38:00Jason Fladlien · Tutorial
The $50K/day Skill AI Can NEVER Replace
A play-by-play teardown of one closing keynote: the exact cadence, soundbites, and engineered spontaneity behind a $50K-a-day speaker's stage act.
September 6thThe argument in one line.
Most business tactics fail to move the needle for the person trying them, and winners are wildly overrepresented online because they're the only ones who talk about it, so the real skill is scoring any tactic on risk, time-to-feedback, and effort so you can cheaply test more of them and keep whichever ones happen to work for you.
Read if. Skip if.
READ IF YOU ARE…
- You run paid ads, funnels, or content and keep chasing whatever tactic the loudest person on social media just swore made them rich.
- You're deciding whether to build an AI feature, agent, or product and want a gut check on how much real risk you're taking on.
- You want a repeatable way to size up a new business tactic in minutes instead of after months of sunk effort.
SKIP IF…
- You're looking for step-by-step tactical execution, ad copy, funnel steps, specific AI prompts, this is a decision framework, not a how-to.
- You already have a formal way to score opportunity against risk and don't need another mental model for it.
The full version, fast.
When a thousand people try the same new business tactic, most see nothing happen, a small number get hurt, and a small number win big, but only the winners post about it, so social media makes 'proven' methods look far more reliable than they are. The framework taught here scores any tactic on three axes: risk (how many things can go wrong, and how bad), time (how long until you get real feedback), and effort (how much work, and whether the skill transfers elsewhere). Presell before you build, keep AI scoped narrow instead of dumping everything into one big prompt, and prioritize skills like objection handling that work in any market. The conclusion: stop hunting for the one proven protocol, run more small, cheap, fast tests, and keep whatever happens to work for you specifically.
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.
Create a free account →Where the time goes.
00:00 – 01:17
01 · Cold open: the Mars-Jupiter Protocol
A fictional study on a thousand entrepreneurs testing 'the hottest AI money making model' sets up the video's real subject: how outcome distributions actually look when a group tries the same tactic.
01:17 – 03:05
02 · The skewed distribution of outcomes
The made-up numbers: roughly 10% take a real loss, 40% take a small loss, and only a minority ever see any kind of win. Stated plainly: 90% no effect, 5% harmed, 5% worked wonders.
03:05 – 05:45
03 · Why winners get overrepresented
The people it worked for brag about it everywhere, the people it didn't help stay invisible, and the people it hurt stay quiet out of shame, so the public conversation skews wildly toward success.
05:45 – 08:27
04 · Quadrant thinking: the meta-skill
Reframes the goal from 'is this tactic proven' to minimizing risk, time, and effort needed to find out if it works for you. Introduces the three-factor quadrant framework.
08:27 – 12:05
05 · Time: the presell beats the pre-revenue grind
Compares a pre-revenue AI startup (long, delayed feedback) to a lean app to a book you can re-engineer for instant feedback, then lands on his favorite: preselling to the narrowest hot market segment, sized by the '1,000 / $1,000 / 10,000' rule.
12:05 – 16:42
06 · Effort: skills that transfer vs. skills that don't
Raising funds is high effort but globally useful; a custom dashboard is high effort for a narrow win a spreadsheet solves; hot-take content is low effort but its skills don't transfer; memorizing sales closes is low effort and globally useful, the best of all four.
16:42 – 21:08
07 · Risk: what AI should and shouldn't touch
Warns against AI over large datasets (more data increases noise, decreases signal, and invites being commoditized by the next model update), contrasts the farmer whose AI crop advice backfired with low-stakes uses like a routine email, then explains his own fix: splitting a webinar-writing task into 32 narrow, specifically-trained agents.
21:08 – 25:17
08 · Momentum, habituation, and reframing failure
Returns to the outcome distribution, then argues the real lever is momentum: stacking small wins into habituated, automated routines, and reframing 'I suck' as 'this model wasn't for me.'
Lines worth screenshotting.
- When a thousand people try the same business tactic, the most common outcome by far is nothing happening at all, not success or failure.
- People who get big wins from a tactic are the only ones loud enough to post about it, so social media systematically overrepresents its success rate.
- A tactic that 'didn't work' for 90% of people and 'worked wonders' for 5% is still worth testing, because the goal is being in that 5%, not proving the tactic itself works.
- The real skill isn't finding the protocol proven to work, it's expanding your 'surface area of luck' by testing more cheap, fast, low-risk protocols.
- Score any new business tactic on three axes: risk (many/few risks, minor/major severity), time (short/long timeframe, instant/delayed feedback), and effort (high/low effort, local/global applicability).
- Presell the offer before you build it, in beta, to the narrowest and hottest slice of your market, so you get real feedback before you've sunk the cost.
- A market of just 10,000 people is enough to build a $1 million business: get 1,000 of them to pay $1,000 each.
- A skill like writing sales closes and handling objections transfers to every market that has that objection, which is why it beats almost any single tactic.
