A quarter-billion-dollar webinar seller grades AI task by task and finds one job it actually wins.
Posted
2 days ago
Duration
Format
Essay
comedic-rant
Views
1.4K
31 likes
57 · 43
Big Idea
The argument in one line.
Language models default to the safe, predictable answer, so they fail at strategy and big-picture content; their real value is narrow idea stimulation, mining existing material, and filling gaps where nothing exists.
Who This Is For
Read if. Skip if.
READ IF YOU ARE…
A marketer or founder who has built elaborate prompt chains or agent pipelines and suspects the output is not worth the engineering.
A creator deciding whether to let AI write your emails, reels, or book, and wanting a grade before you learn it from the reviews.
Someone whose webinar or funnel numbers dropped after switching to AI-written copy.
A business owner who keeps hearing about 100 agents working for you and wants a grounded case for what that noise actually costs.
SKIP IF…
You sell AI tools or services. The speaker says that is where the money is right now, and this video is not about how to sell them.
You want prompts, templates, or a tool walkthrough. There is one brief look at a prompt folder and no how-to.
TL;DR
The full version, fast.
Language models are pattern-matchers, so when asked for strategy they return the probable consensus move while real strategy lives in the unpredictable one; a Harvard Business Review study calls this trend slop. The useful move is to bring a direction you already know is right and have the model generate 50 variations, keep the one or two promising ones, and repeat. Big-picture content fails for the same reason, compounded by context rot, and the speaker grades his own team's results: email copy C-plus, Instagram reels D-minus, vibe-coded apps C-minus. Where the model wins is mining: summarizing 50 videos before you watch two, counting clips in a transcript to hold editors accountable, turning recorded walkthroughs into first-draft SOPs. Use it narrowly, resist letting it write for you, and swap in a human the moment you find one.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Too many shovels, not enough gold. AI is the most overhyped technology since sliced bread and promises productivity while making you less productive. Promise: a paradigm switch, no course pitch.
00:58 – 03:47
02 · AI cannot strategize
HBR study: ask LLMs for strategy and get trend slop. Strategy is context dependent, not content dependent. Models predict the probable; strategy is the unpredictable, initially unsound move.
03:47 – 06:15
03 · Stimulation, not strategy
Bring a direction that is already right and have the model generate 50 ways it could show up. Keep the 1 to 4 promising ones, refactor, repeat. Breakthroughs are accidents; the model cross-pollinates. A good human is still better.
06:15 – 10:03
04 · Macro content and context rot
A fitness author's AI-written book earned 1.9 stars. Long chats confuse old and new ideas and lose the thread. New chats, context monitors, and agent bucket brigades are space architecture: if it is losing context, you are asking too much.
10:03 – 14:33
05 · Micro: 47 prompts for one webinar
The speaker's webinar system: a 2,000-word prompt for a 3-minute pop quiz, 6,000 words for the challenges section, 47 prompts in total. A net win for C-minus clients, a slight edge for him, and every output is human checked.
14:33 – 15:03
06 · Mastermind pitch
In-person 12-hour mastermind in Los Angeles, five entrepreneurs, once a month, 10,000 dollars a seat, high six figures a year minimum. Info in the bio.
15:03 – 17:20
07 · AI as a maker, graded
Email copy C-plus. Instagram reels D-minus: fewer than 2 percent of a couple thousand weekly reels are AI scripted. Vibe-coded apps C-minus and fragile. Website design C-plus, useful only as throwaway MVP or vaporware.
17:20 – 23:05
08 · AI as a miner, where it wins
NotebookLM summaries at A-minus. Summarize 50 videos, watch two. Hypothesize before consuming. Transcript analysis sets a 40 to 45 clip quota for editors. Recorded walkthroughs become first-draft SOPs. Charts and technical call-outs for a book.
23:05 – 26:53
09 · Exponential bullshit
Incremental improvements get in the way of the 10x move. A risk-reversal guarantee let a bad funnel do millions. AI never finds 10x; it scales noise. No 55 agents, no 1,000 slop videos. Make hundreds of cheap bets instead.
26:53 – 30:26
10 · The incremental exception
During the 57 million dollar launch every email made money, so an email where none exists is valuable and C-plus ships. About a 3 percent gain on the current book. Resisting the siren song is a real cognitive cost, and a 2 percent edge everyone has is net zero.
30:26 – 31:35
11 · A hammer, not a screwdriver
AI is short of a tenth of a tenth of its promise. Right-sized, it is a hammer for specific nails. Mostly it is a placeholder until the right human is found, who is cheaper and better in almost every scenario.
Atomic Insights
Lines worth screenshotting.
Asked for strategy, a language model returns the consensus move from popular management books, because it predicts the probable answer and strategy lives in the improbable one.
Strategy is context dependent, not content dependent, and a model that only sees content cannot see the board.
Nothing in marketing is dead or alive; every tactic is valuable or not relative to whatever else is on the board right now.
Ask a model for 50 ways an idea could show up and expect 45 to be bad, 4 to be promising, and 1 worth refactoring into 50 more.
A good human still beats the model at idea stimulation; the model only beats having nobody.
