Ten seats
Never more than twelve. That is the whole room.
Ten people in a room, and the decisions you actually have to make. Not a lecture, not a certificate, and nothing you could have read.
Most AI training teaches tools. Six months later the tool has changed and the training is worthless. The decisions do not change that fast — what to commit to, what to hold, what stays human, what you answer for when a machine got it wrong. Almost nobody teaches that, because almost everyone teaching is also selling something.
Ten is deliberate. Above that people stop saying the true thing. Below it there is not enough disagreement to be useful.
Never more than twelve. That is the whole room.
A proper room, in a city we announce. Not a hotel suite.
Where distance makes the room impossible. Not as the default.
No vendors. Nothing sold from the front.
We travel. Fees reflect it.
First time in the seat, and the first thing on the agenda is a decision about AI nobody prepared you for.
You have been handed a transformation and a budget, and told to work out the rest.
Not the demo, not the hype. What it does, what it cannot, and where the judgement still has to be yours.
You are accountable for it. You would like to interrogate what you are told, in your own industry's language.
Sold per seat — you do not need an organisation behind you to sit in one. Rooms open city by city. Tell us which one you want and we will tell you when it runs.
You are making decisions about something nobody has explained to you properly.
What the models actually are and how they differ. What an agent is, and how it is not a chatbot. Where the tools sit, what they cost, what they are genuinely good at and what they reliably get wrong. The shape of the whole landscape — so when the next thing arrives you can place it instead of starting again.
The afternoon turns to what is being sold to you. Every demo works. Every business case is optimistic. What separates a real capability from a good deck, and the questions to put to the next vendor who walks in.
You leave with — the ability to look at any AI tool and say what it is, what it would replace, and what could go wrong with it. Plus a written framework for interrogating the next proposal you are shown. Every other room assumes this one.
You have delegated AI to people who cannot make the decisions it requires.
You leave with — a scored read on your own organisation, the four decisions you cannot delegate written down, and a ninety-day plan you signed.
Register interest →You are accountable for decisions you cannot yet see or evidence.
You leave with — an accountability map for your own organisation, the questions to put to the people building it, and what a regulator will actually ask for. Calibrated to your industry.
Register interest →What has to change in how you think, decide and design?
You leave with — your assumptions surfaced and tested, a redesign of one real decision loop from your business, and a peer holding you to it.
Register interest →Can your 2015 operating model carry a 2030 strategy?
You leave with — a map of your operating model against the ambition it carries, the three bottlenecks costing you most, and what to change first.
Register interest →Not sure which room? Tell us the decision in front of you and we will tell you which one fits — or that none of them do. Talk to us →
Per seat. Ten to a room. We travel — fees reflect it.
Request a seat →Every room above runs closed, for one organisation. Same format, same ten people — except they all work for you, and the material is your decisions instead of worked examples.
This is where most of the work happens. An executive team gets further in a day together than five of them get separately over a year.
Governance, accountability and the questions to put to management. Calibrated to your industry and your regulator.
One room, one decision loop, everybody who has to live with the answer already in it.
The people running the work, with the workflows they actually run rather than a case study.
One theme at a time, with someone who has spent a career inside it. Not a trend report — a view on where this thing is in five years, in ten, and in 2050, and what someone living it should be doing about that now.
Each runs the same way, and each step decides whether the next one is worth taking.
Podcast, with the specialist
Written. Five, ten, 2050.
Ten people living it
Only if the table asks
Insurance first. Then wherever the specialist is better than us.
Building something now, with tools that did not exist when the playbook was written.
People who left the seat and built something of their own. What actually happened.
Who holds authority in the AI era, who does not yet, and what changes that.
The series opens shortly. Register interest in a theme →
Ninety minutes, the same ten people, no charge and nothing sold. The diagnostic goes out beforehand and the result is yours either way.
Conversations with the people who actually answer for this, not the people selling it.
Coming.
What we learn inside the work, written down. Free, and nobody has to give us an email address to read it.
A certificate is something you are given. Proof here is the difference between your score walking in and your score a year later.
Nothing is recorded and nothing is watched alone. The value is the other nine people.
Sending someone in your place defeats the point. The shift has to land in the person making the decision.
Ten seats. Then the room is full.
Tell us which city, or bring your own ten.
UNLEARN · RELEARN · REINVENT
Continue reading → What we learn inside the work, published. · Next · 06 · Atlas