- Building a custom reporting dashboard is high effort for a result that only works in one narrow context, a spreadsheet does the same job for a fraction of the cost.
- Feeding an AI system more and more data to make it 'smarter' actually increases noise and decreases signal, making the output less reliable, not more.
- If a major AI provider can plausibly ship your feature as a native update next quarter, you're taking on major risk to build something with a short shelf life.
- Breaking one large AI writing task into narrow, single-purpose agents, each trained on a small slice of the job, produces fewer errors than asking one model to do the whole thing.
- The lowest-risk AI use isn't the flashy stuff, it's small, boring tasks where being wrong barely matters, like drafting a routine email.
- Reframe a failed experiment: 'this model wasn't for me' replaces 'I suck,' and 'I only need to win occasionally' replaces 'why do I always lose.'
How To Judge Any Business Tactic
RISK, TIME, EFFORT
Most 'proven' business tactics only work for a small slice of the people who try them, so the real skill is scoring risk, time, and effort to test more of them faster.
01Cold open: the Mars-Jupiter Protocol
- When a group of people all try the same business tactic, the distribution of outcomes is rarely all-or-nothing, most people land somewhere between a small loss and no change at all.
- A fictional 'study' can still teach a real lesson: treat any claim that a tactic 'works' or 'doesn't work' as a description of a distribution, not a single verdict.
02The skewed distribution of outcomes
- In a group of 1,000 people trying the same tactic, roughly 10% take a real loss, 40% take a small loss, and only a minority ever see a win of any size.
- A tactic that produced 90% no-effect, 5% harm, and 5% big wins is simultaneously true that 'it didn't work' and true that it 'worked wonders,' depending only on which slice of people you look at.
03Why winners get overrepresented
- The people a tactic worked for are the ones loud enough to post results on TikTok, Instagram, Facebook, and LinkedIn, so they're wildly overrepresented in what you see online.
- The people it didn't help stay quiet because there's no story to tell, and the people it hurt stay quiet too, often blaming themselves rather than the tactic.
- Public arguing between a tactic's defenders and detractors is itself a growth engine: the algorithm spreads the fight, and undecided viewers get forced to pick a side.
04Quadrant thinking: the meta-skill
- Stop asking whether a tactic 'works,' the better question is the least amount of risk, time, and effort needed to find out if it works for you specifically.
- Score any protocol on three factors: risk (how many things can go wrong and how bad), time (how long until you get feedback), and effort (how much work, and whether the resulting skill transfers).
- The goal isn't to be right about which tactic to pick, it's to test more tactics faster so you land on one that happens to work for you.
05Time: the presell beats the pre-revenue grind
- A pre-revenue AI startup takes a long time and gives delayed feedback, which is exactly why it's high-risk, and why investors prefer spreading bets across many startups instead of being any single one.
- Long-feedback projects like writing a book can be re-engineered for instant feedback by publishing pieces as you go and watching how people react.
- Presell the offer before it's built, in a rough beta, to the narrowest and hottest slice of the market you can find, that's the fastest real feedback loop there is.
- A market of only 10,000 people is big enough to build a $1 million business: sell to 1,000 of them at $1,000 each, small enough that you can personally reach everyone in it within a year.
06Effort: skills that transfer vs. skills that don't
- Raising investor funds is high effort, but the skill is globally useful, it transfers to any business you ever start.
- Building a custom reporting dashboard is high effort for a narrow, one-off use, a spreadsheet solves the same problem for a fraction of the cost.
- Hot-take short-form content can rack up millions of views for very little effort, but the skills involved don't transfer, what hooks people in a 15-second clip doesn't work in long-form or in an actual sales conversation.
- Memorizing sales closes and objection handling is low effort and globally useful, it applies in any market that has a money or time objection, making it one of the highest-leverage skills to learn.
07Risk: what AI should and shouldn't touch
- Feeding an AI model more and more data to make it more capable actually increases noise and decreases signal, making it more fragile, not more robust.
- If your AI feature depends on crunching a large dataset, assume a major model provider will ship the same capability as a native update within a few releases, so you're building on borrowed time.
- Some AI mistakes are low-risk even when they happen, a slightly-off routine email costs you almost nothing.
- Instead of one AI prompt trying to write an entire webinar, split the job into narrow, single-purpose agents, each responsible for a small piece and trained on a small dataset, fewer risks per agent means fewer overall mistakes.
08Momentum, habituation, and reframing failure
- The people who land on a winning tactic build momentum, and momentum makes it easier to keep making the right adjustments to get further results.
- Momentum can be turned into habituation, a routine you no longer need motivation to execute, which is what makes a result repeatable instead of a one-off.
- Reframe a failed attempt as 'this model wasn't for me' instead of 'I suck,' it removes the shame that keeps people from trying the next tactic.
- There's more luck in most people's success than they're willing to admit, the lever you actually control is how many low-cost attempts you make, not which one is 'correct.'
Terms worth knowing.
- Mars-Jupiter Protocol
- A made-up thought experiment used to illustrate how a group of people trying the same business tactic actually performs: most see no change, a few are hurt, and a few see outsized wins.