A popular fitness author shipped an AI-written book and got a 1.9-star Amazon rating, an impulse that did not exist before the option did.
If you need new chats, context monitors, or agent handoffs to keep a task alive, the task is too big for the tool.
One webinar system runs 47 connected prompts whose combined length exceeds any webinar the author has written, for a slight edge over writing it by hand.
A team publishing a couple thousand Instagram reels a week scripts fewer than 2 percent of them with AI.
Vibe-coded apps grade around C-minus; they win only in markets with no solution, and only until a disciplined builder shows up.
Summarizing 50 videos before watching two flips the signal-to-noise ratio, which is the opposite of what most AI use does.
An AI pass over a recording transcript can set the expected clip count, so an editor who delivers 20 of 45 clips gets sent back for the rest.
AI will never find your 10x move; what it scales exponentially is noise.
During a 57 million dollar launch every email shipped made money, so a C-plus draft beats an empty slot, but only once you have exhausted your capacity to ship quality.
Expect roughly a 3 percent gain in quality and speed from AI research assistance on a book you can already write.
The dominant use of AI right now is as a placeholder until the right human is found.
Takeaway
AI mines well, makes badly, and never strategizes.
WHAT TO LEARN
Hand the model content to crunch and ideas to multiply, keep every context decision and every finished piece of writing human, and treat any gain as a placeholder until the right person shows up.
02AI cannot strategize
Asked for strategy, a language model returns the consensus move from popular management books, because it predicts the probable answer and strategy lives in the improbable one.
Strategy is context dependent, not content dependent: the same webinar tactic wins in one market and loses in another, and the model cannot see the board.
Nothing is dead or alive in marketing. Every tactic is valuable or not relative to what else is on the board right now.
03Stimulation, not strategy
Use the model to right-size a direction you already know is correct: ask for 50 ways the idea could show up, expect 45 to be bad, and refactor the handful that are promising.
Breakthrough strategies are usually exposed by accident, so the model's real job is cross-pollinating odd ideas you would never combine yourself.
A good human is still better than the model at idea stimulation; the model only beats having nobody.
04Macro content and context rot
A popular fitness author published an AI-written book and earned a 1.9-star Amazon rating; the option to outsource writing created an impulse that did not exist before.
The longer a chat runs, the worse the output: it confuses old ideas with new ones, re-proposes discarded ideas, and loses the thread.
If you are building new chats, context monitors, or agent handoffs to keep a task alive, the task itself is too big for the tool.
Over-prompting to remove em dashes costs more than paying a copy editor to remove them.
05Micro: 47 prompts for one webinar
The speaker's webinar system uses 47 connected prompts whose combined length exceeds any webinar he has written, to produce short snippets that still need editing.
That setup only pays off because clients who write C-minus openings get an A-minus draft; for someone who already writes well it is a lot of work for a slight edge.
Every output in the chain is sanity checked by a human, and clients who skipped that step now sell worse on webinars than they did before AI.
07AI as a maker, graded
Email copy from a model lands at a consistent C-plus no matter how much engineering goes into the prompt.
A team publishing a couple thousand Instagram reels a week scripts fewer than 2 percent with AI, because it cannot script reels cheaper than an expert.
Vibe-coded apps grade around C-minus; they can win in a market with no solution, but a disciplined builder will take that market the moment they arrive.
The best maker use is throwaway MVP websites, copy, and mock-up apps to demonstrate an idea, never to sell.
08AI as a miner, where it wins
Summarizing 50 videos on a topic before watching two of them flips the signal-to-noise ratio, which is the opposite of what most AI use does.
Write your hypotheses about a piece of content before consuming it, then compare; a summary in advance makes this retention method practical.
An AI pass over a recording's transcript can set the expected clip count, so an editor who delivers 20 clips out of 45 can be sent back for the rest.
Recorded process walkthroughs become first-draft SOPs automatically, but the model cannot invent the process; it only formalizes one that already exists.
09Exponential bullshit
Six simultaneous 10 percent improvements lose to one never-been-done move, like a risk-reversal guarantee that lets a weak funnel do millions.
AI will never find your 10x move; most humans cannot either, and what it scales exponentially is noise.
You do not need 55 agents, 1,000 generated videos, or 100-hour weeks on speculative tooling; make many cheap bets you can resolve quickly instead.
10The incremental exception
An email where none exists is worth money: during a 57 million dollar launch every email shipped made money, so a C-plus draft beat an empty slot.
That logic only applies once you have exhausted your capacity to ship quality and already have a working money printer.
Expect roughly a 3 percent gain in quality and speed when using AI as a research assistant on a book you can already write.
Resisting 'let me write that for you' is a real cognitive cost, like refusing junk food at 9 pm, and giving in erodes the narrow advantage.
11A hammer, not a screwdriver
AI is a hammer, not a screwdriver: useful for specific nails and short of a tenth of its promise everywhere else.
The dominant use right now is as a placeholder until the right human is found, because the right human is cheaper and better in almost every scenario.
Glossary
Terms worth knowing.
