot’s memory. In the coming months, we will start incorporating other activity patterns generated from all of your everyday apps including Teams, Outlook, Word, Excel, and PowerPoint. Stay tuned for additional updates on enhancements to Copilot memory.
Understanding of data that provides work context about your Microsoft 365 tenant data is enhanced by the semantic index. Rather than relying solely on keyword or lexical matching, the semantic index enables Copilot to perform meaning‑based retrieval of your data, producing a bounded set of relevant candidates for downstream processing. Data ingested into the customer tenant using Copilot Connectors is also included in the semantic index. Indexed data retains customers’ existing security, privacy and governance policies – including permissions, sensitivity labels, and tenant boundaries.
Finally, we’re adding business understanding to Work IQ to provide Copilot and agents with a more comprehensive view of customers’ Power Apps and Dynamics data in Dataverse. To do this, we’ve created a layer of semantic understanding, consisting of ontologies and glossaries that capture procedural knowledge from existing business workflows. The result: Copilot can have an expert understanding of the tasks performed by the people, teams, customers, suppliers, and other business entities working together to run your business.
Skills & tools
Work IQ includes agentic skills that provide specialized instructions to Copilot & agents. These skills are designed to help Copilot perform specific tasks with more speed and accuracy. Microsoft is continuously adding skills to Work IQ that enable Copilot to deliver experiences that are highly tailored for specific tasks like “schedule a meeting”, “find and retrieve data from an external source”, or “access meeting details and transcripts”. For example, we have deployed skills designed to optimize retrieval of work content in response to complex user queries. The result is improved Copilot responses to queries that may contain deeply sourced references to vague and hard-to-find archived information.
If skills describe what to do, tools do it. We’re developing custom toolsets designed to execute against the intent of the agentic skills that power Work IQ experiences. We’re selecting the best-use tooling available – for example MCP server tools, agent flows, APIs and plugins – to help Copilot observe, retrieve, reason, and execute using the tool we create. Today customers can build agents and add skills and tools that, in combination with our orchestration services, can also be used with Work IQ.
Going back to our complex content retrieval query, these are the tools that when combined, help us search, open and find content as described by the skill. This supports faster and more accurate responses and actions - while respecting governance policies, privacy and security. We expect to continuously experiment with and release new skills and tools for Work IQ.
Security, Privacy, and Compliance
We know how important it is that customers trust the AI services they use. Like the other components of Copilot, Work IQ is designed from the ground up to respect our customers’ already-existing user permissions, Security Group assignments, sensitivity labels, and Data Loss Prevention (DLP) policies. Similarly, we are committed to compliance with the legal and regulatory requirements of countries and regions around the world in which we operate, including the GDPR and EU Data Boundary.
Experiences and Extensibility
Work IQ is deeply integrated into Copilot. Licensed Microsoft 365 Copilot users will experience Work IQ in the responses and actions offered in Copilot Chat with the Work toggle activated and with Copilot in the M365 apps like Word, Excel, PowerPoint and Teams. Licensed Microsoft 365 users working with Copilot in Dynamics and Power Apps will find work IQ enriched with Dataverse data. In the coming months we will unify Work IQ experiences across all licensed Copilot surfaces.
Developers can also integrate Work IQ into their own apps and agents. The Work IQ API exposes Copilot intelligence through a standard RESTful interface, enabling developers to build agents grounded in the live work context accessible through Copilot Chat - while inheriting Microsoft’s enterprise‑grade identity, security, permissions, and regulatory compliance. The Work IQ API will be available in Public Preview later this month. Note that we support CLI today, and we’re working to offer MCP and A2A support in the coming months.
Below are sample prompts that showcase Work IQ in action. We’ve left some blanks so you can test them in your own environment:
- Copilot Chat: Over the past [timeframe] with I had a meeting with [person's first name], I asked about [specific topic or project]. Can you look for the information [person's first name] shared around this in that meeting?
- Copilot Chat: Identify tasks or action items assigned to me from my manager in this week's emails, Teams chats, and meeting notes, and compile them into a checklist.
- Copilot in Word: Draft an executive summary of [project] that highlights the purpose, progress, and implications of its impact on the business.
- Copilot in Outlook: Recommend how to resolve conflicts on my calendar for tomorrow.
- Copilot in Dynamics 365 Sales: Find the email with the latest price changes, analyze impact to my opportunities, and recommend next steps for me to communicate the changes.
- Researcher: What can you tell me about [Competitor name]? How do the offerings in [Competitor name] compare with [Company name] offerings? If you were in a compete situation against [Competitor name], what would you leverage as [Company name] response to our tooling and how a customer can accomplish the same thing? Pull in context from emails and Teams chats shared by the [Customer name] account team.
Conclusion
Work IQ combines powerful skills and tools with your work data and business context to help make Copilot more personalized, accurate, and trusted. We’re excited about Work IQ today and for the role it will play in helping our customers with their AI transformation. Whether you’re a Microsoft 365 Copilot user, an AI power user or a professional developer, we can’t wait to see what you do with Work IQ.
Microsoft