How I Use AI (And Where I Never Will)

Why publish this at all?

Nobody asks their favorite author whether a typewriter wrote the book.

People are asking a version of that question about AI, and they should. The internet is filling with content nobody thought about. Words assembled by a model, prompted by someone who never lived them, published under a face that may not exist. You can feel it when you read it. Something answers, and nobody is home.

So the question I get, more than almost any other one: "Scott, do you use AI for your content?"

Yes. Openly, heavily, and with a hard line running through the middle of it.

Most creators answer that question with either denial or a shrug. I'd rather show you the whole thing. Partly because trust is becoming the most expensive asset in business, and I'd rather pay for it up front. Partly because I teach Entrepreneurs to build with AI, and it would be strange to hide how I do it myself.

The line: thinking versus distribution

Everything on this page comes down to one distinction.

Thinking is deciding what's true. What I believe about business, money, energy, scaling. Which advice fits which person. What a framework means and where it breaks. What I'd say to you if you sat across from me with your real numbers and your real fear.

Distribution is everything that happens after the thinking exists. Turning a talk into an article. Turning a story I've told on stage into an email. Making the thing I said to twelve people on a call readable for twelve thousand.

AI lives entirely on the second side. It has never crossed to the first, and it won't, because the first side is the business. People don't work with me because my paragraphs are tidy. They work with me because of fourteen years of pattern recognition across thousands of Entrepreneurs... and you can't prompt your way to that.

Here's the part most people find backwards: I tell my clients that content should be the last place they use AI, and operations the first. I stand by it. If AI writes your ideas before you've had any, you've automated the one thing that was supposed to be you. The reason I can use AI in content at all is that the ideas were already mine, already said, already on record.

Where everything starts

I talk for a living. Client calls, group sessions, workshops, podcasts, my own books... most weeks, hours of me explaining, arguing, coaching, and telling stories, all recorded, all transcribed.

That archive is the raw material of everything you read from me. Years of transcripts. Hundreds of thousands of words that came out of my mouth before any machine touched them.

When you read a post of mine about, say, why revenue plateaus around a million dollars... nobody generated that idea. I said it, somewhere, to someone, usually more than once. The article is the accessible version of a conversation that already happened.

I think of it like this: I've already written everything. I just said it out loud first.

I didn't start recording everything because I foresaw AI. I started because I kept noticing that the best explanations I ever gave happened live, under pressure, answering a real person's real situation... and then they were gone. The version of an idea I'd write at a desk was always thinner than the version I'd said on a call three weeks earlier. Recording started as a way to stop losing my own best work. It turned into the foundation of everything you're reading now.

How a story travels

Stories are where AI content usually goes to die. Models are happy to invent a struggling client, a rock-bottom moment, a tidy comeback. Fake specificity is the house style of the machine.

So stories carry the strictest rule in my system: every story is real, and it stays the way I told it.

When I lost my company in my twenties, when I built the next one from a negative bank balance, when a client turned $150K into $700K... those things happened, to real people, in real years. When one of those stories shows up in my content, it's pulled from a recorded telling. Names appear only with permission. Numbers stay exactly what they were. If I told a story three different ways over the years, we use the words on the tape, and where my memory drifted, we correct to the record, never away from it.

No story in my content has ever been invented by AI. That rule has no exceptions, and it's enforced by process, not by mood.

How my voice stays mine

A while back I learnt something building an AI version of my own mentorship: sounding like someone and thinking like them are two completely different problems. You can teach a machine my tone with a style guide. Teaching it my judgment took months of work, real people, and codified decision rules... and even then, it only reasons inside the lanes I built.

My writing system works off more than a hundred thousand words of my real, transcribed language. Not "write like a smart marketing guy." My actual sentences, my rhythms, the way I explain things when nobody's polishing me.

On top of that sits a banned list. Words I'd never use. Constructions that smell like a machine. Phrasings that sound like everyone else's AI content. Every piece gets checked against it, mechanically, before I ever see it. And then I see it. Nothing ships without my eyes on it, and plenty gets sent back or killed.

When I built an AI version of my own mentorship, the technical build took about two weeks. The whole project took five months and two full-time people. The gap between those numbers is the real story: almost all of the work was going through years of my own material and resolving the places where I contradicted myself. Which advice was the real position, and which was me talking through a specific person's specific situation? A machine can't decide that. I had to.

That project taught me what "AI in my voice" really costs, and it's why I don't take the phrase lightly when someone claims it after uploading three PDFs.

The test I hold everything to is simple: if a longtime client read it, would they hear me? Not "is this well-written." Would they hear me.

