Before you start: VS Code open · GitHub Copilot active · Alex extension installed · any workspace folder open
1
Meet Alex
Initialize
- Open VS Code
- Press Ctrl+Shift+P and type Alex: Initialize Architecture → press Enter
- Wait for the confirmation notification (~10 seconds)
- Check the sidebar — you should see Alex's Welcome View
Open Alex in Agent Mode
- Open the Copilot Chat panel (click the Copilot icon in the sidebar, or press Ctrl+Shift+I)
- Click the agent selector at the top of the chat panel (it may say "Ask Copilot")
- Select Alex from the list
Your first message
Type this — filling in your real information:
Hello! My name is [your name]. I'm a [your role] working in [your field].
I prefer [brief/detailed] explanations and a [formal/casual] tone. Example
Hello! My name is Maria. I'm a research analyst working in healthcare policy.
I prefer detailed explanations and a formal tone. Read Alex's response — it will greet you by name and save your preferences.
Verify the memory works
What do you know about me? Alex should recall your name, role, and preferences. This persists across sessions.
2
Dialog Engineering
Structure every substantive request with this pattern:
I'm a [role], working on [real task or project].
I need [specific deliverable].
Keep it [constraint: length, format, audience, tone]. Marketing
I'm a marketing professional working on [campaign, launch, or content initiative].
I need [a campaign brief / content strategy / messaging framework] for [audience/product].
The goal is [awareness/conversion/retention]. Tone: [brand voice descriptor]. Then iterate:
Good. Now write the hook for the first social post — 280 characters, same tone.What would the biggest critic of this messaging argue? How do I address it? The observation: Three turns. No restarts. Each built on the last. A single prompt cannot do what three iterative turns can do — that's Dialog Engineering.
3
Save Your First Insight
Save something you learned
Think of one thing from today — about Alex, dialog engineering, or AI in general. Tell Alex to save it:
Save this as an insight called "[short title]": [what you learned, in your own words] Here's a strong example for your discipline:
Marketing
Save this as an insight called "Campaign brief context window": The more specific
the audience description and the explicit tone constraint in the first message,
the less time I spend editing AI marketing copy. Brand voice always goes first. Find it again
Now ask Alex to search your knowledge base:
What do I know about dialog engineering? Your insight should appear. It's stored in your Global Knowledge base — available in every project where you use Alex.
See everything you've saved
Show me my knowledge library Bonus: If You Have Extra Time
Try Explain-Like
Explain blockchain like I'm a business executive who understands databases
but has no background in cryptography. Use a real-world analogy. Try Creating a Diagram
Create a flowchart showing the decision process for approving a new project
in our organization. Include: proposal submission, budget review, technical review,
executive approval, and the possible outcomes at each gate. Try the Rubber Duck
I'm going to think out loud. Just listen and help me organize my thoughts.
I'm trying to figure out whether we should centralize our data analytics team
or keep them embedded in each business unit. The centralized model gives us
consistency but the embedded model gives us speed. We tried centralized before
and it created bottlenecks. But the embedded model has led to duplicate work
and inconsistent methodologies...
What pattern do you see in what I just described?