You have likely seen conversations about people building AI bots that check each other’s work, debate perspectives, or serve as “teammates” with different skill sets.
A founder, an economist, a therapist, and a historian will often approach the same problem completely differently. A designer and an engineer will disagree on which problems matter most. An optimist and a skeptic will notice different risks. There’s value in giving ourselves access to perspectives different from our own - just like there is value to diversity in teams.
This week, we’re going to build a room full of AI personalities and let them debate ideas, critique each other, and collaborate.
Before we start: if you are getting this email forwarded to you and are not yet part of the lab - you can join the tiny AI lab here.
The Challenge:
This week, you're going to create an AI roundtable discussion.
pick a topic, problem, or decision
create multiple AI personalities with different worldviews or expertise
let them interact with each other
You can make this as practical, philosophical, or funny. It’s totally your call. The goal is to see AI adopt these different perspectives and understand how you can use these multiple perspectives or personality structures to shape the value that you get from the tool.
The prompt to get you started
Drop this into your AI tool of choice:
I want you to simulate a roundtable discussion between multiple AI personalities discussing a topic I care about.
First, help me design the room. Ask me:
what topic or decision I want debated
how many personalities should be involved
whether I want realistic experts, fictional characters, historical figures, or personality archetypes
what kind of tone I want (serious, strategic, chaotic, funny, philosophical, etc.)
Once we’ve designed the room, run the discussion.
Each participant should:
have a distinct worldview, communication style, and priorities
challenge and respond to each other directly
build on or disagree with previous points naturally
avoid sounding repetitive or overly agreeable
Structure the conversation like a moderated roundtable with:
opening statements
discussion and debate
challenges or critiques between participants
a moderator summary of agreements, disagreements, and surprising insights
Too Easy?
Try creating specialized agents with different jobs inside the conversation.
For example one agent researches, one critiques assumptions, one plays devil’s advocate.
Then ask the system to refine itself over multiple rounds.
I want this discussion to behave more like a coordinated AI system. Assign each participant a specific role in the reasoning process. Include:
a researcher
a skeptic
an optimist
a synthesizer
and a moderator
Have each participant improve or critique the others’ reasoning before arriving at a final synthesis.
Why this matters?
One of the biggest shifts happening in AI right now is the move from single prompts to systems and agents. Increasingly, companies are experimenting with AI agents that review and improve each other’s work.
Much like a team of people would work together to own parts of a process, give each other feedback, and make the overall output of the work stronger. Instead of one AI generating a final answer, you get teams of agents working together.
This challenge is a playful version of that idea.
That’s it for this week.
Hit reply and tell me what you thought of the challenge this week.
I read every response, and your experience helps me shape these challenges. If you get stuck or have questions, reach me at [email protected]