Trend slop
Generic, buzzword-heavy strategy advice that sounds sound because it parrots popular management books instead of addressing the specific situation. The term comes from a 2026 Harvard Business Review study of LLM strategic advice.
Context rot
The decline in a chatbot's output quality as a conversation grows long. Old and new ideas get confused, discarded suggestions reappear, and the model loses the thread of the task.
Space architecture
The speaker's name for building workarounds on top of workarounds, such as chat resets, context monitors, and agent chains, to force a model to produce an output a person could produce directly.
Bucket brigade
An agent pipeline where each AI agent does part of a job and hands the result to the next agent with a fresh context window, like firefighters passing buckets along a line.
Rube Goldberg machine
An absurdly complex chain of mechanisms built to perform a simple task. Used here for multi-agent systems assembled to write something piece by piece.
Vaporware
A mock-up or demo of a product that has not been built, used to show a concept or test interest before investing in the real thing.
Risk reversal
An offer structure that shifts the buyer's risk onto the seller, such as a guarantee that refunds the price and pays a penalty on top if the promised result does not happen.
Vibe-coded app
Software built by describing what you want to an AI coding tool and accepting what it produces, with little or no traditional engineering discipline.
NotebookLM
Google's research tool that ingests documents, videos, and audio you upload and produces summaries and answers grounded only in those sources.
Resources
Things they pointed at.
01:12linkResearchers Asked LLMs for Strategic Advice. They Got 'Trendslop' in Return (Harvard Business Review, March 2026)
06:25bookThe Aesthetic Revolution by Dr. Mike Israetel (1.9-star Amazon example)
08:51toolClaude Code (shows context window in real time)
“AI is a great shovel to sell. But when there are too many shovels being sold and not enough gold to be dug, then my friends, we have a problem.”
Self-contained metaphor, delivered by someone who admits he sells the shovels.→ TikTok hook↗ Tweet quote
00:22
“AI is promising you productivity out the ass while making you less productive.”
One line, one contrarian claim, profane enough to stop a scroll.→ IG reel cold open↗ Tweet quote
03:00
“Nothing is dead or alive. Everything is valuable or not valuable related to the other pieces that are on the board in the market.”
Kills the 'X is dead' genre in one sentence.→ newsletter pull-quote↗ Tweet quote
03:25
“Strategy is not about predictability. Strategy is about doing the thing that is unpredictable that the market doesn't know what it wants until you put it in front of it.”
The core argument for why LLMs cannot strategize, no setup needed.→ IG reel cold open↗ Tweet quote
07:20
“I've seen guys over-prompt their ass off just to get rid of an em dash. You know what's cheaper than that? Hiring a copy editor.”
Instantly recognizable pain for anyone using AI for writing.→ TikTok hook↗ Tweet quote
09:50
“If it's losing the context, you are asking it to do too much.”
Eight words that reframe every context-management hack.→ newsletter pull-quote↗ Tweet quote
18:20
“Most AI is doing the opposite. It's increasing signal this much, but it's increasing noise this much. And this is why you're becoming less productive with it.”
Hand gestures make this visual; the claim is the whole video in one beat.→ IG reel cold open↗ Tweet quote
25:05
“AI isn't going to find your 10x move. It never will. Most humans can't find 10x.”
Short, absolute, and contrarian to how agents are sold.→ TikTok hook↗ Tweet quote
25:54
“Agent is a dumb concept that humans come up with to give a human characteristic to a non-human entity. You don't need 55 agents.”
Direct shot at the current hype word.→ TikTok hook↗ Tweet quote
31:10
“What we're using AI for right now more than anything else is to put it in place until we find a human that can do it better. Then we plug out the AI, plug in the human.”
Clean closing thesis, works as a standalone take.→ newsletter pull-quote↗ Tweet quote
The Script
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
17px
metaphoranalogystory
AI is a great shovel to sell. Given my talents, I know how to sell it better than any other. But when there are too many shovels being sold and not enough gold to be dug, then my friends, we have a problem.
AI is that very problem. It is the most overhyped technology since sliced bread. And it's worse than the promise that social media made you, the promise to connect you more to your friends while leaving you feeling more disconnected than ever.
AI is promising you productivity out the ass while making you less productive. Bye, my friend. all is not lost there is tremendous amount of gold to be dug here with ai if you can switch your paradigm and i'm going to show you without selling you an ai course an app or an agent at the end of this video my name is jason fladlin and i advise the top ai companies in the space who pay me 5k an hour to help them better sell their ai i recently had a client go from 1 million a month to 2 million a month with just one shift in their ai approach because as good as ai is it can't replace strategy which is where we're going to start ai strategy is there was a harvard study that was done in the harvard business review where researchers they asked llms for strategic advice and they got trend slop in return so trend slop is essentially the tendency for ai to offer these generic buzzwordy sounding like sound strategy advice that parrots like popular management books like who moved my cheese and all that bullshit
instead of giving practical context specific solutions the ai cannot strategize and i will prove it to you i'll let you prove it to yourself take the best campaigns that you know of that have been run in history run them as simulations in ai and ask it for its advice and hypotheticals on which strategy it should take and it will always default to the safe normal strategy aka dog shit strategy.
