Publish Date: November 18, 2026
Executive Overview
As foundation models evolve on a biweekly cadence and become uniformly accessible across cloud providers, raw model intelligence has rapidly ceased to be a durable source of competitive advantage. Every enterprise can license the exact same frontier models; what differentiates market leaders from laggards is everything surrounding the model—specifically, institutional knowledge, operational workflows, collaborative history, governed business data, and domain-specific decision rules. Yet in most organizations today, this vital enterprise context remains trapped in disconnected silos across productivity suites, ERP platforms, analytical warehouses, and document repositories, forcing engineering teams to rebuild fragile grounding pipelines from scratch for every new AI agent.
At Microsoft Ignite 2026, Yitzhak Kesselman, Corporate Vice President of Microsoft IQ, formally introduced Microsoft IQ—a unified enterprise context management platform engineered to capture, govern, and activate an organization’s unique intelligence across every agent and application. By converging four specialized intelligence layers—Work IQ, Fabric IQ, Foundry IQ, and Web IQ—alongside more than 350 third-party connectors and reusable business plugins, Microsoft IQ establishes a centralized, permission-aware context fabric. Crucially, the platform is architected to be open and model-diverse, exposing governed enterprise context not only to Microsoft Copilot, GitHub Copilot, and Microsoft Foundry, but also to third-party ecosystems including ChatGPT and Claude via the Model Context Protocol (MCP). For Chief Data Officers, Chief Information Officers, and AI strategy leaders, Microsoft IQ represents the institutionalization of context engineering, transforming fragmented corporate knowledge into a compounding, governed enterprise asset.
Features
Microsoft IQ consolidates enterprise grounding, semantic modeling, knowledge graphs, live web retrieval, and continuous agent optimization into a single modular architecture. The core technical features, extracted strictly from the official announcement, span five primary capability areas.
- Unified Four-Layer Intelligence Architecture and Business Plugins: Microsoft IQ integrates four complementary context layers alongside over 350 third-party connectors to build a comprehensive organizational understanding. At the center of reusability are new business plugins, which allow teams to package governed context, domain skills, and authoritative data sources once and distribute them across agents, teams, and platforms. Business plugins can be built and consumed across Microsoft Copilot, GitHub Copilot, Microsoft Foundry via Model Context Protocol (MCP), and third-party assistant marketplaces including ChatGPT and Claude.
- Work IQ Operational Grounding and Governance: Work IQ supplies context on how people collaborate, encompassing documents, conversations, meetings, and past decisions, while transitioning from passive understanding to active execution. Newly generally available capabilities include Business Applications in Work IQ (adding CRM and ERP records to the grounding set), Retrieve (returning grounded, permission-trimmed Microsoft 365 content and federated connector data in a single call), real-time meeting context (allowing agents to act while meetings are in progress), long-running tasks and agent-to-agent support, granular MCP policies (scoping access by path, audience, and trusted application rather than a tenant-wide switch), and OpenAI and Anthropic plugins extending Work IQ into ChatGPT and Claude. In public preview, Work IQ adds Memory, native Word, Excel, and PowerPoint creation and editing, Listen event subscriptions, and agent policy governance with human approval gates.
- Fabric IQ Semantic Business Context and Ontologies: Fabric IQ delivers the governed state of the business through structured data, business entities, relationships, and rules. Power BI semantic models in Microsoft Copilot are already generally available, with Fabric data agents reaching general availability soon. New preview capabilities include Fabric IQ in Edit with Copilot within Excel (enabling natural-language queries against governed Power BI semantic models to generate live, Fabric-connected PivotTables), Fabric IQ connectors for ChatGPT and Claude, Fabric IQ ontologies inside GitHub Copilot for developers, embedded conversational experiences via Fabric apps, and ontology support in IQ Sharing to securely share business context with external customers and partners.
- Foundry IQ Knowledge Graphs and Serverless Retrieval: Foundry IQ manages curated institutional knowledge such as corporate policies and authoritative documents, engineered to deliver higher accuracy with fewer tokens. Now in preview, Knowledge graphs in Foundry IQ allow agents to reason across complex entity relationships spanning unstructured data and enterprise applications, paired with Microsoft Retriever 1, a new retrieval model enhancing agentic grounding. The indexed SharePoint knowledge source with full permissions support is now generally available. Additionally, Foundry IQ Serverless will reach general availability in December 2026, introducing usage-based pricing with free monthly allowances so teams pay strictly for knowledge consumed rather than reserved capacity.
- Web IQ Broad Availability and Enterprise Controls: Web IQ is now broadly available to all Azure customers without a waitlist, starting at $1 per 1,000 calls. It introduces Auto and Auto Advanced modes that unify Web, Images, Videos, News, Places, and Weather grounding while optimizing retrieval for complex queries. Web IQ spans 207.5 million real-world entities and weather data, and adds Commerce in private preview covering structured product data across more than 40 billion listings and offers. Enterprise controls include Scopes (to explicitly include or exclude specific websites and domains) and Bring Your Own License (to blend licensed publisher content directly with live web results).
