Browse the guide
Guides
Presenter Tips
Narration tips, pacing guidance, and recovery strategies for demos.
General Principles
Narrate Before Running
Always explain what you’re about to do before executing a prompt:
- Intent — “I’m going to ask Copilot to…”
- AI Action — Run the prompt, let the audience watch
- Validation — “Notice how it picked up on…”
This structure helps the audience follow along even if the output is dense.
Keep Diffs Small and Visible
- Prefer one backend service (
order-serviceorpayment-service) for consistency - Avoid prompts that touch many files simultaneously
- Use VS Code’s diff view to highlight changes
Manage Expectations
- “AI output varies each run — the structure will be consistent, but details may differ.”
- “We always review AI-generated code before merging.”
Pacing
| Audience | Suggested Total Time | Focus Areas |
|---|---|---|
| Executives | 10 min | Plan mode, Code review, Security overview |
| Engineering leads | 20 min | Full end-to-end sequence |
| Developers | 30+ min | Full sequence with live coding follow-ups |
Timing Tips
- Allow 10–15 seconds for Copilot to generate responses
- Don’t rush to fill silence while Copilot is working — narrate what’s happening
- If a response is still generating, preview what to expect
Demo Environment Setup
Recovery Strategies
Copilot Gives a Broad or Off-Topic Response
Follow up with: “Make this minimal and repo-specific.”
This grounds Copilot back to the project context.
Response Takes Too Long
- Switch to a pre-prepared example while waiting
- Narrate: “While this generates, let me show you what the typical output looks like…”
Code Doesn’t Compile or Tests Fail
- Don’t panic — this is a teaching moment
- Narrate: “This is exactly why we review AI-generated code. Let’s see what needs adjusting.”
- Fix one issue live to show the iterative workflow
Agent Doesn’t Respond
- Check that the agent name is spelled correctly (e.g.,
@bdd-specialist) - Try refreshing the Copilot Chat panel
- Have a backup screenshot of typical output
Storytelling Arc
For maximum impact, structure the demo as a story:
- Setup — “We have a bicycle e-commerce platform with 6 microservices…”
- Challenge — “The PM wants a wishlist feature. Let’s see how Copilot helps across the entire SDLC.”
- Journey — Walk through Plan → Code → Review → Test → Deploy → Secure
- Resolution — “In 20 minutes, we went from idea to implementation plan, working code, tests, CI pipeline, and security validation — all with AI assistance.”
Common Questions & Answers
Q: Does Copilot replace developers?
“No — it’s a force multiplier. Developers still make design decisions, review output, and own quality. Copilot handles the repetitive parts so developers focus on what matters.”
Q: How does it know about our codebase?
“Copilot reads the repository context — file structure, imports, naming conventions, configuration files — and uses that to generate contextually relevant code.”
Q: Is the generated code secure?
“AI-generated code should go through the same review process as human-written code. That’s why we showed Copilot code review and CodeQL — they catch issues regardless of who wrote the code.”
Q: What about data privacy?
“Copilot processes code in context but doesn’t store or train on your private repository code. Check GitHub’s Copilot trust documentation for the latest details.”