FACILITATOR GUIDE

Demo Scripts

Step-by-Step Live Demo Walkthroughs

Demo Scripts

Step-by-step scripts for every live demo in the session. Keep this on your second screen.

Pre-Demo Setup

Before the session starts:

  1. Open VS Code with a clean demo workspace (not the Master Alex workspace!)
  2. Initialize Alex: Ctrl+Shift+P → “Alex: Initialize Architecture”
  3. Verify the Welcome View appears in the sidebar
  4. Open Copilot Chat panel (Ctrl+Shift+I or click the Copilot icon)
  5. Test one quick message: /status → confirm you get a response
  6. Set Gamma API key if demoing presentations: $env:GAMMA_API_KEY = "sk-gamma-xxx"
  7. Set Replicate API key if demoing images: Use Alex Welcome View → “API Keys & Secrets”
  8. Test audio output if demoing TTS

Demo #1 — Say Hello (Module 1, ~0:12)

Purpose: Show Alex’s personality and personalized greeting.

Steps

  1. In Copilot Chat, type:

    Hello! I'm presenting you to a room full of knowledge workers today.

  2. Point out to the audience:

    • Alex responds with personality (not a generic “How can I help?”)
    • Alex may run a self-actualization check (health, version, synapses)
    • The sidebar avatar may change to reflect the cognitive state
    • Alex uses the name from the user profile

  3. Follow up with:

    What do you know about me?

  4. Point out: Alex recalls the profile — name, role, preferences, technologies. This is persistent memory.

If it fails

Say: “Alex greets you by name, checks its own health, and shows you its cognitive state. It’s like starting a workday with a colleague who remembers what you worked on yesterday.”

Demo #2 — Dialog Engineering (Module 2, ~0:33)

Purpose: Show a multi-turn conversation using the 5 prompting patterns.

Steps

Turn 1 — CONTEXT-GOAL-CONSTRAINTS:

I'm a marketing director preparing for our annual strategic planning session.
I need a framework for evaluating which of our 12 product lines to invest in, maintain,
or sunset. The audience is C-suite executives who want data-driven recommendations,
not opinions. Keep it to a one-page summary format.

Wait for response. Point out: “See how the context, goal, and constraints shaped the response? No ambiguity.”

Turn 2 — ITERATE:

That's a strong framework. Two adjustments:
1. Add a "customer retention risk" dimension — some products are low-revenue but high-retention
2. Make the scoring criteria more explicit — I want executives to see the math
Keep the one-page format.

Point out: “I didn’t start over. I built on what Alex gave me. Each turn adds context.”

Turn 3 — CHALLENGE-ME:

Now challenge this framework. What are the blind spots? What would a skeptical CFO
poke holes in? Give me the three strongest objections and how to address them.

Point out: “THIS is the power move. Most people never ask the AI to push back. Alex is now stress-testing your thinking.”

Turn 4 — CHECKPOINT:

Summarize our complete framework — the original, my additions, and the objections
with responses — in a clean one-page format.

Point out: “Four turns. No restarts. The final output is better than anything the first turn could have produced.”

Alternative prompts (for academic audience)

Replace “marketing director” with “doctoral researcher preparing a dissertation defense” and “product lines” with “research methodologies” for an academic demo.

Demo #3 — Gamma Presentations (Module 3, ~0:42)

Purpose: Show AI-generated slide decks from a conversational prompt.

Steps

  1. In Copilot Chat, type:

    Create a 10-slide presentation about the future of AI in higher education.
Audience: university deans and provosts. Tone: visionary but grounded in evidence.
Export as PowerPoint.

  2. While it generates, explain to the audience:

    • Alex calls Gamma AI to design professional slides
    • AI selects appropriate images, layouts, and color schemes
    • You can specify tone, audience, slide count, and image style
    • The result is a fully designed deck, not bullet-point outlines

  3. When the result appears:

    • Show the Gamma link (opens in browser)
    • Show the PowerPoint export option
    • Point out the visual quality vs. “I typed this in PowerPoint”

  4. Follow up:

    That's great. Can you also create a version as a document format
for those who prefer to read instead of watch slides?