What this looks like on a Tuesday

Abstract principles are easy to claim, so here's the shape of one real piece of content.

I run a client session. Somewhere in hour two, answering a question about why a launch underperformed, I explain something about audience readiness I've explained fifty times... except this telling is sharper, because the question was sharper.

The session is recorded and transcribed, like everything. Later, the system surfaces that explanation as strong source material: this idea, said this way, in this telling. A draft gets built from it. The draft is checked against my voice rules and my banned list, mechanically. Any number in it is checked against my records. Any story in it is checked against the original telling.

Then it lands in front of me. I read it as the person whose name goes on it. Sometimes it ships. Sometimes I rewrite the opening because the emphasis drifted. Sometimes I kill it, because reading it made me realize I no longer fully agree with myself... which is its own kind of useful.

What never happens: nobody asks a machine "write a post about launches." There is no blank prompt anywhere in the chain. The blank prompt is where everyone starts sounding identical.

What AI does here, day to day

The honest list:

• Turns my spoken material into drafts. Talks become articles. Calls become emails. A workshop becomes a guide. The thinking is lifted from the recording. The drafting is the machine's job.
• Repurposes. One real idea, expressed once, becomes the versions different people need... the short version, the deep version, the version for the person who reads and the person who watches.
• Assembles research. When a piece needs data or sourcing, AI gathers and organizes. I decide what it means.
• Runs the checks. Voice rules, banned language, factual claims against my records. Machines are better at consistency than people... which includes checking the machine's own work.
• Runs operations. Follow-up, organization, routing, the repeatable layer of the business. That's a bigger story than this page, and it's where AI matters most by a wide margin.

And yes: this page itself was drafted by that system, from my recorded ideas, under my rules, and reviewed by me before it shipped. It would be a strange transparency page that hid how it was made.

What AI never does

• It never originates an idea. If I haven't said it or written it somewhere real, it doesn't get published as mine.
• It never invents a story, a client, a result, or a number. Ever. Proof either exists on record or it doesn't get used.
• It never takes a position for me. Every opinion in my content is one I hold, including the uncomfortable ones.
• It never plans my business. Offers, direction, pricing, who I serve... human decisions.
• It never advises my clients. People pay for my judgment. They get it from me. Where clients use AI tools we've built together, those tools run on documented frameworks, and everyone knows what they're talking to.
• It never fakes my presence. No AI avatars pretending to be live. No voice clone answering your DMs pretending it's me. If something is automated, it's allowed to look automated.

What it replaced (and what it didn't)

In 2019, my content team cost $45,000 a month. Writers, editors, managers, a machine of people turning what I said into what you read. That was the price of being consistently visible, and I paid it for years.

The people are gone. The function isn't. What that team did, the system does... faster, closer to my actual voice, and without the game of telephone that happens when four people separate your mouth from the page.

I won't pretend that shift is comfortable to talk about. It's the same shift arriving in most industries, whether we're honest about it or not. But notice what didn't get replaced: the thinking, the living, the fourteen years of receipts. AI compressed the cost of distribution to almost nothing. It compressed the value of judgment not at all... which is exactly why I protect the line the way I do.

Why this matters more every month

The cost of producing words is heading to zero. Which means words, on their own, are worth almost nothing.

You can already see the flood: AI-written content is losing visibility in search while content carrying real human experience regains it. The platforms are learning to smell it. So are readers. When anyone can generate ten thousand words of advice in a minute, the only question left is: whose words were lived first?

That's the bet this whole page sits on. In a world of infinite content, provenance becomes the product. Where an idea came from, who tested it, what it cost them to learn... that's the scarce thing. I'd rather own that ground openly than compete with the flood on volume.

Information went free. It's been heading that way for a decade, and AI finished the job. What people pay for now is what was never in the information to begin with: judgment about their specific situation, proof that a path works, and proximity to someone who has walked it. Every one of those gets more valuable as generated words get cheaper.

Which means the creators in real trouble are the ones whose entire value was assembling information... and the ones with the most to gain are people with a real body of lived work who've been too busy living it to distribute it. AI flips the economics for exactly the people who deserved better distribution all along. That's the version of this technology I'm building on, and it's the version I teach.

What I'm still figuring out

A transparency page that claims certainty about all of this would be lying by tone.

The line between thinking and distribution is clear at the extremes and blurry in one spot: when the system drafting from my material finds a sharper way to say something than I said it. Is a better sentence distribution, or is it thinking? My working rule: phrasing belongs to distribution, claims belong to me. If a draft ever says something truer than I said, it goes back to source... I either say it myself somewhere real, or it doesn't ship. That rule works. I also expect it to get tested harder every year as the tools improve.

