I've been trying to work out for a while why so many people who are brilliant at their work never quite get round to building the system that would do half of it for them. Even now, with AI tools sitting right there.
Not won't, exactly. Never quite get round to it.
Because the tools are usually fine, and honestly, most of the AI training people get is too.
People leave knowing what the tool can do.
And then I'll talk to some of the same people a few months later, and their week looks more or less the way it did before. The same report, built the same way, by hand, very carefully, by somebody who is genuinely very good at it.
I don't think that's stubbornness.
I think learning a new way of doing a thing costs you a round, and the deadline is today. So the round never gets spent.
Week after week.
I know this one from the inside, by the way. I was that person for years, doing everything by hand, and I would not have thanked anybody for pointing it out. (I'd have been too busy.)
So this programme starts somewhere a long way from a laptop.
The Potter from Belagavi is the game at the heart of AI-Augmented Systems Thinking, my systems thinking training. It's one of the 7 games.
The photos on this page show an early print of the cards. A few numbers have changed since, and the rules below are the current ones.
First, the game
You are a potter in Belagavi. You start with five coins and a pair of hands.
Most rounds you can make three pots by hand, carry them to the santhe, and sell them. Keep at it for seven rounds and you will have made a perfectly respectable pile of coins.
You will also lose.
Because somewhere in round two, one person at your table stops making pots for a moment and buys a wheel instead. Then a mould. Then a kiln that fires the whole batch at once.
By round five they are not making pots at all. The workshop is.
And they are sitting there with more coins than you, having done visibly less work.
In a lot of teams I've worked with, the busiest person is quietly the one the team leans on hardest and the one who has the least room to change anything. That's an uncomfortable thing to notice about somebody who is clearly trying so hard, which is probably why it mostly goes unnoticed.
Why a potter, and why Belagavi
Because I am tired of simulations set in a fictional company called Acme with a fictional product called Widget. I have never managed to care about Acme.
A potter's workshop is something everyone in an Indian room already understands without a single slide of setup.
The wheel is the Kumbarana Chakra. The mould is the achchu. The kiln is the avige. The bullock cart that gets you to the big santhe is the Ettina Bandi.
The cards are terracotta folk art on cream stock, and people pick them up and turn them over before I have finished explaining the rules. A room that wants to touch the thing is a room that has already started playing.
How it actually runs
Three things happen every round, in the same order, and the whole game fits on one printed sheet.
- Invest. Buy one card you can afford, or none. Skills work the round you buy them. Tech starts next round. That one-round delay is where the entire lesson lives.
- Make. Three pots by hand, plus whatever your tech makes for you. The kiln adds four pots every round, but it needs the wheel running first. And on rounds five and seven you are away at the santhe yourself, so your hands make nothing. The workshop keeps going without you.
- Sell. Pots times the price at the santhe. The cart adds a coin on every pot you sell, up to 12 pots.
Most coins after round seven wins.
Seven rounds is deliberately not many. It is just enough time for compounding to beat effort, and not a round more, so the people who kept their heads down get to watch it happen rather than hear about it afterwards.
Two ways to get better, and only one of them compounds
There are skill cards. Nimble Fingers. Steady Rhythm. Fine Finishing. Quick Trimming. Master's Touch.
Each one makes you a slightly better potter, immediately, with no waiting. They are lovely and they feel brilliant to buy.
And there are tech cards. The wheel, the mould, the kiln, the helpers, the cart.
These cost the same coins for the same pots, do nothing at all on the round you buy them, and need each other to work. The mould needs the wheel. The kiln needs the wheel. But once they are running, they make pots every round until the game ends, whether you are standing over them or not.
Skills stop when you stop. Tech keeps going.
The game never tells anyone this. The scorecard does, in round five, in front of everyone.
What you will watch happen at the table
Forty-five minutes, cards and a pencil, and the working habits people carry into Monday come out in the open.
- Why the person who buys nothing for three rounds feels the most productive at the table. Three pots a round, coins piling up nicely. It feels a lot like being careful. It's mostly being busy, and the two feel identical from the inside.
- The skill-card collector, and the ceiling they can't see coming. Fast, satisfying, visible progress every round, right up until round five.
- The quiet one who takes the early loss to build in round two. Occasionally insufferable about it later. Watch how long the table takes to copy them.
- The moment somebody says "wait, show me your sheet." This is where the session turns, and nothing I say beforehand creates it. Somebody loses to a person doing less work and really needs to know how.