- Continuous Improvement and Post-Training Loop: Microsoft IQ integrates directly with Microsoft Foundry’s continuous improvement engine, carrying proven context, skills, and evaluations across workflows. Agent Optimizer evaluates agents against quality, latency, and cost per successful outcome to rank improved configurations for human approval. When configuration tuning plateaus, teams can apply the exact same production signals to model weights via Managed Training (automated customization) or Programmed Training (now in public preview), which enables AI engineers to write custom supervised fine-tuning and reinforcement learning loops for supported open-weight models on Foundry-managed compute.
Benefits
Implementing Microsoft IQ as a centralized enterprise context management layer produces measurable strategic, operational, and economic advantages over ad-hoc grounding architectures.
- Elimination of Redundant Context Engineering: In typical enterprise AI initiatives, dozens of product teams independently build overlapping RAG pipelines, duplicate SharePoint connectors, and hardcode inconsistent business definitions into system prompts. Microsoft IQ eliminates this massive duplication of effort by establishing a build-once, reuse-everywhere context fabric. When a domain expert defines a business plugin or ontology in Fabric IQ or Foundry IQ, every downstream agent inherits that exact, authoritative logic—dramatically compressing time-to-value for new AI initiatives.
- Reduced Token Overhead and Predictable Grounding Economics: Grounding autonomous agents against sprawling unstructured repositories frequently bloats context windows and drives up inference bills. Foundry IQ’s new knowledge graphs and Microsoft Retriever 1 model specifically target higher retrieval precision using fewer tokens. Paired with Foundry IQ Serverless usage-based pricing (with free monthly allowances) and Web IQ’s transparent $1 per 1,000 calls entry point, FinOps teams gain predictable, consumption-aligned economics that scale smoothly from early prototypes to enterprise production.
- Cross-Ecosystem Governance Without Assistant Lock-In: Modern workforces rarely rely on a single AI interface; developers work in GitHub Copilot, knowledge workers use Microsoft Copilot, and specialized teams frequently adopt ChatGPT or Claude. Historically, supporting third-party assistants meant sacrificing enterprise data governance or manually copying sensitive data outside the corporate perimeter. Through granular MCP policies, permission-trimmed Retrieve calls, and native Work IQ and Fabric IQ connectors for OpenAI and Anthropic ecosystems, organizations can meet users in their preferred tools while ensuring corporate permissions, human approval gates, and intellectual property protections travel seamlessly with the data.
- Compounding Accuracy Across the Agent Lifecycle: By coupling enterprise context directly with Foundry’s Agent Optimizer, Managed Training, and Programmed Training, operational improvements no longer vanish inside isolated chat sessions. Production signals, evaluation criteria, and validated skills compound across the entire agent fleet, steadily improving decision quality while keeping proprietary intellectual property firmly under organizational ownership.
Use Cases
Microsoft IQ’s four-pillar architecture and reusable business plugins directly address high-stakes, context-intensive operational challenges across energy, manufacturing, supply chain, and corporate finance.
- Real-Time Critical Infrastructure Grid Operations: As spotlighted in the announcement, the Australian Energy Market Operator (AEMO) co-engineered NEM-GEM (National Electricity Market Grid and Energy Management) with Microsoft in less than 12 weeks. Powered by Microsoft IQ, NEM-GEM synthesizes thousands of live operational signals alongside institutional grid management procedures, enabling control-room operators to cut through signal noise and make faster, high-confidence decisions to keep Australia’s power grid secure—while keeping human experts firmly in control.
- Predictive Maintenance and Root-Cause Quality Investigation: An industrial manufacturer deploys a governed Microsoft IQ business plugin combining equipment telemetry ontologies in Fabric IQ, engineering maintenance manuals and safety policies in Foundry IQ, and past shift-handover discussions in Work IQ. Maintenance engineers and autonomous agents reuse this single plugin across factory sites to accelerate predictive maintenance triage and pinpoint root causes of recurring production quality defects.
- Accelerated Supplier Risk and Disruption Response: A global supply chain organization builds a cross-functional disruption response plugin uniting ERP order and inventory records (via Business Applications in Work IQ), governed Power BI supply chain semantic models (Fabric IQ), authoritative supplier contracts (Foundry IQ), and live geopolitical, weather, and news alerts (Web IQ). When a regional weather event or port strike occurs, agents instantly assess affected shipments, surface contract clauses, and trigger human-approved mitigation workflows—whether accessed from Microsoft Teams, Excel, or an external assistant via MCP.
- Live Financial Analysis and Cross-Organization Context Sharing: Corporate finance teams utilize Fabric IQ in Edit with Copilot directly inside Excel to query governed Power BI semantic models in natural language and generate live, Fabric-connected PivotTables without waiting on BI engineering queues. Simultaneously, using IQ Sharing for Fabric IQ ontologies, the enterprise securely shares standardized product or supply chain business definitions with key external partners and customers without duplicating underlying data lakes.