Why AI cannot do strategy worth a shit is because strategy is context dependent. It is not content dependent. And this is something you've noticed with AI all over the board is it's really good in certain content situations.
Like if you give it a bunch of content and you have it analyze, it can do reasonably well, but it breaks context, which we'll talk more about here in a bit. Context with strategy is everything because I've worked in markets where the go -to strategy would be a webinar, but in other times the go -to strategy would be to not do a webinar.
I've been in markets where free webinars were the strategy and in other markets where paid webinars were the strategy. This is like this whole stupid idea of X is dead or Y is dead or Z is dead is dumb because nothing is dead or alive. Everything is valuable or not valuable.
related to the other pieces that are on the board in the market but see ai does not get that especially llm ai which is mostly what we're talking about here these are language models language models are based on pattern recognition and predictability and strategy is not about predictability strategy is about doing the thing that is unpredictable that the market doesn't know what it wants until you put in front of it so you have to understand this AI is going to default to safe and sound, but strategy is initially at least unsound.
It's that whole concept of everybody thinks you're a raving madman the day before the breakthrough occurs, and then they look back at you and call you a genius. So don't use AI for strategy. Instead, use it for stimulation.
So here's what I mean by this. i will know the strategy to perform generally for most of my clients but what we need is the difference between a meter and a millimeter is a little bit of word but a lot of distance and so we need to write fit strategy we need to take the thing that is generally and directionally correct and make it specifically and exactly precise and one of the ways that we do that is by asking it questions like here's my idea give me 50 different examples of how this idea can show up and then what will happen is i'll ask it for 50 45 of its inputs will be terrible four of them will be promising one will be really promising and then i'll ask it to refactor those and give me 50 -ish more And then I can repeat this until I get something that is interesting to me to pursue from a strategic standpoint.
Now, here's what's the greatest thing about strategy. Nobody gets this at all. Almost every great strategical initiative was exposed by accident.
Viagra, accident. They tried to help chest pains and they gave something else instead. Penicillin, accident.
Something was left out. Oops. Great insight occurs.
most almost all major multi -million dollar launches that occurred in my life from the companies that I've ran did not come about because we forced them into existence they came about because we were exposed to an opportunity we weren't aware of and then we recognized the signals when they occurred and so AI is really good at creating cool ideas to copulate with each other, to have the offspring of the offspring of the offspring of strategy.
And the hard thing to do on your own is to have that cross -pollination of weird ideas that normally don't mix together to give the accidental breakthrough. So AI can help you with that. Now I'm going to say something that's going to piss some people off, and it's this.
A real person still better at this so AI is better than nothing for idea stimulation for strategical stimulation for you to help right -size an idea but a good person is better than the AI so if AI isn't good at strategy is it good at content creation and the answer is it is bullshit at what we call macro content creation It is not good at big picture stuff.
There's this YouTuber, I really like him, Dr. Mike Isretel. And he's a big bodybuilding kind of dude and very popular, millions of subscribers.
And he launched a book recently on Amazon called The Aesthetic Revolution. And basically the whole thing was written with AI. And even the teaser copy to sell the book on Amazon is all the AI tells underneath of it.
And he's paying the price for it. 1 .9 average rating. on amazon one of the worst ratings that you could ever seen because he tried to get ai to write bunch of shit versus how I use AI to write books and he would have never done this before AI he would have never had the impulsiveness or whatever you want to call it to try to get it to do something it can't do because there was no option but now because there is an option for people to use a hairdryer in the shower they have to put a warning label don't use hairdryer in shower you will die and this is what I'm seeing a lot of otherwise really smart really productive really successful people People do AI really, really dumb.
So what do we do? So what we don't do is this. We don't try to kill the em dash.
So I've seen guys over prompt their ass off just to get rid of an em dash. You know what's cheaper than that? Hiring a copy editor to just get rid of your em dashes for you.
The problem with having it do large scale task AI is that the longer you chat, with it the worse the response so here's what it does it will very quickly confuse old ideas with new ones and that's a problem it will re -propose re -propose discarded ideas things you told it stop doing it will only do it like a little petulant child for a few minutes and then it will start doing it again and then it typically it loses its thread So what you have to do to deal with this is you have to have a strategy now that you create to overcome context rot.
And the way that this shows up is typically people will have to new chats often. So you're constantly creating new chats, which is what I do. And I don't do this because I'm trying to get it to write macro.
I'm going to give you a better solution here in a second. But just in general, I'm like a schizophrenic with my AI chats because I know very quickly it loses the plot. So you can frequently create new chats or you can use programs like Cloud Code, which will show you the context window in real time.
And you'll see things like, oh, okay, my context window is filling up. I should clear it out. Here's what's eating up my context so I know in the future what is context heavy and what is not context heavy.
