What You’ll Learn

Most AI tools agree with you. You open Claude, Gemini, or ChatGPT, you already know what you want, and the tool hands it back with a smile. That feels productive. It’s actually how bad ideas get built fast.

In this session, James walks through the fix he uses every day: an Advisory Council of digital advisors that challenge a decision before any work gets done. You’ll see the five permanent “seats,” the special sixth seat, and a live demo where the council talks James out of a pricing mistake.

The Problem: A Yes-Man in Your Chat Window

Look back at how you started your last few AI sessions. In most of them, you already had the answer in your head. You just asked the tool to produce the output. The AI agreed and delivered.

That’s the trap. Newer models like Opus 4.8 push hard against being a yes-man, but they still lean toward agreement. And almost every agent workflow people build is after-the-fact work: research, organizing, posting, creating. Nothing in that stack stops you and asks, “Why are you even doing this?”

The Advisory Council fills that gap. It’s the front half of the work, before the doing starts.

The Fix: Five Lenses That Disagree on Purpose

Think of a table. There are five advisors seated, one guest chair, and one seat for you. These advisors aren’t real people. They’re five fixed lenses that pressure-test any idea of consequence. They are set to disagree on purpose, so the real angle survives and everything else gets stripped away.

1. The Customer

This advisor forces you to explain the idea in your customer’s own words. It strips away your insider language and asks, “What does this actually mean to the person you serve, and where are they right now?”

2. The Skeptic

The skeptic names the obvious objection you’d rather hide under the table. Want to sell a $2,000 cohort when you’ve only sold a $97 course? The skeptic asks the question that saves you from flopping: “Will they really pay for that?”

3. The Contrarian

This is the lens that flips everything. What if you did the opposite? Does that work better? It pulls ideas out of you that you’d never reach by playing it straight.

4. The Numbers

The numbers advisor asks the uncomfortable money question: does this make money, save money, or save time? If it isn’t something you can sell or measure, “I can do it” isn’t a good enough reason to do it. This is the seat most educators and coaches skip — and the one they need most.

5. The Brand Voice

This lens checks fit. Is this on-brand? Is it the way you do business and the kind of people you want to serve? It catches the moment a serious brand starts posting off-key jokes and drifting out of character.

The Sixth Seat: The Guest Lens

The guest chair is the interesting one. It isn’t a fixed role — it’s an open seat the council fills automatically based on your problem. The AI nominates a legendary thinker’s framework to apply, not the celebrity themselves. Talking pricing? The council seats a value-equation lens. The rule to remember: it’s the thinking method that joins the table, not the person.

The Live Demo: Talked Out of a Pricing Mistake

In the demo, James talks to Dean, his chief orchestration officer inside the Campus AI Operating System. A routine request (“remind everyone about Monday’s sprint”) doesn’t trigger the council — it’s low-stakes and reversible.

Then he raises a real decision: charging $2,000 for a cohort to customers who’ve only paid $97. The system stops. It flags the 20x jump, refuses to just act, and recommends running it through the council before committing.

“This is exactly the kind I’m supposed to pressure test before we build anything — not just log along.”

The council goes to work. The customer lens asks who this is really for. The skeptic flags a guarantee that’s outside James’s control. The numbers lens questions the list. The guest value-equation lens reframes it as a perception problem. The output isn’t a yes — it’s a sharper plan: run a founders’ beta, decouple the risky guarantee, and do a 90-day post-mortem before scaling.

Why This Changes Your Outcomes

AI is great at the back half of the work — organizing, collating, producing the outputs you already know you want. The Advisory Council adds the front half: it makes you bring your own expertise and judgment to the table before the work begins. That judgment gets embedded in everything the agents build afterward.

There’s also a reporting layer that closes the loop. After Dean implements a plan, the council can ask what worked and what didn’t — so the next time you weigh a course, a cohort, or a big price change, it remembers the last one.

Key Takeaways

• AI tools default to agreement — that’s how half-baked ideas get built fast.
• An Advisory Council adds the missing front half of the work: pressure-testing a decision before any output is created.
• Five fixed lenses — Customer, Skeptic, Contrarian, Numbers, Brand Voice — are set to disagree on purpose.
• A sixth “guest” seat is filled automatically with the best thinking framework for your specific problem.
• A gate keeps it from challenging everything — only decisions of real consequence or that can’t be reversed trigger the council.
• A reporting layer remembers what worked, so each big decision learns from the last.

Your Next Step

You don’t have to build this from scratch. James packaged it as the Strategy & Research team you can install directly into Claude Cowork, Claude Code, or Codex — the team you talk to first, before you ask anything to get done. Learn more at trainingsites.io/research, and join the free community to see how it fits inside a full Campus AI Operating System.