I stopped using AI and started building with it

Quick note before we dive in.

You might notice things feel just a little different around here. A little more talk on AI, right?

When I started Fuel & Focus, I was writing for "ambitious builders" — a broad category that included anyone trying to ship something meaningful.

But here's what I've realized: the most valuable thing I can share is what I'm actually doing every day.

And that's building AI marketing systems.

After 15 years in marketing, I've spent the last year going deep on AI — not just playing with tools, but building actual systems that produce content that sounds human, not robotic.

So that's where we're headed.

Same energy. Same soul. More focus on what I know best: making AI work for marketers (and founders who have to be their own marketing team).

Let's get into it.

We're in the 1995 moment of AI partnership.

Let me explain.

In 1995, most people thought the internet was for nerds. Email was a novelty. Websites were curiosities. The people who saw what was coming — and started building — created the infrastructure that runs the world today. (after Bill Gates, Steve Jobs, etc.)

I still get a chuckle about David Letterman making fun of the Internet at this time.

We're in that moment again.

Right now, most people use AI like a fancy search engine. They ask ChatGPT questions. They generate images for fun. They maybe use Copilot to write emails faster.

That's like sending emails in 1995. Useful. But missing the point entirely.

The real shift isn't AI as oracle. It's AI as operator.

This weekend I set up an AI agent that:

This isn't ChatGPT with a fancy wrapper. This is infrastructure for human-AI partnership.

The framework I'm using is called OpenClaw. Open source. Runs on a $5/month server and token from an already existing subscription. It’s connected to Claude Opus 4.5 as the brain. (the most human-like LLM on the market).

And it took the Twitter/X and tech world by storm over the past week.

But here's what matters:

OpenClaw isn't the point. The paradigm is.

Bill Gurley tweeted this week about AI fluency becoming table stakes in the workforce. Alec Coughlin has been documenting how the people who can work with AI systems are pulling ahead while everyone else falls behind.

This isn't a prediction anymore. It's happening.

The question isn't "will AI change work?"

The question is: Are you building infrastructure, or just consuming output?

Most people will keep asking ChatGPT questions.

A few will build agents that do the work.

The gap between those two will define the next decade.

The AI Partner Stack

Here's a framework I've been developing. It's the difference between a chatbot and a Chief of Staff.

Most people stop at layer one. They give AI a prompt and hope for the best. Better people at least provide helpful context.

The leverage is in layers 3-5.

When your AI partner has memory, access, and autonomy, it stops being a tool and starts being a collaborator.

It catches things you missed. It follows up on tasks you forgot. It researches while you're in meetings. It drafts while you're at dinner with your kids.

That's not automation. That's partnership.

Here's the thing: none of this requires hard coding. The setup takes an afternoon. I wrote a full tutorial on exactly how to do it — every command, every file, every step.

If you want to build your own AI Chief of Staff this week, the guide is here.

If you want to hear more about my experience in the first 48 hours, that is here.

This was a big week. Let me show you what actually happened.

Last week I posted "How I Built My AI Chief of Staff" — the story of what happened.

This week I finished the companion piece: "How to Build Your Own AI Chief of Staff" — the step-by-step guide.

The goal: become a go-to guide for marketers and founders who want this leverage.

Bookmark it. Build yours this weekend.

I didn't just set up the agent. I put it to work.

I pointed Triton (my OpenClaw bot’s name) at a folder of documents and said: "Review everything here. Summarize what you find. Tell me what's missing."

Within an hour:

I didn't write anything. I just made decisions. Triton did the prep work.

That's the shift: I stopped being the researcher and became the decision-maker.

I built myself a dashboard this week. Dark mode. Glowing accents. Shows my tasks, ideas, pipeline, and goals.

Actually, I had Trident build it.

Small wins matter.

Next up: getting Triton connected to more tools — Gmail, Notion, image generation, calendar. The more access, the more leverage.

Here's what clicked:

I taught my AI partner how to think like me.

Not how to write. How to understand context, identify what matters, and draft outputs in my voice.

That's different.

Next up, I'm teaching it to follow up on things automatically.

The week after, maybe it handles scheduling.

This is what building infrastructure looks like — one capability at a time.

Worth your time this week:

📝 How I'd Build a One-Person Business in 2026 (Dan Koe) — Essential read. His thesis: info products are evolving into AI-assisted "learning experiences." The moat is no longer information — it's specific knowledge applied uniquely. "Build the thing only you would think to build." [LINK]

🔧 OpenClaw Starter Pack (@aiedge_) — If you want to explore AI agents, this is the definitive resource collection. 1,400+ likes for a reason. [LINK]

🔒 Bill D'Alessandro's Security Setup — Before you give an AI access to your life, read this. The "assume the server is already hacked" mental model is the right one. [LINK]

📖 My Tutorial: Build Your AI Chief of Staff in an Afternoon — The full guide. Every step. For people who want leverage, not just another tool. [LINK]

I'm not just using AI. I'm building with it.

I'll be documenting the shift in real-time.

And I'm showing others how to do the same.

That's the work now. And I’m happy to work with you.

If anything here resonated, hit reply. I read everything.

— Brian

P.S. If you've been thinking about setting up your own AI agent but haven't started, this week is the time. The tutorials are limitless. The infrastructure is ready.

The only question is whether you're building or watching.