Links, recaps & takeaways from our sessions
Quick Links
I'd be delighted to connect on LinkedIn — follow me for more on AI, facilitation, and business analysis.
What landed, what surprised people, and where the friction was across both days. You're not alone in any of this.
✦ What clicked
You can paste a screenshot directly into the tool — no need to describe what's on your screen. For many people this was the biggest eye-opener.
"I wish I'd known — I could have used it to role-play difficult conversations." Most people hadn't tried it. Now you know it's there.
You can push back, redirect, ask follow-up questions, and steer the output. The more you engage, the better the result.
This is the difference between single-shot prompting (one prompt, take what you get) and multi-shot prompting (an ongoing back-and-forth where each exchange sharpens the result). Think of it less like Googling and more like briefing a smart collaborator.
Personas that might have taken hours to research, write, and design can be drafted in minutes. With image generation, you can give them a face too — useful when you're moving fast or don't have a dedicated research team.
AI models are trained on human-generated data, which means they inherit human biases. But AI can also help surface your own blind spots. Try asking "What am I missing?" or "Who might be left out of this?" — a surprisingly powerful prompt.
⚡ Where the friction was
This is partly prompting skill, partly mindset shift. AI has built-in variability — we looked at what's happening under the hood (temperature, probabilities). That same variability becomes a strength in persona work.
Version control — some strategies were shared in the room; ask if you missed them.
Code → another tool — generating a PlantUML diagram means leaving the AI environment to render it.
Copilot + PlantUML — results varied by participant. Try asking Copilot directly why — different orgs may have different setting lockdowns.
Principles worth keeping close as you keep exploring.
Stuck? Unsure? Just ask. AI is remarkably good at explaining itself and won't judge you for asking something "basic."
If the output isn't right, say so. "That's not what I meant — try again with more focus on X" is a completely valid prompt.
There's no single best AI tool. Ask the same question in two tools and compare. Use one to check another's work.
Try voice mode, paste a screenshot, upload a photo of handwritten notes. The more context you give, the more useful the response.
When you find a prompt that works well for a recurring task, save it. Over time you'll build a personal library of prompts that reliably get you to a good starting point.
Things we built together — click any image to open it full size.
Interactive artifact
Participant work samples
CRUD Matrix — Coffee Shop System
Data Flow — Original vs Updated
CRUD Matrix — E-Commerce Order & Payment Lifecycle
Customer Experience Matrix — GreenMart with Personas
Coffee Ordering Experience Matrix
Package Story — MVP vs Walking Skeleton
Feature Test Case — Morning Pastry Availability
Use Case #2 — Make Payment (Original vs Updated)
AI-Generated Personas — BeanStop Coffee