The Copilot architecture and what runs where - SkilBrill

Microsoft 365 Copilot Training The Copilot architecture and what runs where

Microsoft 365 Copilot Training

0% Completed

M1 — How Copilot Works in Your Tenant

The Copilot architecture and what runs where

Grounding: how Microsoft Graph supplies the context behind an answer

Permissions mirroring: Copilot can only surface what the user can see

What Copilot does with prompts, telemetry and your tenant data

The difference between Copilot, Copilot agents and standalone chat

Where the model ends and your organisation's content begins

M2 — Daily Work: Writing That Lands

Prompt structure for drafting, rewriting and summarising

Word: drafting from a document, rewriting for an audience, tightening

Outlook: triage, drafting replies that sound like you, meeting prep

Looping a person in or out of a thread without losing the point

Using Copilot pages as a workspace rather than a one-shot answer

Turning a messy first draft into something worth sending

M3 — Analysis and Slides

Excel: asking questions of a table and checking the formula it wrote

Spotting the difference between an analysis and a confident guess

PowerPoint: generating a deck from a document, then fixing the story

Speaker notes and the slide that should have been a paragraph

Business Chat: pulling a cross-app picture for a status update

Verifying an output before it reaches a customer or a board

M4 — Meetings and Collaboration

Teams: recap, action items and who actually owns what

Recap against a meeting the user missed, and its limits

Whiteboard and Copilot for structuring a working session

SharePoint: finding and summarising content across sites

OneNote integration and capturing decisions where they live

Privacy: what attendees should know before the assistant joins

M5 — Administration and Tenant Setup

Licensing: what is required and what is optional

The Copilot admin page, settings and feature rollout

Preparing data: storage, sharing hygiene and indexing readiness

Network, latency and the readiness checks that prevent week-one complaints

Search and sensitivity labels and their effect on what gets grounded

Support model: who answers what after rollout

M6 — Governance, Security and Compliance

Microsoft Purview sensitivity labels and how they govern access

Data loss prevention applied to AI interactions

Audit logging: what was asked, by whom and what was returned

Retention, eDiscovery and AI-generated content in a record

Oversharing: the inherited permission problem that surfaces in Copilot

Writing a tenant AI usage policy people can follow

M7 — Adoption and Change

Why adoption fails in week two and what predicts success

Role-based training: different prompts for different jobs

Champions, office hours and a feedback loop that surfaces real blockers

Measuring adoption beyond licences assigned

Handling scepticism without overselling the tool

Publishing an internal prompt library that stays maintained

M8 — Capstone

Run a readiness assessment on a real tenant configuration

Draft the licensing and staging plan for a phased rollout

Apply Purview controls to the content that must not be surfaced

Build role-based prompt training for three different job families

Define adoption and time-saved metrics with a baseline to beat

Deliver the rollout plan, policy document and support playbook

Announcements Course Info