The context problem Copilot couldn’t solve alone
Enterprises that deployed Microsoft 365 Copilot encountered a consistent limitation: the assistant was fluent and fast, but it lacked organisational context. It surfaced documents without knowing which ones actually drove decisions. It summarised meetings without understanding which outcomes mattered. It had no model of who the real decision-makers were, how information moved across teams, or where approvals typically stalled. The problem was not the underlying model. It was the absence of a persistent, structured understanding of how work actually flows.
Microsoft described this as the “context gap” at Ignite 2025: the inability of AI tools to maintain a persistent understanding of an organisation’s real workflows — not its org chart or SharePoint taxonomy, but the actual patterns of collaboration, communication and decision-making. Work IQ is Microsoft’s direct response. It entered preview on 30 September 2026, with rollout continuing through October 2026, according to the Microsoft Dynamics 365 Blog.
What Work IQ actually is
Work IQ is not a new application. It is an intelligence layer that sits beneath Copilot and custom agents, grounding their reasoning in approved business data from Microsoft Dynamics 365 and Microsoft Power Platform. On the Microsoft 365 side, it semantically indexes SharePoint documents, Outlook emails, Teams meetings and chats — not as isolated files, but as a connected web of knowledge that captures relationships, timelines and collaboration patterns. On the business application side, it reasons across Dynamics 365 CRM (Sales and Service), Dynamics 365 ERP (Finance and Operations), Power Apps, and third-party line-of-business applications, as the Dynamics 365 Blog confirms.
A useful architectural distinction separates Work IQ from Microsoft Graph. Graph provides raw data access — mail, files, Teams, calendar, users. You query it and build the logic yourself. Work IQ codifies how work gets done from the daily flow of information, then makes that understanding available to Copilot and agents without requiring every developer to reconstruct it independently. For organisations building multiple agent experiences, that distinction has direct implications for build cost and consistency.
Work IQ sits within a broader architecture. Microsoft positions it alongside Fabric IQ (business intelligence and data analysis) and Foundry IQ (agent reasoning capabilities) as three components of a unified Microsoft IQ offering. Work IQ is specifically the execution layer — the component that brings AI-driven decisions into the actual flow of work inside Microsoft 365, according to the Microsoft TechLab repository.
The Work Chart and persistent memory
Two capabilities warrant close attention from operators evaluating the platform.
The Work Chart is Work IQ’s model of actual collaboration patterns — who works with whom regardless of reporting lines, which relationships drive decisions, and how information moves across teams. It deliberately distinguishes formal hierarchy from real workflow. For Copilot, this means recommendations grounded in how an organisation actually operates rather than how it is theoretically structured, as Devoteam’s analysis describes.
Persistent memory means Work IQ retains context across sessions: user preferences for communication style and document formatting, repetitive task sequences, and previous project discussions. Copilot can recall prior context without users re-establishing it each time. This addresses one of the most consistent complaints about enterprise AI assistants — the requirement to re-establish context at the start of every interaction.
For developers and coding agents, Work IQ introduces a dedicated MCP Server, a command-line interface (CLI), and a Work IQ plugin. These bring approved business context directly into the coding environment. The developer begins with the intended business outcome; the coding agent determines which skills, tools, and sequence to use. The developer remains responsible for architecture, validation, testing, and user experience. Connected experiences reuse the same approved business context and process definitions, which the Dynamics 365 Blog describes as enabling faster builds of governed agent experiences.
Governance is the operative question
Most coverage of Work IQ focuses on what AI can now do. The more consequential question for enterprise operators is who controls what AI is allowed to know.
Work IQ answers that through Microsoft Agent 365. IT administrators choose which business context agents can access and control which users, agents, and applications can reach business data. The Dynamics 365 Blog is specific about the terms: observable means understanding how AI uses data and evaluating the value it creates; governable means the IT admin decides what business data becomes available to AI experiences. This is not a single new portal — it is connected decisions across familiar administrative experiences.
The strongest counterpoint to the Work IQ proposition is legitimate: centralising intelligence also centralises risk. If Work IQ’s model of your organisation is wrong — if it misidentifies real decision-makers, or surfaces sensitive business context to the wrong agent — the error propagates across every experience drawing from that shared layer. The governance controls in Microsoft Agent 365 exist because Microsoft acknowledges this failure mode. But controls are only as effective as the policies an organisation actually configures. Deploying Work IQ without a defined data access policy does not reduce risk; it concentrates it.
A related concern worth addressing explicitly: the Work Chart and persistent memory capabilities raise legitimate questions about employee monitoring. Understanding who collaborates with whom, and retaining that context persistently, is valuable for AI grounding — and it is also the kind of data that requires clear policy boundaries and transparent communication with employees before deployment begins.
What operators should do during preview
Work IQ is in preview, not general availability. That status matters: the feature set, performance characteristics, and governance controls remain subject to change before full release. This is the right moment to evaluate and design, not to deploy at scale.
Three things need to happen in parallel. First, audit the business data you would connect — Dynamics 365 CRM, ERP, Power Apps, and any third-party line-of-business applications. Identify which data should never reach an agent and define that boundary explicitly before preview access begins. Second, map your actual collaboration patterns independently. Building your own baseline gives you something to validate Work IQ’s model against, rather than accepting its construction of your organisation uncritically. Third, assign named ownership for the governance configuration inside Microsoft Agent 365. This is an architectural decision with security implications — not a setting individual teams should configure in isolation.
Work IQ represents a meaningful shift in how enterprise AI accesses organisational context. The value it creates depends on the quality of the policies built around it. The intelligence layer functions only when the governance layer is designed first.
One action: Before enabling Work IQ in preview, convene a single working session with IT, security, and at least one business application owner to define the data access boundary and agree on employee communication. That conversation — not the feature activation — is where the real work begins.
— Eagentix
Eagentix helps growth-focused enterprises redesign and automate manual business processes. We combine executive strategy, implementation support, and managed services to build dependable operations across Southeast Asia.
Sources
- Source: Work IQ: Business and workplace intelligence in the flow of work – Microsoft Dynamics 365 Blog (
- Source: Introducing Microsoft Work IQ: The Intelligence Layer for Agents / Blogs / Perficient (
- Source: WORK IQ. | Josh Cook (
- Source: What is Microsoft Work IQ? – Infinity Group (UK) (
- Source: GitHub – microsoft/TechLab-Building-Intelligent-Solutions-with-Microsoft-Work-IQ · GitHub (
- Source: Microsoft Work IQ: Building The Intelligence Layer for Enterprise AI | Devoteam (