Alternatives
Enterprise data and AI architects evaluating Microsoft IQ must weigh its integrated four-pillar context platform against alternative industry approaches to enterprise knowledge grounding.
- Custom In-House RAG Pipelines and Standalone Vector/Graph Databases: Engineering-heavy organizations frequently build proprietary context layers combining open-source ingestion frameworks, standalone vector databases (such as Pinecone, Qdrant, or Weaviate), independent graph databases (such as Neo4j), and custom permission-filtering middleware. While this approach offers maximum architectural control and zero vendor lock-in, maintaining real-time document ACL synchronization across Microsoft 365, SharePoint, CRM, and ERP systems—while also building web grounding and MCP governance—imposes a massive, ongoing engineering tax that few enterprises can sustain cleanly.
- Competing Hyperscaler Knowledge and Enterprise Search Platforms: Rival cloud ecosystems provide integrated grounding alternatives, such as Amazon Q Business paired with Amazon Bedrock Knowledge Bases, or Google Cloud’s Vertex AI Search and Gemini Enterprise connectors. These platforms offer strong retrieval accuracy and enterprise connectors within their respective cloud estates, but they lack native, permission-trimmed real-time access to Microsoft 365 meetings, Office document editing, Power BI semantic models, and Excel Copilot workflows that anchor daily operations in Microsoft-centric enterprises.
- Independent Enterprise AI Search and Context SaaS Vendors: Specialized third-party enterprise search platforms (such as Glean or Coveo) offer turnkey, permission-aware indexing across heterogeneous SaaS applications including Slack, Jira, Confluence, Salesforce, and Microsoft 365. These vendors excel in highly diverse, non-Microsoft productivity environments and deliver rapid time-to-deployment, though they typically operate as standalone search and assistant layers rather than integrating deeply into custom model post-training (Programmed Training), Fabric semantic ontologies, and Azure developer workflows.
An Alternative Perspective
While Microsoft IQ presents a remarkably cohesive vision for solving enterprise AI’s context fragmentation problem, rigorous architectural scrutiny exposes several structural challenges regarding data readiness, governance complexity, and platform maturity.
First, Microsoft IQ cannot magically manufacture institutional clarity where underlying enterprise data hygiene is broken. Business plugins, Foundry IQ knowledge graphs, and Fabric IQ ontologies depend heavily on accurate Power BI semantic models, cleanly maintained SharePoint permissions, and well-structured CRM/ERP records. In reality, most large enterprises suffer from decades of overshared SharePoint sites, conflicting departmental KPI definitions, and stale policy documents. Activating Work IQ Retrieve and Foundry IQ knowledge graphs across poorly governed repositories risks amplifying data leakage—surfacing technically permitted yet sensitive legacy files—or encoding contradictory business rules into widely reused plugins. Organizations must invest heavily in foundational data stewardship and Microsoft Purview permission hygiene before scaling Microsoft IQ enterprise-wide.
Second, the staggered release maturity across the four IQ pillars creates real architectural complexity for production deployment planning. While core Work IQ grounding, indexed SharePoint sources, and Web IQ are generally available, many of the platform’s most transformative differentiators—including Foundry IQ knowledge graphs, Microsoft Retriever 1, Work IQ Memory and Office editing, Fabric IQ ontologies in GitHub Copilot, IQ Sharing for ontologies, ChatGPT/Claude Fabric connectors, Web IQ Commerce, and Programmed Training—remain in public or private preview, with Foundry IQ Serverless scheduled for December 2026. Enterprise architects must carefully segregate generally available production foundations from preview features to avoid building mission-critical workflows on shifting preview interfaces.
Third, extending enterprise context into third-party assistant ecosystems like ChatGPT and Claude via MCP and Work IQ/Fabric IQ plugins introduces a nuanced governance boundary. Even though granular MCP policies scope access by path, audience, and trusted app, once grounded enterprise context is transmitted into an external assistant’s reasoning loop, Chief Information Security Officers must verify data retention agreements, zero-data-retention guarantees, and cross-tenant audit logging across every participating third-party vendor to prevent silent shadow-AI data exfiltration.
Final Thoughts
The launch of Microsoft IQ at Microsoft Ignite 2026 addresses the single most consequential bottleneck in enterprise artificial intelligence: turning scattered institutional knowledge into a governed, reusable, and continuously improving operational foundation. By unifying Work IQ, Fabric IQ, Foundry IQ, and Web IQ behind standardized business plugins and open MCP interoperability, Microsoft enables organizations to capture their domain expertise once and compound its value across every internal and external agent.
For technology and data leaders, the path forward should follow the pragmatic discipline outlined in the announcement itself: start with a single high-value business process where superior context directly drives measurable operational outcomes, clean and govern the underlying semantic models and document permissions, package that verified know-how into a reusable business plugin, and connect it to Foundry’s continuous improvement loop. Executed with strong data governance, Microsoft IQ transforms proprietary organizational context from a static archive into an enduring competitive moat.