What You’ll Learn
James built a product yesterday. Today he launches it — live — and he does the part most educators dread (the selling, the copy, the setup) with one prompt. In this session he hands the entire go-to-market to Dean, his AI orchestrator, running on Fable 5, and watches it build the course, the emails, the social posts, the community campaign, and the shopping cart. This walkthrough shows exactly how that works and why it matters.
By the end you’ll understand the three stages of shipping a digital product, what a launch actually requires, how a single well-formed prompt kicks off the whole thing, and the memory system that makes the orchestrator this capable.
The Setup: You Only Talk to the Orchestrator
James opens the paid version of Claude on the desktop with his campus operating system in the working folder, and says hello to Dean. That’s the whole interface — he doesn’t touch individual tools or employees.
“It’s not me trying to use the tools. I’m letting Dean do the work — he orchestrates all of the employees through the playbooks he’s built.”
💡 In Plain English: you’re the boss handing a project to a manager, not a person clicking through ten apps. You describe the outcome; the team executes.
The Three Stages of a Digital Product
James frames every product as three parts:
• Stage 1 — Pre-work: the research and strategy (handled by his Atlas team; covered in a future video)
• Stage 2 — Build: actually creating the product (he did this yesterday)
• Stage 3 — Launch: everything required to sell it — this session
The One Prompt That Started It
Here’s the striking part. The entire launch began with a single instruction. James told Dean, in effect: run phase 4 launch of the Prep Agents Team, here’s the working document, follow it exactly — stages one through three (the build) are done, so the copy team writes all the buyer-facing copy and the course, mint the run ID, and start by condensing the research brief.
That was it. He answered five or six approval questions along the way — but that one prompt drove the whole thing.
What the Launch Actually Produced
Dean first condenses the research brief from the build stage so it knows what the product is, its purpose, features, and benefits, plus the price ($97) and the ideal customer profile. Then it builds the full go-to-market: the course, the email marketing campaign, the social media campaign, the community campaign, and the shopping cart with product descriptions.
Crucially, James stays in the loop at the points that matter. The playbook identifies where he wants to approve — he isn’t handing everything off blindly. Dean pauses, confirms the brief, and only then proceeds. The whole launch ran in about 90 minutes.
Why It Works: Memory Plus a Built-In Wiki
The reason the orchestrator is this capable is context. On startup Dean reads James’s goals, outstanding tasks, ideal customer profile, offerings, and connected tools — so it already understands the business. On top of Claude’s default session memory, James’s operating system keeps the last eight days of working memory and trims from there.
The compounding piece is a built-in Karpathy-style LLM wiki. At the end of each session James just says “wrap up the session,” and it updates the memory and tasks, logs every output, and writes what was learned into the wiki. He’s even ingested 700 of his YouTube videos into it. So each new session starts with recent memory and two-plus months of accumulated context — far more than default Claude memory alone.
The Takeaway
Three hours from building a product to shipping a product. If you’ve believed creating and launching products takes too long, James’s point is that it no longer does — the barrier was never the work, it was doing it alone. Give an orchestrator good context, a clear playbook, and a few approval checkpoints, and the launch runs itself. If you want the system, the campus operating system is at trainingsites.io/os.
Teach more, and let the agents do the rest.
This tutorial is a recap of a live launch session, and the companion to the build session on the same product.