Microsoft IQ Overview (Work IQ, Foundry IQ and Fabric IQ)

Microsoft IQ Overview

Key Concepts:

• Microsoft IQ: A unified system designed to provide AI agents with comprehensive enterprise knowledge and context, encompassing work history, curated institutional knowledge, and real-time business state.
• Work IQ: Focuses on personalization and memory, learning from user interactions (IMs, emails, meetings, documents) to understand individual work patterns and objectives.
• Foundry IQ: Enables curated AI-powered search across specific knowledge sources (documents, SharePoint, web sources, OneLake data) with granular control and AI-driven relevance.
• Fabric IQ: Defines entities, relationships, rules, and constraints within the OneLake data environment, providing operational intelligence and a semantic layer for analytics and operations.
• Semantic Index: A method of indexing data based on meaning and relationships, allowing for natural language searches beyond keyword matching.
• One Lake: Microsoft Fabric’s unified data platform for analytical and operational data, including streaming data, SQL databases, and Cosmos DB.
• Agent 365: A framework for observing, securing, managing, and governing AI agents within an enterprise environment.

Understanding the Need for Microsoft IQ

The speaker begins by highlighting the widespread recognition of AI’s business value – boosting employee productivity, optimizing operations, enhancing customer experiences, and accelerating innovation. However, current generative AI models, while powerful, lack inherent knowledge of specific enterprises. These models are trained on public data and require supplemental information about an organization’s data, tools, and processes to be truly effective. The quality and relevance of this supplemental data directly impact the AI’s performance and the quality of its outputs.

Currently, enterprise information is fragmented across various platforms: M365 (messages, documents, chats, Dynamics 365 customer signals), institutional knowledge stores (policies, contracts, knowledge bases), and data lakes like Microsoft Fabric (analytical, operational, streaming data, SQL databases, Cosmos DB). While tools like the Graph API provide a semantic index for natural language searches, they lack a deep understanding of how the organization works and lack consistency across experiences. This fragmented landscape limits the ability of AI agents to leverage the full potential of enterprise data.

The core challenge is transforming this wealth of data into usable “enterprise knowledge” grounded in entities, relationships, and constraints – mirroring the understanding of a highly effective employee. Microsoft IQ aims to bridge this gap, providing AI agents with the same level of contextual awareness. As the speaker states, “The goal when I think about the Microsoft IQ, I need to get AI agents that same level of knowledge.”

The Three Pillars of Microsoft IQ

Microsoft IQ is comprised of three interconnected components, or “IQ’s”: Work IQ, Foundry IQ, and Fabric IQ.

1. Work IQ: Adding Memory and Personalization

Work IQ focuses on providing AI agents with a personalized understanding of individual users and the organization’s culture. It achieves this by:

• Learning from Interactions: Analyzing IMs, emails, meetings, and documents to understand user communication styles, work patterns, and collaborative relationships.
• Tracking Objectives: Monitoring user goals and projects to proactively offer assistance.
• Understanding Company Rhythm: Recognizing planning cycles and workflows to anticipate needs.
• Utilizing Explicit and Implicit Memory: Remembering specific information provided by the user (explicit) and learning from ongoing interactions (implicit).
• Tuned Inferencing: Optimizing interactions across data, apps, and workflows based on learned patterns.

The result is a “super rich context” about an employee’s work and collaboration, enabling more relevant and effective AI assistance.

2. Foundry IQ: Curated Knowledge for AI Agents

Foundry IQ addresses the need for controlled and reliable knowledge sources for AI agents. Key features include:

• Curated Information Sources: Allowing administrators to specify trusted sources like documents in Blob storage, SharePoint sites, unstructured data from OneLake, and specific web sources.
• Granular Control: Providing the ability to “put blinders on” the AI, restricting its knowledge base to designated sources.
• AI-Powered Relevance: Utilizing AI to identify the most relevant information within the curated sources, tailoring interactions and queries for optimal results.
• Knowledge Base Management: Facilitating the creation and maintenance of institutional knowledge bases.

Foundry IQ complements Work IQ by adding a layer of curated institutional knowledge, enhancing the overall context available to AI agents.

3. Fabric IQ: Operational Intelligence from Data

Fabric IQ focuses on grounding AI agents in the real-time state of the business. It achieves this by:

• Defining Entities and Relationships: Establishing a semantic layer on top of the OneLake data environment, defining key business entities and their relationships.
• Incorporating Business Rules and Constraints: Adding operational context, policies, and constraints to the data model.
• Enabling Operational Intelligence: Providing a foundation for both analytical and operational applications.
• Abstracting Data Complexity: Allowing users to interact with entities without needing to understand the underlying data infrastructure.

The speaker illustrates this with an example of defining “robots” and “missions” within an ontology, linking them to relevant data across OneLake. This allows users to act on entities directly, without needing to navigate the complexities of the underlying data.

Integrating Microsoft IQ with AI Agents & Governance (Agent 365)

The speaker emphasizes that Microsoft IQ is designed to be accessible to AI agents built using various platforms, including Agent Builder (low-code), Copilot Studio (no-code/low-code), and Microsoft Foundry (pro-code). The goal is to provide a unified data and control plane, enabling agents to not only access information but also perform actions and operations.

The speaker highlights the importance of meeting users where they are, embedding agents into existing tools like Copilot to avoid disrupting workflows. Microsoft IQ provides a consistent source of “enterprise truth” for these agents.

Finally, the speaker introduces Agent 365, a framework for governing and securing AI agents. Agent 365 provides:

• Identity Management: Assigning identities and roles to agents.
• Auditing: Tracking agent interactions.
• Security: Protecting agents from threats.
• Discovery: Enabling agents to discover each other.

Conclusion

Microsoft IQ represents a significant step towards enabling AI agents to operate effectively within complex enterprise environments. By providing a unified system for delivering enterprise knowledge and context – encompassing personalized work history, curated institutional knowledge, and real-time business state – Microsoft IQ empowers AI agents to perform tasks with the same level of understanding and insight as a human employee. The integration of Work IQ, Foundry IQ, and Fabric IQ, coupled with the governance features of Agent 365, positions Microsoft IQ as a critical component of the future of AI-powered business operations. As the speaker concludes, “We’re going to provide a way for our agents to get that same level of knowledge, that same context, so they can perform all manner of enterprise capabilities.”