Practical. No hype.
The most enjoyable thing I’ve built with AI this year has nothing to do with surgery. It’s a debate council — six historical figures convened around a table to argue any question I throw at them.
Socrates asks questions that destabilise everyone else’s answers. Marcus Aurelius finds equanimity in the cosmic indifference of it all. Machiavelli calculates. Mary Wollstonecraft spots the argument that excludes everyone without a seat at the table. Sun Tzu identifies the actual objective before anyone has noticed there isn’t one. Nietzsche burns the whole thing down and offers something better from the ashes.
There’s a seventh chair. I fill it depending on the question — whoever has the vantage point the other six cannot generate between them. The council debates. Then it offers a synthesis: where agreement was found, and what remains genuinely unresolved.
I call it the Ancient Council. I use it to think — not clinically, for anything. Ethics, difficult decisions, questions I’m genuinely uncertain about and don’t want to resolve too quickly. The council disagrees in ways grounded in each thinker’s actual writing, not synthetically, not diplomatically.
The second one has football managers. Cruyff on space and intelligence. Sacchi on collective organisation. Clough on simplicity and trust. Shankly on the moral dimension of the game. Mourinho on winning the psychological battle first. Bielsa burning with idealistic intensity and regarding a cautious approach as a kind of dishonesty.
The Football Council debates match-ups, tactical questions, hypotheticals, transfer decisions. Cruyff and Mourinho agree on almost nothing. Clough punctures tactical intellectualism with blunt authority. Put any question about the game in front of them and it becomes a genuine fight — because the differences between these philosophies are real, not invented.
Neither of these setups is productive in any measurable sense. They won’t save you an hour. They won’t write your referral letters. They have no clinical application.
They are play things. That’s the point.
The useful tribe exists too
The work setups are covered in this setup guide and this modes explainer. A project that writes in your voice. Another with your guidelines uploaded for literature review. A thinking partner that pushes back rather than agrees. Those handle the work.
But here’s the thing: building a work tribe and building a council of ancient philosophers is the same skill. You’re deciding who is in the room. What they know. What makes them genuinely different from each other. What each one notices that the others won’t. The difference is what you want from the conversation.
The councils are the purest version of the exercise — there’s no practical constraint, no goal beyond “who has an interesting perspective on this question?” — which makes the logic of the skill clearer, not less. Once you’ve worked out how to make Nietzsche and Wollstonecraft fight productively about equality, writing a system prompt for a letter-writing assistant is fifteen minutes.
How to build yours
Pick a topic where you have genuine opinions and where people reasonably disagree. Football, wine, political history, music, architecture, Formula 1, literature — anything with real fault lines. The topic doesn’t need to be serious. It needs to matter to you.
Pick four to six voices with genuinely different frameworks — not just different conclusions. Cruyff and Mourinho don’t just reach different answers; they are working from entirely different premises about what football is for. That’s the level of difference that produces good debate. If your voices agree too easily, they’re not different enough.
Write each one so it can’t be confused with another. Give them a distinct register, a set of core beliefs, specific texts or a specific career to draw from. Let them address each other by name. Let them be rude.
Then ask them something.
The councils work because the friction is real. Sacchi and Clough on collective organisation versus individual freedom is a genuine philosophical disagreement with decades of documented evidence behind it. That friction produces something a single-perspective response cannot. The same principle builds everything else in the tribe — the more precisely you define the voice or the perspective, the more useful the setup becomes.
I use the councils for thinking I don’t want to do alone: questions worth arguing about but not in a meeting room, ideas I want stress-tested without a social cost. The philosophers for ethics and principle; the managers for anything involving competition, strategy, or pride.
Build the useful tribe. Build the pointless one too. Only one of them is work — and you’ll probably learn more about AI from the one that isn’t.
Want help building a tribe? Logged-in members can book a session — 30 minutes (£75) to set up a work project, or 60 minutes (£150) to build the whole thing: the useful setups and whatever fun project you’ve been thinking about.
See also: Setting up Claude properly: a 15-minute guide for surgeons · Chat, Cowork, Code, Skills, Plugins, and Prompts: what is what and when to use each