>_ The Manifest

An agent can have excellent instructions and still miss what matters if it does not understand the work around the request. Work IQ gives agents permission-aware organizational context, so they can connect people, projects, conversations, meetings, and files instead of treating every prompt like an isolated question. Watch the session, then skim the recap below.

▶ Context Is Everything: Building Agents That Know Your Work Context from the Microsoft 365 Copilot Agents Playbook. Open on YouTube.

What you’ll see

Work IQ is the workplace intelligence layer behind Microsoft 365 Copilot. It builds a semantic understanding from signals across emails, chats, meetings, documents, people, and business systems, while preserving each user’s existing permissions. The result is not another pile of search results: it is agent-ready context assembled around the user’s intent.

  • Chat, context, tools, and workspaces as the four parts of the Work IQ platform
  • Grounded synthesis that answers a question and links back to the source material for verification
  • A2A, MCP, and REST as entry points for agent collaboration, tool use, and applications
  • Natural-language access from the command line, including GitHub Copilot CLI with the Work IQ plugin
  • Building and provisioning an agent with Work IQ Developer Tools instead of hand-authoring every step
GitHub Copilot

Find the most recent PowerPoint files that I modified.

I found the most recent PowerPoint files you modified and ranked them by last activity. Each result includes its location and a link back to the source so you can verify it.

The important distinction is that Work IQ is the intelligence layer your agent uses at runtime, while WIQD is the developer toolset used to build, validate, provision, and manage the agent. They complement each other, but they solve different problems.

⚠️ Warning

Work IQ Developer Tools are in preview. Use the current documentation and command help as the source of truth, expect breaking changes, and keep preview workflows off the critical path of production deployments.

The takeaway: grounding is not about giving an agent access to more documents. It is about giving the agent enough permission-aware context to understand which information matters, how it relates to the request, and where the user can verify the answer.

Resources

Have questions or want to share what you're building? Connect with me on LinkedIn or check out more on The Manifest.