- The late convert in round six. They buy the wheel, knowing it is too late, and buy it anyway. That is probably the most useful thing anyone feels all day.
Then the debrief, and this is where AI comes in
The game is the Experience. I mark it with one line - "That was the Experience. Let's unpack that." - and we go straight into the debrief, which runs on the EPPA loop.
- Patterns. The room names what kept repeating. Who bought nothing. Who bought every skill. When the table started copying the wheel.
- Principles. The room connects that to how they work, in their own words. Effort adds, systems multiply. It lands because they got there themselves.
- Application. This is the stage where people usually stall, and it's the only one that changes Monday. So it's the one AI gets to help with.
No AI at the table during the game, ever. The laptops only come out at Application.
And at Application, it goes like this:
- Name one wheel. Each person names one thing they do by hand every week. A small give-away while we're here: if you do it every Monday and could explain it to a new joiner in five minutes, it's probably a wheel.
- Build it rough. With AI, they build the first rough system that does it. Rough on purpose. A wheel that wobbles still beats three pots by hand.
- Run it once before they leave. One person, one AI assistant, supervised, on their own real work. That's the first step on the AI adoption ladder, and it's plenty for a first step.
- Put a date on it. A named behaviour, a real situation, a date. All three, or it didn't happen.
Imagine somebody walking back to their desk with a small, slightly wobbly system they built and ran on a task they were doing by hand that very morning.
That's the whole point of the programme, more or less.
What stays human
AI drafts the system. It doesn't decide what to stop doing by hand, and it doesn't decide what the freed time is for.
Those two calls stay with the person. Which is, I think, exactly where they belong.
The Potter from Belagavi also opens three of my AI programmes. For managers using AI in their own week, there is AI for managers. For senior leaders, there's AI for business leaders, the bigger version with the thinking frameworks underneath it. And if what you're after is hands-on AI training for employees across a whole team, that one is probably the better fit.
A little about the person running it
I've been training founders and leaders since 2019. The founders have been in the US, Canada, Australia, Singapore and India.
On one programme at a Fortune 100 company, six session-days and 113 respondents, the satisfaction score came out at 8.82/10.
On a separate one-day programme for a leading HR-tech platform, 12 of the 13 people in the room said they would apply what they learned. Which is probably the more useful number of the two.
Neither of those is a score for this exact programme, to be clear. They're just the best honest picture I have of what a room with me in it tends to feel like.
Who it's for
- Individual contributors and team leads who are very good at their job and doing all of it by hand.
- People who've been given AI tools and are quietly not using them, mostly because learning the tool costs a round and the deadline is today.
- L&D and HR teams who want a session that ends with something built with AI and already run once.
- Any day on AI or digital fluency that needs an opener people will still be talking about at lunch.
When I'll tell you it's not the right fit
- When the real problem is that nobody has time. The game says invest a round now to earn more every round after. If your people genuinely have no round to spare, it will feel like a taunt. Fixing the load first would probably help them more, and I'll say so.
- When you want team dynamics. This is about how one person spends their own effort. If you'd like to watch a team talk to each other, Sticky Fingers or Ripple Effect would serve you better.
- When you need training on one specific platform. There's no tool demo here. A vendor's own training does that job well, and this can sit in front of it to make people want it.
The questions people usually ask
"We've already done AI training."
That's fair, and honestly most of it was probably fine. This one is mostly about getting one real task off someone's plate, built and run once, before they leave the room.
"My people don't have time for a game."
That's a fair worry, and it was mine too. The game is forty-five minutes. What I've found is that it's the forty-five minutes that makes the AI part land, because by then nobody needs convincing that the round is worth spending.
"Will this work for people who aren't technical?"
I think so, yes. The game is cards and a pencil, and the first system each person builds is rough on purpose. The only real ask is noticing one thing they do by hand.
"Can AI really help with something this small?"
Probably more than with something big, in my experience. Small and weekly is where a first rough system pays back fastest, which is the whole lesson of the wheel.
Worst case, we have a conversation and you come out of it with a clearer brief.
Come and see if it fits your team
Tell me a little about your people and the work they do by hand, and I'll tell you honestly whether this is the right programme for them.
Come and have a look at the calendar
Make the ceiling visible. Make the compounding visible. Let somebody lose to it in front of the room, then build their own wheel. That's the whole design, which is either elegant or lazy depending on which day you catch me.
With love from Bengaluru, this is Arvindh saying over and out.