Run one yourself.
You do not need to be an AI expert, and you do not need our permission. If you have a room, a couple of hours, and a group who are up for making something rough together, you have enough to start. The kit below is free, and it is yours to change.
Setting the stage
You've probably got most of this already.
Tables that seat 3–5, a screen you can project to, and wall space or a whiteboard. Movable chairs matter most.
One leading from the front, one roaming between tables. One person can do it, and the roaming is what catches teams that have stalled.
At least one device per team, and free-tier access to Claude, ChatGPT or Gemini. Teams share screens; not everyone needs an account.
Sticky notes, markers, scrap paper. Several boosts need nothing else, and paper prototypes often beat the built ones.
Group size
How many people, and how many facilitators.
Below the minimum the energy drops and peer feedback thins out. The room stops being a room. Above the maximum, teams stall without anyone noticing.
| Length | Participants | Facilitators |
|---|---|---|
| 2 hours | 12–25 | 2–3 |
| 3 hours | 16–32 | 3–4 |
Three to five people. Pairs run out of perspectives during the Designer phase; six-person teams leave someone with nothing to do during the build.
Teams that mix roles (a teacher, a coach, someone from operations) produce sharper problem statements than teams of people who all do the same job.
The kit
What is in the kit.
Everything below is free to use. Open the slide deck on the day and work through it with the room.
Run of show
The full arc, phase by phase, with what each one ends in and what to cut when time is short.
Slide deck
The deck you run the room from, for the two hour and three hour versions. Make a copy and edit it for your room.
AI boosts
Ready-to-paste prompts across the three phases, in four languages.
Human boosts
Table activities with step-by-step facilitation instructions and tips.
Prototype gallery
Where teams post what they built so the whole room can browse it mid-session.
Adaptation examples
How the same three roles work across different challenge areas and audiences.
Access
Two ways to run one.
Take the free kit and run it yourself, or bring us in to train your facilitators and unlock the companion app.
Run it yourself
Everything on this page. The boost libraries show a starter set, chosen so a room can navigate it without much help from you.
Training and the companion app
We run private labs and train facilitators for groups of 25 or more. Smaller groups, start with the free kit above.
We train your facilitators, and you get the AI companion app. It runs the process alongside you as you deliver it to your own community, and it fills each team's own words into every prompt. Training groups get the full boost library, and first call on which new boosts we build next.
Before the day
What to prepare.
- Pick your challenge area, or decide not to. Teams naming their own challenge produces more ownership, and a pre-set challenge saves twenty minutes. Either works.
- Play with an AI tool for half an hour. That is enough. You are running the room, not teaching the tool. If you have used Claude or ChatGPT once, you are ready.
- Read the run of show once. Know what each phase ends in. That is the only thing you have to protect when the timing slips.
- Test the room. Wi-Fi for the number of devices you expect, and check whether your district blocks the AI tool you planned to use. This is the single most common thing that goes wrong.
FAQ
Frequently asked questions
Which length should I run?
Two hours is the common one, and the one we would recommend by default. Every phase survives at that length. Go to three hours if you have a professional development day and want real time for teams to push back on each other's findings.
People ask about ninety minutes. It can be done, and something has to give. Pre-set the challenge before the room arrives, offer one boost per phase, and expect the reporting round to get thin. If you have the choice, take the extra half hour.
Can I run this alone?
Possible, not recommended. Two is the minimum that works well, because a team that has stalled is the failure mode that costs the most and only the roamer catches it.
Do I need judges?
No. This is not a competition. It is a community of experimenters sharing what they found, and scoring gets in the way of that: a team being judged hides what broke, and what broke is the useful part.
Think instead about who is already in the room who could lead the feedback. Someone who has run the thing you are all circling, or who works closest to the people you are designing for. Ask them to respond to what teams report, and the room does the rest.
What if I have never used AI tools?
You do not need to be an expert. Every AI boost is a ready-to-paste prompt, and teams fill in their own words. Your job is the room, not the tool. Half an hour with Claude or ChatGPT beforehand is enough preparation.
Can I adapt it for a different challenge area?
Yes, that is the design. The three roles work whether you are tackling student mental health, teacher professional development, scheduling, or something that has nothing to do with classrooms. The process stays the same; you center it on your own context and learners. See adaptation examples.
Do participants need expertise in the challenge area?
No. The session walks them through understanding the challenge as part of the design work. Curiosity and a real stake in the topic matter more than prior expertise.
What if a team builds something harmful or inappropriate?
Ask probing questions before you redirect. Often the team has already noticed and has not said so. "Who could this hurt?" and "What happens when this is wrong?" do most of the work. Flag it to the lead facilitator, and if something is posted to the gallery it passes an automated moderation check first.
Can this be done virtually?
The format is designed for in-person and we have not run it fully remote. The parts that would need rethinking are the human boosts that depend on being at a table together. If you try it, we would like to hear how it went.
What does it cost?
Depends. The kit on this page is free and stays free, and everything in it runs on free-tier AI accounts. You are not required to tell us you ran one, and we would like to know.
Facilitator training and access to the companion app are priced per group. The price moves with the size of the group, how many facilitators you want trained, and how much of the delivery you want us to carry. Tell us about your group in this short form and we will work it out with you.
Tell us you ran one.
We are collecting what happens when other people run this: what you changed, what broke, what your teams built. It is how the format keeps improving.