But man, this is a lot of what we call space architecture. Space architecture is where you're building things to circumvent things to circumvent things to circumvent things to get an output that a human can do without having to hack it in order to get the outcome. The other way that you can do this, and this is, I think, catastrophic.
is routing agents all over the place so agent a takes the bucket of water he hands it over to agent b who has a fresh context she takes the bucket of water hands it over to agency and it's a bucket brigade how they used to put out fires they'd walk down to the ocean and then they'd one person at a time hand a bucket all over until you got to the building that was on fire and this is a strategy that can work but man it's a lot of architecture unnecessary building to solve a problem that doesn't need to be solved in the first place because if it's losing the context you are asking it to do too much so what do we do instead we focus on the micro I don't have it write a full entire book for me.
And I don't create a Rune Goldberg machine to try to get it to assemble a book for me by piece and agent, by piece and agent, by piece and agent. A Rune Goldberg machine, by the way, is like that mousetrap game back in the day where this thing hits that thing, it triggers this thing, it drops this thing, and then the mouse gets caught kind of bullshit.
We don't want to do that. Instead, we want to do micro. So I'll give you an example.
This is one way that I've designed. ai to help assist me with writing webinars so i have a very specific kind of webinar that i use where we start with a pop quiz pop quiz is about two to three minutes of run time in the webinar it's about 2 000 words of prompting for me to get it to give me a decent enough version one of the pop quiz and then after that i go into the challenges you face then i run through 10 to 12 to 15 challenges so i have actually two prompts that do this.
First is a buyer profile prompt that I have run pre -webinar creation. And then I will have it analyze the market and then follow another prompt to create the output for the challenges. And these two prompts are weighing in at about 6 ,000 words in total.
So I've now created 8 ,000 words of prompting to write less than 8 ,000 words a webinar script like this is insane and the bigger problem here is I can do a pop quiz faster with my brain than an AI can do it and better than an AI can do it I can do challenges you face significantly better than an AI but because my clients who are D minus or C minus at these openings can't off the top of their head rip off an A plus or an A minus script this is a net win for them and this is why i develop these to help me help my clients but it's a lot of work for a microscopic movement forward in the webinar and then we do the same thing like we have the excitement stack by the end of this webinar you're going to discover this this and this and you'll be able to do this this and this and this and this and this will no longer be a problem and then we have the why listen to me section And then, you know, it's the whole like, here's how I discovered this.
Here's where I was at. Here's where I'm at now. Here's why I'm uniquely qualified to help you.
Here's why this specific thing is the thing I'm teaching you over every other thing that I could teach you, et cetera, et cetera. And then we go on. And at the end of the day, I have about 47 different prompts that are connected together that are longer in totality of words inside of those prompts than any webinar I've ever written word for word.
So we are doing a lot of effort to get the AI to produce small snippets of copy at a high degree of value with little bit of effort for editing. And you know, we're getting a slight edge over a human doing it. Now I'll take slight edges and they make sense when you multiply them.
So if I'm selling you an AI tool that does this, I'm incentivized to write these over -engineered prompts to sell it to you. But if I'm using this myself and I'm not an AI business or an AI agency or I'm not trying to sell an AI shovel, this shit doesn't really make sense.
It only makes sense if you're solving for one major problem and you can build a lot of little AIs out there to help assist you. And AI is with humans in the loop because it's either me or somebody on my staff that is sanity checking every AI along the way. So is it helpful?
Yeah. Is it really helpful if I'm selling a shovel to you? Sure.
That's where the real money is. Now, is it helpful beyond and above what we could do already? No.
In fact, most of my clients now are worse at webinars because they're using AI slop that looks good but isn't good. And then they're not selling on these webinars as well as they used to. and they're coming to me to solve their problems now so in that way ai has been good for me personally because i've seen people that were really good get less good pay me to help them get back to where they were and then beyond where they were but if you're going to use ai to create content have it create micro not macro hey real quick we do these in person five at a time masterminds i do one full day for 12 hours training five entrepreneurs in person here in los angeles once a month 10 000 a person if you are interested and you make a decent amount of money at least high six figures a year information is in the bio if you want to attend one of these in -person seminars that i run but even then it's problematic because ai as a maker is bullshit it cannot make content very well to save its life so i'm going to give you a couple use cases email copy
it is not very good no matter how much you prompt no matter how much you engineer it no matter how you try to get it to do things that it just isn't very good at and we've had it write a lot of copy and the only consistent copy that we can produce is average that at about a c plus rating it's not world class it's not terrible it's decent -ish and you know we publish a couple thousand instagram reels a week believe it or not because you know we're insane like that guess how many of them are scripted with ai less than two percent because we have found that it doesn't know how to script ig reels no matter how much you try to train it or teach it at a degree where it's cheaper than hiring an expert to do it so it's dog shed at instagram reels now what about vibe coded apps Garbage.
If you work really hard at it, these vibe -coded apps are about a C -. Now, in some markets, with no solution, a C - solution can make you millions of dollars, right? So this is...
dying in the desert, no water, somebody goes and gives you tap water out of a hose, and you think it's the greatest thing alive. So in these instances, vibe -coded apps can work, but that's very fragile. It's very short -lived.
This is not an advantage that is going to last you very long, and I've seen this too many times to count, easy come, easy go. The moment somebody with some discipline who builds a proper app comes in, they will crush you in the marketplace because you've been addicted to junk food apps, and this is not a... viable long -term solution.
