This page has two uses, and each topic below has a matching part for each:
- Instructors: use the Discussion points to lead a conversation with the room.
- Attendees: waiting on the agent, or would rather read than listen? Use the Reading instead. You don't need to read it all. Pick the one topic that interests you most, read its Start here link, and come back for more if you have time.
Everything linked here is public documentation.
Discussion points:
- Which of your customers could be good modernization candidates?
- Who has aging Java or .NET apps, framework upgrades, technical debt, or apps they're already discussing moving to Azure?
Reading:
- Start here: Modernization agent overview. It frames a customer conversation as Assess → Plan → Execute.
- Go deeper: GitHub Copilot modernization documentation. A good page to bookmark, with Java, .NET, C++, JavaScript/TypeScript, CLI, assessment, and migration guidance.
Discussion points:
- Which of these would your customers actually use?
- Where would each one fit in a customer's modernization journey?
Reading: pick the tool you're curious about.
- Custom skills: teach Copilot your organization's standards. Adding agent skills for GitHub Copilot CLI
- Agents window in VS Code: assign work to agents and review the results across projects. Use the Agents window
- Modernization CLI: run modernization from the terminal. Quickstart: install and use the modernization agent
- Modernization plugin: a multi-agent workflow for assessment, planning, and execution in Copilot CLI. GitHub Copilot modernization repository
Discussion points:
- How do you choose and switch models in GitHub Copilot?
- When would a different model give better results for a modernization task?
Reading:
- Start here: Changing the AI model for GitHub Copilot Chat. How to switch models in VS Code, and why they have different strengths.
- Go deeper: AI language models in VS Code. A simple rule of thumb: use a faster model for quick questions and small edits, and a stronger reasoning model for complex refactoring, architecture decisions, or multi-step work.
Discussion points:
- Model choice isn't only about quality, it's also about cost. How would you explain that to a customer?
- How does token usage drive cost?
Reading:
- Start here: Changing the AI model for GitHub Copilot Chat. Different models use AI credits at different rates.
- Go deeper: AI language models in VS Code. Turning up a reasoning model's effort produces more thinking tokens, which uses more credits.