How will you help your team members learn AI?
Use AI to interview yourself about ideas, producing shareable product briefs from bullet points in under 10 minutes. Read: "Everything Starts Out Looking Like a Toy" #260
Hi, I’m Greg 👋! I write weekly product essays, including system “handshakes”, the expectations for workflow, and the jobs to be done for data. What is Data Operations? was the first post in the series.
This week’s toy: weather forecasts in your browser console (or terminal). It’s taken us a long time to realize that low-fi information, well-designed, is still quite useful. Why not add more terminal-based widgets instead of building a fancy UI? Edition 260 of this newsletter is here - it’s July 21, 2025.
Thanks for reading! Let me know if there’s a topic you’d like me to cover.
The Big Idea
A short long-form essay about data things
⚙️ How will you help your team members learn AI?
We wanted to help teammates to start using AI as a practical tool and stop thinking about it as magic, and thought a "speed dating" style demo would help. The goal was to have each person or team left with a concrete example of an idea transformed into a product brief making it easier to visualize their idea.
In the workshop, I ran a "try AI in 5 minutes" demo 9 times over a 90 minute time period. Why did we do this? It's easier to answer questions and edit a document than it is to start from a blank page.
Most people are great at having ideas but struggle to turn them into something shareable. The gap between "that's a good idea" and "here's a product brief" is where good concepts often get stuck. Most people don't have time for a full document but do have time for a 5 to 10 minute facilitated interview.
Iterative prompt building is the key insight
An approach where you do a multi-step prompt to AI turns the daunting task of "write something great" into the manageable task of "answer some questions." By turning the product (and process) development process into structured questions, you get people to be more creative by unsticking them.
This approach might be worth trying if you feel your team is sitting on ideas that never get shared, or if the time to turn ideas into features just takes too long.
Start with the template below and let the AI improve it. Now, see what happens when you treat idea development as a conversation rather than a writing assignment.
A structured interview process
Here's how to create your own iterative prompt-building process:
Step 1: Create the initial prompt
You'll need a template to defined the desired content, role, and outcome. You're not building a final output, but a prompt that will help the chatbot focus and give you a solid template to improve.
Step 2: Let the AI improve it
Next, feed your initial prompt back into the AI and ask it to create an interview prompt. The AI understands the context and creates a customized interview.
Step 3: Conduct the interview
Now, answer the questions! The AI interview handles the task of feeding you questions in the right order, capturing you response, and building a coherent narrative.
Step 4: Generate the output
When you're done, you simply say "give me the output" and the AI produces a polished, structured document ready to share.
This process breaks down the overwhelming task of "write a product brief" into manageable, conversational chunks. Instead of staring at a blank page, you're having a focused conversation.
Turning the interview process into a demo
When I discovered this process, I didn't initially know how to turn it into an engaging demo. Then I thought: this is a great way to demonstrate to teammates directly how they can interact with AI, even if they don't know exactly how it works.
To customize this playbook for each demo, I asked the teams:
What role do you do at the company?
What's something you'd like to change?
What's one thing that might improve the thing you'd like to change?
Say you're a customer service rep at a SaaS company and you want to propose an improvement to the product to make it easier to send reminders to the internal team when there's an open issue for a customer. You'd like for the reminder to be automatic and then stop when the Jira issue is closed.
You might start with bullets like:
Auto-reminders for open customer issues
Stop when Jira ticket closes
Notify internal team
Reduce manual follow-up work
Then, use the prompt template and the interview process to reach a product brief to share with other teams.
Here's the prompt template I used in my demo.
The prompt template
Start here - paste this prompt into Gemini, Claude, or similar:
You are a [provide role] for [company name], [company description] that provides [product or service]. You're really interested in [background, backstory of motivation]
Let's write a brief (1 page) pitch for a [intended outcome, like new feature description].
I want you to give me a prompt that I could use later with a chatbot to interview me iteratively about [use case, like feature development] with the goal of producing short (1-2 page) outputs that I can share with [intended audience, like product development] to demonstrate how we can improve [example, like "our product" or "related services" or intended outcomes]. I have 5-10 minutes to spare at a time and I'm not particularly technical, though I have used ChatGPT before and I know a lot about [company name] because I work there.