And then, you know, website design. This one is actually the thing that we have found to be the most useful for it. So if we can throw up like an MVP website and we can throw up some MVP copy, whether it's email or website copy and an MVP app, not to sell, but merely for demonstrative purposes or what we'd call vaporware to demonstrate a concept of an idea that hasn't been developed yet.
mock -up style then this can be useful but it is not maker content and that's what sucks now what can it do instead it's really good at mining content this is where it wins so it can take a lot of data and synthesize it and crunch it for you so for example i use notebook llm probably more than any other ai out there i know that sounds insane because it can summarize youtube videos for me at about an a minus it's not perfect it would be better if i did it but it's hard for me to hire a person that can summarize it as good as Notebook LLM can.
So instead of me having to go out there and trying to personally consume a lot of content to figure out what I should follow and what I shouldn't, what is good and what isn't good, the AI gives kind of the round one for me. So it can run and analyze 50 videos on a topic. I can see the summary of those videos.
From those videos, I can determine all the key ideas and points ahead of consuming any one of those videos. And then I only consume a couple of those videos. And because the noise to signal ratio has been switched, I've reduced noise and I've increased signal.
Most AI is doing the opposite. It's increasing signal this much, but it's increasing noise this much. And this is why you're becoming less productive with it.
But because I can go in, I can have it analyze podcasts, I can analyze YouTube videos, I can get core ideas, and I can hypothesize before I consume the concept. Because the most effective way you can learn anything is before you study it, You create an idea around what will be covered in it, and you create your own conclusions in advance.
I bet this will cover this. I bet this will be the insight. I bet this will be the aha moment.
I bet it will advise me against this. And then as you consume it, you see if you were right or wrong. And then when you're done, you compare your hypotheses, what actually happened.
And this is hard to do if you don't have a summary in advance, but it's pretty easy to do now. So my retention and my understanding of ideas has increased. And my ability to cover more on a specific topic has increased.
The effectiveness has went up. But this is research. Nobody gets their panties all wet because I'm a better researcher now.
That's not going to sell clicks and sell courses, but that's where a lot of the value is. Another way that we use this is we analyze what we call car videos. So I get over a million views a week driving in my car.
Actually, my wife does all the driving, so I'm the talent, right? So she will ask me a question. We have two DJIs posted up on suction cups against the window.
I answer it hot take fashion. And we do this a couple times a week anytime we're really driving anywhere. And it produces a lot of content for us, hundreds of videos.
And from those hundreds of videos, we get millions of views. Now, the challenge with this is how many clips should we get from each session in the car? if we hand these over to editors to do that they tend to take the easiest clips because you know this is how people work and they clip it and ship it and then they throw away a lot of good content just because they don't want to do the extra work so how do we hold them accountable so we run it through an ai system to analyze the transcripts and the ai system has been programmed to tell us you should get between 40 and 45 clips from this session So then when we go to the editors and the editors produce only 20 clips for us, we say, bro, you're missing half of the clips.
Go back there and do the rest. Otherwise, I'm not going to assign you any more content. So this is how we're able to get a workforce of editors to crunch lots of content for us with AI.
But again, as a supplemental tool to help you with something that is already a high skill threshold. nobody's going to get excited about that but i'm finding that's where a lot of the value of ai is so it's supporting the content it is not creating the content so first draft sops are another great place where ai has helped us tremendously Because I can walk through a process on a training call with a client or with a staff member, and then it will take that and summarize this.
This is a Gemini function, essentially, when we upload it in Google Drive, and it will give me the first draft of the SOP, and then I can hand it over to somebody to finalize it. So we are now able to create very quick SOPs in a fashion that is more effective than any other approach we've ever tried in history. But that means you have to have the process.
The AI is not going to create the process for you. The AI sucks at creating process. But the AI is really good at pulling out existing process and turning it into a formalized document.
And then here's actually lately the thing I've been doing most because I've been writing this book is I've been having it create for me charts, summaries, technical explanations. And by the way... For me, again, Google wins here.
The AI summaries when you go to Google and ask it like rapid fire questions when you're writing your book. So I'll be writing my book and I'll be like, okay, cool. Give me a chart that will summarize this concept.
And I use ChatGPT for that. But a lot of times it's like, what's the technical explanation of this? And then you can have that little like call out box that you would have in a book.
And it does really good for that. Or if I feed the manuscript of each chapter into an AI tool. I'll say summarize the key points here, and it will.
And then sometimes if I want to give actual step -by -step, that is, again, technical, first do this, then do that, it can take what I've written and turn it into a technical explanation. So it is a good research assistant, which is awesome if you're doing amazing things and you need somebody to support you with the research, but it doesn't know on its own.
what is good and what isn't good when it comes to creating content and last exponential it is bullshit it sucks at being exponential which is funny because this is how it's totally sold it's this tool that's going to exponentialize your business because you could have 100 ai agents working for you all simultaneously but this is where we increase noise way much more than we increase signal so in general You should never try to increment anything in your business.
