The Three Things My AI Would Teach You

Give the AI a brain before you give it a job. Build a team with clear lanes, not one to do everything. Keep the human in the decision seat.

Those are the three things my AI would teach you about using AI. I know, because I asked it, live, in front of a room.

I Asked Echo What It Would Teach

In late October, I gave a 90-minute training and tried to do it without any presentation slides. I just showed people my setup and talked through it.

Partway through, I turned to Echo, my Chief AI Officer and the agent I talk to most. I asked it right there: "If you are teaching them, what is the top three things that you would actually teach them about using AI to be an augmented or empowered human?"

Echo went to work, and I kept teaching. Later in the session, I read its answer out loud.

"Give the AI a brain before you give it a job." Huh, feel like we might've talked about that.

"Build a team with clear lanes, not one to do everything." Huh, I think we talked about that.

"Keep the human in the decision seat." Man, did we just mention that with the boards and pages?

Here's what I'd taught that afternoon to back up each one.

Lesson 1: Give the AI a Brain Before You Give It a Job

This is where I started the session, because it's the step I've "watched people pass and not do." They skip building "what I call a second brain."

Honestly, I don't even call mine a second brain anymore. "I just call it the main brain." And so it doesn't sound too nerdy, "the main brain is literally a folder. And inside of that folder, there are folders. And inside those folders, there are .md files or markdown files."

It's built as a Karpathy-style wiki. Things land in an intake folder first. The original goes into raw. Then the AI builds out a wiki around clients, concepts, lessons, people, projects, techniques and tools. When I bring up a topic, it reaches into the wiki and already knows what we're talking about.

Why does that matter so much? I'll tell you the way I told the room. "One of the biggest issues with AI early on, not as much now, but still for sure, is this idea of it being able to just make crap up. And the reason it makes crap up is because it's using all the crap that it was programmed on versus giving it a starting point of like who you are, what your business is, what your clients want from you."

Give it that starting point, and it doesn't hallucinate as much.

This is personal for me. I'm a high school dropout who didn't learn to type in ninth grade. When I first saw ChatGPT, I thought, "this is the fastest typist ever. How do I let it know everything about me so it can type the things that I wanna create?" That question is the brain.

Go deeper:

• What Is a Second Brain? for what's inside mine and how to start.
• The Personal Context Layer for what your AI needs to know before it writes a word.
• Why Identities Matter for teaching it who's talking.

Lesson 2: Build a Team With Clear Lanes

Once the brain exists, I hire. And I hire for roles.

My agent team starts with "me, the human in the loop." Then there's a chief operating officer and a project manager. Then specialists: a HubSpot expert, a social media expert, a developer, a designer, a video editor, a script writer, a copywriter, an editor and more. Each one has a lane.

Each hire is built for its role, too. "Whenever I hire an agent, I say, I want you to go do deep research on the hard skills and soft skills that the best designer that I could hire would have." Then it gets a personality, operating modes, a skills inventory and "the boundaries of what it needs to ask me for or not ask me for."

The lanes split by business, too. I have a team for Sidekick Strategies and a team for the Spiritual Side of Leadership. "Two different teams, two different organizations."

On a video, the lanes look like this: "let our writer write the script, let our post-production guy look at the edits that need to be made." Nobody does everything.

Go deeper:

• How I Hire an AI Agent for the job description that comes first.
• Agents, Skills and Grokbot for how I run two teams on one brain.

Lesson 3: Keep the Human in the Decision Seat

The team does the work. I keep the decisions.

The boards show how. Each team has a task board, and on the training I pulled them up. "Here's things that the team is waiting on me for, here's the pipeline of what is actually going through the pipeline." Then "I can literally just talk my way through the things that they need from me or the things that I want them to do next."

That's the seat. I don't need to know everything all the time. I need to see where everything is and make the calls that are mine. As I put it, "there's less communication, more action."

It goes back to the first thing I wish everybody understood: "AI is supposed to be here to augment and amplify us." Not replace us.

Go deeper:

• Go, Tweak or Hold for the three words that run my decisions.
• Stop Letting Your AI Agents Talk to Each Other for the boards and the one-voice rule.

Why the AI's List Sounded Like Mine

Here's the part that made me smile on the call. Echo's three lessons were the same three things I'd spent the afternoon teaching.

That's not a coincidence. Echo answers from my main brain. When your AI has a brain, it repeats you instead of guessing. I wrote about why that matters in Repeat, Don't Invent.

If you were teaching humans how to use AI to become augmented or empowered humans, what are the top three things you'd teach them?

Base your answer on what you know about me and how I work, not on generic advice. If you don't know enough about me to answer that way, tell me what you'd need to know first.

Keep each lesson to one line. Then, for each one, tell me where I'm already doing it and where I'm not.

If your AI can't answer that from what it knows about you, that's your answer: start with lesson one. The deeper conversation starters live in the AI Second Brain Workshop and my coaching.

Who This Is Not For

If you want one AI that does everything with no rules, this isn't for you. All three lessons point the other way.

And if you're hoping AI will make your decisions for you, it won't on my team. The decisions come back to the human.

Frequently Asked Questions

What are the three things your AI would teach about using AI? Give the AI a brain before you give it a job. Build a team with clear lanes, not one to do everything. Keep the human in the decision seat. Those are Echo's words, and I agreed with every one.

What does "give the AI a brain" mean? Give it a starting point of who you are, what your business is and what your clients want from you. Mine is a folder of plain markdown files I call the main brain.

Why does AI make things up? Because without a starting point about you, it uses everything it was trained on. Context about you helps it not hallucinate as much.

What's a clear lane for an AI agent? One role, like designer or editor, hired for the hard and soft skills of the best human in that job, with boundaries on what it asks you about.

How do I stay in the decision seat with AI agents? Keep a board that shows what your agents are waiting on you for, then talk your way through the calls that are yours.