What You’ll Learn

After watching a Daniel Priestley video on the “podcast pyramid” — the idea of guesting on other people’s shows to borrow their audience — James wanted more traffic but had no time to chase it. So he handed the whole job to Dean, his campus operating system, live. This walkthrough shows what happened and, more importantly, the mindset shift behind it: stop asking AI how to do something and start asking it to do it.

By the end you’ll see how a single outcome-focused request turns into a running outreach system, and why the operating system’s memory — your goals and ideal customer profile — is what makes that possible.

The Shift: Outcomes, Not Instructions

Here’s the change worth internalizing. Early on, you might have asked ChatGPT to “find a list of podcasts I could be on,” then figured out who runs them, how to contact them, and why they’d say yes — you doing the project management at every step. James does the opposite:

“I’m not asking how it should be done. I’m asking you to actually go out and physically do the work.”

💡 In Plain English: you’re not asking for a recipe, you’re hiring the cook. You describe the meal you want; the team handles the rest.

The Request

James gives Dean one instruction: I want more traffic beyond YouTube and the free community. I want to be a guest on two video podcasts a week, talking about AI for educators, coaches, and creators. Find the shows, handle the outreach, I’ll do the interviews — walk me through it as you go.

What Dean Actually Did

1. Checked it against the goals. Dean doesn’t just say yes. He first asks whether the request serves the bigger number — the standing business goals he already holds. James had actually paused some marketing activity for lack of time, so Dean confirmed this was worth fitting in before proceeding.

2. Made it real. Instead of treating it as a one-off, he wrote it into the goals file as a standing weekly target: two podcast spots booked and aired per week. Now every session and every department steers toward it.

3. Set the selection rule. Video podcasts for educators that take guests, ranked by audience fit — not vanity subscriber counts. A tightly matched small show beats a giant generic one, because the goal is real traffic into the community and campus, drawn from the ideal customer profile the OS already knows.

4. Found and ranked the shows. Dean pulled a shortlist — the course creator community, Her Empire Builder, an online course show on leveraging AI, plus a dozen more behind them.

5. Wrote the outreach. A personalized pitch per show, in James’s voice, referencing each host’s audience with his proof baked in — staged as drafts so nothing sends without approval. No cold-firing.

6. Kept it running. It sources on a schedule and tracks every show through a pipeline: found → pitched → booked → done. A one-off request became a working system.

Why It Works: The Memory Underneath

This only works because Dean operates inside a system with memory. On startup he reads the goals file that every AI employee can see, so the whole office pulls in the same direction. He also knows the ideal customer profile, so “audience fit” isn’t guesswork — he knows exactly who both James and the podcast should be serving. That shared context is the difference between a lone assistant and an accountable team.

The Takeaway

Next time you see a traffic idea from a speaker or creator you follow, don’t ask Claude how to execute it — ask it for the outcome. “I want to be a guest twice a week on shows in this market.” Then let it figure out the skills, departments, and steps. The first time, Claude will interview you to capture your goals and ideal customer profile; after that it runs. If you want it prebuilt, you can install Dean (link on screen), and James also has a full course on setting Dean up manually.

Teach more, and let the agents do the rest.

This tutorial is a recap of a short live session.