It's poor strategy because all the incremental improvements, they get in the way of the exponential result. If you try to chase 10 rabbits at once, you go hungry. If you focus only on one rabbit, your belly gets full kind of a deal, right?
If you try six different ways to grow your business by 10 % simultaneously, even if you get all six ways, it's going to be less effective one incredibly new way to 10x your business because you will come up with an idea that's never been done before That if you get that one idea it will change everything and I'll give you examples of this So so many people are trying to like improve the sales copy on their funnels, but if you come up with a risk reversal Proposition like I've done in the past better your money back guarantees, you know 10x guarantees will buy your business and pay you $25 ,000 on top of that guarantee then the funnel can still be dogshit You can still be terrible at organic search and all paid advertising and all that and do millions of dollars like I've done because the exponential risk reversal Was a thousand times more effective than any singular incremental move ever could have been so we are allergic to incremental as really strong strategists in making money.
But there is one exception here. And the exception is this. AI isn't going to find your 10x move.
It never will. Most humans can't find 10x. And by the way, you need...
10x you need to exponentialize because when the tides come in because there's always recessions there's always contractions there's always black swan events 10 20 30 40 all of those get wiped out they all die because incremental dies but if a business is exponential it can withstand significant changes in the current so you need 10x but the only thing that AI does really well exponentially is creating exponential noise, which decreases productivity.
So you don't need 55 agents. You don't even need... one agent.
Agent is a dumb concept that humans come up with to give a human characteristic to a non -human entity, which is AI, right? You don't need 55 agents. You don't need a thousand slop videos.
If they have any advantage, it's going to be short -term. The algorithms are going to figure out how to get rid of those pretty soon anyway. And you don't need 100 -hour weeks of investing in speculative AI technology because predicting the future is expensive.
We do not want to predict the future at all. We want to explore different innovations in a low -risk, low -cost fashion, make hundreds of bets on speculative ideas that don't hurt us, that we can find the answer to quickly. And if only one of those pay off, we make a thousand times more than what the market typically produces.
And your AI space architecture isn't going to get you there. Now, here's what will help you. get there and it's incremental but there are a lot of caveats and i don't think a lot of you are going to be able to comply with these caveats and so therefore it's better to just use it to write the kids pe teacher you know excuse my son for being late no as opposed to having it try to perform surgery on your business kind of a deal so an email when none exists then it's very valuable.
And I'll tell you a little story here. When we were doing the plan launch, which was the $57 million launch, we had about 12 copywriters that we were having write emails full -time for us. And we had 300 pages of emails because I was writing different emails for every scenario or having direction for the team to write emails in every scenario.
And every time we wrote and shipped an email, we made money. and if we didn't have an email we didn't make more money and so having any email where none exists we're good enough is sufficient and c plus wins then we ship but that's because we've exhausted our ability to ship quality we are at a human scarcity or a time scarcity function and we have a money printer and only when we have a money printer do we just add more to it at the end because a one percent increase on a 57 million dollar launch is still a lot of moolah baby so there are instances where just having anything in place is better than having nothing at all so when i'm writing my book ai is a better researcher i think i will attribute about a three percent increase in the quality of the content and the speed of productivity of me writing this latest book that i have which is use brain in stores soon hopefully right but it's not a game changer in and of itself it won't substitute a poor written book with research it will be a more efficient way to publish a book that isn't good quicker if i didn't have the skill and the experience and the knowledge and the strategy to write the book on my own now try to resist the urge
to have it right for you it is incredibly hard to do it's very cognitively demanding like if you study habit formation you discover that one of the hardest parts about a habit is the cognition required to force the habit into existence every time you have to make a no decision where you want to make a yes decision that's a cognitive cost so if you're in your ai and you haven't habituated it yet every single time it's a siren song saying me write that for you let me write that for you and you initially have to resist that which is going to deplete your glucose in your brain and your energy in your brain and you're going to be tired and you're going to want to give in like eating that junk food at 9 pm at night after going the whole day with a crystal clear clean pristine diet so you have to resist the urge and that's very difficult and this is the case with a lot of the incremental improvements so the caveat here is incremental instead As long as you use it in narrow scope, as long as you don't give in to the urge, right?
So you've got to stay away from the siren song. And you've got to know that this incremental advantage can vaporize in a second. Because if everybody can increase productivity by 2 % relatively easily, then that is a net zero gain because you don't do business in a vacuum.
So AI, it's mostly bullshit. There is a few instances where you can have leverage. And if you are selling AI, that is the best place to be right now, albeit temporarily.
But as a tool that has promised the moon and the sun and the stars, as of now, it is woefully, woefully. Short of even a tenth of a tenth of a tenth of its promise, but if we right -size it we recognize it that it's a hammer It's not a screwdriver It's a hammer and it can be used to nail in certain nails and this is a tool in our toolkit Let's go for it But oftentimes what we're using AI for right now more than anything else is To put it in place until we find a human that can do it better and then we plug out the AI Plug in the human because in almost every scenario It is cheaper and more effective at the end of the day with the proper calculus to have the right human do it for you as opposed to have the right AI.
