4.5.1. Microsoft 365 Agents for Business Scenarios
💡 First Principle: Microsoft 365 agents live where users already work — Teams, SharePoint, Outlook. Their power comes from proximity to the user's daily workflow, not from architectural complexity. An agent that surfaces the right information inside a Teams chat at the right moment beats a sophisticated agent that requires users to switch to a different application.
Agent Deployment Surfaces in M365:
| Surface | Agent Capability | Design Use Case |
|---|---|---|
| Teams | Conversational agents, meeting agents, channel bots | IT help desk agent, HR policy Q&A, meeting summarizer |
| SharePoint | Knowledge agents grounded on site content | Department-specific knowledge assistant, policy lookup |
| Outlook | Email-integrated agents, scheduling assistants | Email triage, meeting preparation, follow-up drafting |
| Microsoft 365 Copilot chat | General-purpose Copilot with plugin extensibility | Cross-application queries, data synthesis across M365 |
Orchestrating Multiple M365 Agents:
When multiple agents are deployed across M365 surfaces, the architect must design for discoverability (how do users find the right agent?), consistency (do agents provide coherent answers across surfaces?), and governance (who manages which agents, and what data can they access?).
Reflection Question: A company wants AI agents in Teams for IT support, in SharePoint for HR policy questions, and in Outlook for sales follow-up drafting. Each agent accesses different data sources. Design the governance model: who owns each agent, what data boundaries exist, and how do you prevent data leakage across the three agents?