How you use the template to produce the result
You feed it in.
Then you take the prompt it produces and feed it in again.
Then it interviews you
When you're done, you say: "give me the output" or "create the product brief" or "ship it"
Here's an example of what gets produced:
Step 1: Initial prompt
Here’s the prompt you use to start the process:
"You are a customer service representative at a SaaS company. I want to write a brief for an automatic reminder system that notifies the internal team about open customer issues until the Jira ticket is closed."
Step 2: AI-improved interview prompt
You’ll get something like this:"Hi! I'm here to help you develop a product brief for your automatic reminder system. I'll ask you some questions to understand the full scope and impact of this feature. Let's start with the basics:
What's the current process for following up on open customer issues?Who specifically needs to be reminded, and how often?What information should each reminder include?How do you want the reminders delivered (email, Slack, etc.)?What should trigger the reminder to stop?What's the business impact if this feature gets built?
Take your time with each question - the more detail you provide, the better the final brief will be."
Step 3: The interview
The AI then guides you through these questions systematically, capturing your responses and building a coherent narrative.
Step 4: The output
A structured product brief ready to share with your team:
Automatic Customer Issue Reminder System
Problem: Customer service reps manually track open issues, leading to delayed responses and dropped customer concerns.
Solution: Automated reminders that notify internal teams about open customer issues until the Jira ticket is resolved.
Key Features:
Daily reminders for issues open >24 hours
Jira integration to detect ticket closure
Configurable notifications (email, Slack)
Escalation schedule with manager alerts
Success Metrics: 50% reduction in manual follow-up time, 30% faster issue resolution
Business Impact: Improved customer satisfaction through faster issue resolution and reduced manual overhead.
This process produced several ideas that I hadn't heard from the team from a very brief process. So why does this work?
Why this approach works
Conversational flow - The interview format feels natural and engaging. You're not struggling to organize your thoughts—you're just answering questions.
Systematic coverage - The AI ensures you cover all the important aspects of your idea. No more forgetting to address key questions like "who is this for?" or "how do we measure success?"
Iterative improvement - The two-step process (improve the prompt, then use it) creates a much better final result than starting with a basic prompt.
Time efficiency - You can do this in short bursts. Start a conversation, answer a few questions, come back later and continue. The AI remembers where you left off.
The future of idea development
This approach has implications beyond just product briefs. The same iterative prompt-building process could work for:
Technical documentation - Interview yourself about a system you built
Process documentation - Walk through a workflow step by step
Research summaries - Structure findings into shareable insights
Meeting notes - Turn scattered discussion into actionable outcomes
The broader insight is that AI is most effective when it's guiding a conversation rather than trying to read your mind. By creating prompts that interview you systematically, you can turn any half-formed idea into something structured and shareable.
What's the takeaway? The key insight is that we're not just using AI tools to ideate. We're creating a directed collaboration system where the prompt itself becomes a conversation partner.
Links for Reading and Sharing
These are links that caught my 👀
1/ Get started! - If you’re not an engineer yet releasing code, you’ll be much better at your job when you start releasing code, even if you have no idea what you’re doing. Getting the reps is the thing that matters most in driving improvement. (Don’t worry about how cringeworthy your early code might look to you later.)
2/ Vibe Kanban - The next logical step beyond letting agents code some of the work for you is managing that work in a system like a Kanban board. It makes sense that you need a higher-level abstraction to arrange multiple workers, even when those workers are AI bots.
3/ A free open-source video editor - I needed this last week when I was trying to reduce a demo video down to a meaningful size and convert it from an .mov to an .mp4. I eventually figured out how to get it done with iMovie, but this would have been much faster to edit, scrub, and downsize a simple demo video.
What to do next
Hit reply if you’ve got links to share, data stories, or want to say hello.
The next big thing always starts out being dismissed as a “toy.” - Chris Dixon