Thoughts?
The Hook
The bait, then the rug-pull.
The speaker opens by admitting he is better than almost anyone at selling AI, then spends 31 minutes explaining why he would not buy most of it. The title is the thesis, and the parenthetical is where the useful part lives.
Frameworks
Named ideas worth stealing.
02:00concept
Context, not content
Strategy depends on the context of the market, what else is on the board right now. Language models are content machines built on pattern recognition, so they default to the safe, probable answer, and strategy is by definition the unpredictable one.
Steal forAny argument about what to delegate to AI: delegate content tasks, keep context decisions human.
04:10list
The 50-ideas stimulation loop
Bring a direction you already know is right
Ask for 50 examples of how it could show up
Expect about 45 bad, 4 promising, 1 really promising
Refactor the promising ones into 50 more
Repeat until something is worth pursuing
A way to use a model to move from directionally correct to exactly precise. The model's job is volume and cross-pollination of odd ideas, not judgment.
Never ask the model for the whole book, webinar, or app. Ask for small snippets at high value, with a human sanity-checking every piece. If the task needs context workarounds, the task is too big.
Steal forScoping any AI writing or coding task.
10:29list
Webinar prompt chain
Buyer profile (pre-webinar)
Pop quiz
Challenges you face
Bone to pick copy
Excitement stack
On this webinar
Why listen to me
Paradigm shift
Criteria 1..N (why, what, how, next)
About 47 connected prompts, each producing one short section of a webinar script. Roughly 8,000 words of prompting to produce fewer than 8,000 words of script. Worth it for clients who cannot write an A-minus opening themselves.
Steal forModular script generation where each section has a fixed job.
15:00model
Maker versus Miner
Maker: email copy C-plus
Maker: Instagram reels D-minus
Maker: vibe-coded apps C-minus
Maker: website design C-plus (MVP only)
Miner: NotebookLM summaries A-minus
Miner: clip counting for editor accountability
Miner: first-draft SOPs from recordings
Miner: charts and technical summaries for a book
AI grades poorly whenever it has to create content and well whenever it has to crunch existing content into something smaller. Use it to support content, not to create it.
Steal forDeciding which parts of a content workflow to hand to AI.
18:46list
Hypothesize before you consume
Get a summary of the content in advance
Write what you bet it will cover and conclude
Consume it
Compare your hypotheses to what actually happened
A retention method that was hard before summaries were cheap. Raises understanding and lets you cover more on a topic.
Steal forAny research or learning workflow.
23:32concept
Incremental versus exponential
Six 10 percent improvements lose to one never-been-done move, and incremental gains get in the way of finding it. Exponential businesses survive contractions; incremental ones get wiped out. AI only scales noise, so it never supplies the 10x.
Steal forPrioritization. Spend on many cheap speculative bets, not on polishing funnels.
27:43concept
C-plus wins when nothing exists
Once you have exhausted your ability to ship quality and you already have a working revenue engine, any email beats no email. A 1 percent lift on a 57 million dollar launch is real money. This is the one place incremental AI output is worth taking.
Steal forDeciding when good-enough AI copy is acceptable.
CTA Breakdown
How they asked for the click.
VERBAL ASK
14:33product
“Hey real quick, we do these in person, five at a time masterminds. I do one full day for 12 hours training five entrepreneurs in person here in Los Angeles once a month, 10,000 a person. If you are interested and you make a decent amount of money, at least high six figures a year, information is in the bio.”
A 30-second mid-roll at the 14:33 mark, right after the webinar-prompt section where he has just demonstrated expertise. Hard qualifier (high six figures) filters the ask. Honors the opening promise of no AI course, app, or agent pitch. Mastermind b-roll plays under it.
Add Modern Creator as a preferred source and Google shows you more of our breakdowns in Search, Top Stories, and AI Overviews. It only changes what you see, and you can undo it in your Google settings anytime.
Add to Preferred SourcesOpens your Google source preferences with us pre-loaded. Tick the box and you're done.
Jason Fladlien reads nine word-for-word sales closes, breaks down the psychology in each line, and argues that fluency comes from memorizing them, not improvising in the moment.
A single continuous talking-head lecture, intercut with handwritten notepad cutaways, breaking offer design into five levers — Time, Effort, Routine, Money, Status — each with its own named formula, from a designer of 26 seven-figure offers.
00:30
00:49
01:46
02:01
02:33
03:04
03:15
03:40
04:21
04:56
05:19
05:50
06:08
06:43
07:25
07:33
08:08
08:31
08:45
09:40
09:59
10:32
10:40
11:23
11:38
12:07
12:40
13:13
13:37
14:00
14:39
15:24
15:49
16:23
16:37
17:16
17:22
17:57
18:21
18:45
19:08
19:56
20:19
20:43
21:08
21:30
22:12
22:42
23:07
23:46
24:40
25:00
25:19
25:42
26:04
26:31
26:53
27:14
28:21
28:32
29:01
29:24
29:47
30:20
30:35
30:59
14:47
32:31
38:00
Video of the Day30:08
Video of the Day53:07
21:16