Disclosure norms are unsettled too. I don't stamp "made with AI" on every email, the way nobody stamps "made with spellcheck." This page is my disclosure: the system behind everything, documented once, kept current. Whether the wider world settles on per-piece labels or per-creator pages, I don't know yet. I'll follow honesty over convention if they conflict.

And models will keep training on my public work whether I like it or not. Versions of my frameworks will show up in other people's AI content, unattributed and slightly wrong. The only durable answer I've found is the one this page describes: keep being the living source the copies degrade from.

How to test me, or anyone

Transparency pages are easy to write and easy to fake. Here's how you check one, mine included:

1. Ask for the origin. Pick any idea in a creator's content and ask where it first showed up. A real body of work has a trail: talks, older posts, books, interviews. Generated content has no past.
2. Look for cost. Real thinking has scars. Specific numbers, named failures, decisions that aged badly. If every story lands perfectly and costs nothing, someone's inventing.
3. Cross-examine the stories. The same real story told across years stays consistent in the facts and changes in the telling. Fake stories are consistent in neither, or suspiciously perfect in both.
4. Watch the positions. A human with skin in the game holds positions that cost them audience. A machine optimizes for agreement.
5. Ask them how they use AI. Not whether. How. Anyone doing this seriously can answer in detail, without flinching. Anyone hiding will give you a vibe instead of a system.

Run those five on me anytime. My material is fourteen years deep and public.

Here's what that looks like in practice, using my own work as the test subject. Take omnipresence, one of the ideas I'm most associated with. Ask for the origin: it shows up in my talks and writing years before it appeared in any book, born from watching my own ads follow people who weren't ready yet. Look for cost: I spent millions of my own dollars on ads learning the difference between being everywhere and being remembered, and I've published the number. Cross-examine the tellings: the idea appears across old posts, interviews, two books, and this site, consistent in substance, different in words each time, because a human retold it rather than a template reprinting it.

A generated body of work can't survive that walk. There's no trail, no cost, no drift between tellings. That's the whole test. Never whether someone uses AI... whether there was ever a person underneath it.

One more honest layer: yes, this page is also built to rank in search and be cited by AI engines. I want it found. The way I've tried to earn that is the old way... by making it the most complete answer I can give, so the machine citing it and the person reading it get the same true thing. If content ranks because it deserves to, the incentives finally point the right direction.

If you want to build your own version

The principles transfer, whatever tools you use:

• Record everything. Your calls, your talks, your explanations. Your future content is already in your mouth... you're just not capturing it.
• Never let the machine think first. Draft from your material, not from a blank prompt. The blank prompt is where everyone starts sounding identical.
• Write your own bans. Make a list of what you'd never say and make every draft answer to it.
• Keep one human gate. Nothing ships unseen. The day you stop reading your own content is the day it stops being yours.
• Tell people. Publish your version of this page. It costs you nothing except the option of pretending, which was never worth much anyway.

The machine is a printing press, and it's pointed at whatever you feed it. Feed it someone else's thinking and you become a reseller of noise... feed it your own record, and it becomes the widest door your real work has ever had.

FAQ

Was this page written by AI? It was drafted by my system, from my recorded ideas and my written rules, then reviewed and approved by me. That's the same process as everything else with my name on it, which is the point of the page.

Does AI write your emails and posts? It drafts them from my material. I review what ships. The ideas, stories, and positions are mine, on record, before any drafting happens.

Do you use AI images or avatars of yourself? Not to pretend I'm somewhere I'm not. I use AI in production the way anyone uses design tools. Nothing automated ever pretends to be live me.

How do I know a story you tell is real? Every story is pulled from a recorded telling and kept factually intact, with names used only by permission. If you ever want the origin of a specific story, ask me. There's always a tape.

Isn't using AI for content hypocritical if you say AI shouldn't think for you? That's exactly the distinction this page exists to make. AI distributing thinking that already happened is a printing press. AI replacing the thinking is a costume. I use the press. I don't wear the costume.

What AI does your company use? The tools change. The rules don't. Any tool that touches my content operates inside the system described here: my source material, my voice rules, my review.

Do you disclose AI on every individual piece? No, for the same reason nobody discloses spellcheck. This page is the disclosure, it covers everything with my name on it, and it stays updated. Per-piece labels say less than a documented system does.

Does AI touch your client work? Client-facing agents my team builds run on documented frameworks, and clients know exactly what they're working with. My personal judgment about a client's business is never delegated. That's what they're paying for.

Why should I believe any of this page? You shouldn't, on its own. Run the five-question test above against my public record. A transparency page is a claim. Fourteen years of consistent, dated, public material is the evidence.