If Gamma is not configured

Say: “Behind the scenes, Alex calls Gamma AI to design full slide decks with professional layouts, images, and transitions. You describe what you want in conversation, and you get a finished deck you can export to PowerPoint. No design skills required.”

Show a pre-generated example if available.

Demo #4 — Word Documents (Module 3, ~0:48)

Purpose: Show markdown-to-Word conversion with professional formatting.

Steps

  1. First, create some content. In Copilot Chat:

    Write a 2-page executive summary about implementing AI governance
in a mid-size organization. Include a process flowchart using Mermaid syntax.
Structure it with: Background, Key Principles, Implementation Roadmap,
and Governance Framework sections.

  2. When Alex generates the markdown with a Mermaid diagram:

    • Save the output to a .md file in the workspace

  3. Convert to Word:

    • Ctrl+Shift+P → “Alex: Convert Markdown to Word”
    • Or mention it in chat: Convert this to a Word document

  4. Open the generated .docx and point out:

    • Professional formatting (fonts, headings, spacing)
    • The Mermaid diagram rendered as an embedded image
    • Ready for review, printing, or submission

If conversion tool isn’t available

Generate the markdown, show the Mermaid diagram preview in VS Code, and explain: “Alex generates the content with embedded diagrams, then converts it to a professional Word document. Faculty, analysts, and consultants use this daily.”

Demo #5 — Mermaid Diagrams (Module 3, ~0:52)

Purpose: Show how Alex creates visual diagrams from natural language.

Steps

  1. In Copilot Chat:

    Create a flowchart showing the peer review process for an academic journal.
Start from manuscript submission through to publication or rejection.
Include revision loops.

  2. When Alex generates the Mermaid code:

    • Copy it into a .md file
    • Show the Mermaid preview (VS Code has built-in Mermaid rendering)
    • Or use the Mermaid plugin for a live preview

  3. Follow up with a different diagram type:

    Now create a Gantt chart for a 6-month research project with these phases:
Literature Review, Data Collection, Analysis, Writing, Review & Revision.

  4. Point out:

    • Natural language → professional diagram
    • Multiple diagram types available
    • Can be embedded in documents, presentations, reports

Demo #6 — AI Images (Module 3, ~0:55)

Purpose: Show right-click image generation.

Steps

  1. Open any .md file in the workspace
  2. Right-click → Generate AI Image from File
  3. Alex reads the file content and generates a relevant image

Alternative (from prompt):

Generate a professional banner image for a conference presentation
on "The Future of Knowledge Work." Style: modern, clean, abstract.
  1. Point out:
    • Multiple AI models available (Flux for speed, Ideogram for text in images)
    • Costs range from $0.003 to $0.08 per image
    • Results appear in the workspace as PNG files

If Replicate is not configured

Show a pre-generated image and explain the workflow. Keep this demo brief — 2 minutes maximum.

Demo #7 — Save an Insight (Module 4, ~1:03)

Purpose: Show how Alex captures and stores knowledge.

Steps

  1. In Copilot Chat:

    /saveinsight title="Five patterns for effective AI dialog" insight="The five core patterns for productive AI conversations are: 1) Context-Goal-Constraints to set the scene, 2) Explain-Like to control depth, 3) Show-Don't-Tell for concrete examples, 4) Iterate to build incrementally, 5) Challenge-Me for critical thinking. These patterns work regardless of which AI model you're using." tags="dialog-engineering,ai-patterns,communication,knowledge-work"

  2. Point out:

    • The insight gets a unique ID, timestamp, and category
    • Tags make it searchable later
    • It’s stored in the Global Knowledge base, not just this project

  3. Emphasize: “This insight is now available in every workspace where Alex is initialized. You saved it here, but you can find it anywhere.”

Demo #8 — Search Knowledge (Module 4, ~1:05)

Purpose: Show cross-project knowledge retrieval.

Steps

  1. In Copilot Chat:

    /knowledge dialog patterns communication

  2. Point out:

    • Alex searches across all saved insights and patterns
    • Returns results with context, tags, and source project
    • Works even if you saved the insight months ago in a different project

  3. Follow up (if there are other insights in the GK base):

    /knowledgestatus

  4. Point out: “This shows your entire knowledge library — how many patterns, how many insights, what categories, which projects contributed. This is your compound interest.”

Demo Tips

General

Recovery

Timing