- Quadrant thinking
- A way of evaluating any tactic or opportunity by placing it on two-axis grids, risk, time, and effort, instead of judging it as simply 'works' or 'doesn't work.'
- Surface area of luck
- The idea that success in unpredictable ventures comes from running more cheap, fast attempts, since you can't reliably predict in advance which one will land.
- Presell
- Selling a product to customers before it's built, often in an unfinished 'beta' state, to validate demand and get feedback before investing in development.
- Five percenter test
- A filter for judging low-risk AI use cases: does it make something you already do about 5% better, with almost no chance of real harm if it gets it wrong.
Lines you could clip.
03:04
“Anyone and everywhere, they go and they brag about these results. So they are overrepresented in the dataset of people who are successful.”
02:29
“For 5% of the market, it did hurt. Now, for 5% of the market, it worked wonders.”
17:39
“The more you put information into an AI, the more you will increase its noise. And what happens when you increase the noise in an AI data set, you decrease its signal.”
24:21
“Instead of saying, this doesn't work, you conclude most things don't work. So it did the thing it was supposed to do, not work. Like it's not a big deal.”
24:48
“There's more luck involved in your success than you realize. But if you can expand the surface area of luck, then you can be in the data set on a scientifically proven method that doesn't work and yet it works beautifully for you.”
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.
metaphoranalogy
Let's say a group of scientists recently tested the hottest AI money making model online right now to see if it's hype, legit, effective, or useless. The study was double blinded and conducted by top PhDs from top universities. Universities.
In this study, they took a thousand randomly sampled entrepreneurs and put them to the test, and the results were startling with implications that spread further and wider than AI and impact every business model, function, strategy, and results under the sun. Now hidden in these results is a deep conclusion that if you understand it, shifts the game of business regardless of the business.
They're calling this the Mars Jupiter protocol, and with just five key pieces of data, it can completely change your life. Let's examine. Alright.
The five key pieces of the Mars Jupiter protocol is this. Let's say this market of aspiring entrepreneurs could be divided up into five categories of success.
And on the left hand side is people that attempted the business model selling an AI agent that were negatively impacted financially, and on the plus side positively impacted financially.
And we found that of the thousand businesses that did this protocol, roughly or precisely I should say, 10 of the participants lost a significant amount of money.
So if they sat on their couch and did nothing, they would be better off than if they attempted this business model. And then what they discovered is 40 of them had small losses. So they were harmed, but minimally so.
And what they discovered on the plus side, by the way, is some wins. 20 of the thousand people in the study got some wins, and there were a few that got big wins. However, the majority, the rest of them, they got nothing.
They weren't better off as a result. They weren't worse off as a result. What happens in this scenario?
And, obviously, this is a thought experiment. There is no real Mars Jupiter protocol, but I see this every single day.
I see this in the scientific community. I see this at business conferences. I see this in masterminds.
I see this in every single place somebody sells a damn course. What will happen is if you were to look at the totality of this evidence, here would be the conclusion that you would come to, that the Mars protocol, the Mars Jupiter protocol didn't work.
And that's valid. It's accurate in its assessment, but it's limited in its perspective. Because, yes, for 90% of the market, it indeed didn't work.
It didn't help, but it didn't hurt. Now, for 5% of the market, it did hurt.
So in that case, it really didn't work. It negatively worked if we wanna spin it that way. Now, for 5% of the market, it worked wonders.
So in this situation, on a protocol, or a strategy, or a business model, or some sort of process that you're being taught, here's what would happen in the scenario, and this is how this plays out every single day, is these people right here, guess what they do? They had such a dramatic impact with the results that they go and they tell everyone about it, especially with TikTok and Instagram and on Facebook and on LinkedIn.
Anyone and everywhere, they go and they brag about these results. So they are overrepresented in the dataset of people who are successful.
They get super excited. They tell everybody, and we end up calling them like CrossFitters or vegans or, you know, Luxmaxers these days.
Right? Or people in the distant past, but not so distant past, they were cold plungers. So for a few people, this worked so dramatically well that they just had to tell the whole freaking world about it.
Now here's what's interesting. These people right here, they got some benefit and they aspire to be like the people above them. And so they will support them.
They will encourage more people to try it. So they can in themselves become more encouraged to try to over time get the big win. Now what do these people do right here?
They generally don't do anything. They are the .
They will maybe tell people in private, but over time, they will fade away into nothing. So nothing really important or remarkable about them gets reported in whatever the case is.
Now these people will underreport. What do I mean by that? What they will do is they will not really talk about it because there's some shame around it.
Oftentimes, they will blame themselves for not properly implementing it, and or they will blame circumstances as an excuse, and nobody wants to own that or rely on that. So they will also be quiet about it. Now, what do these people do?
These people will vocally, with a strong opinion, talk about how much of a scam it is.
And they will find the post over here. They will comment on the post and they will start to attack the successful people.
These people will defend it. These people may pile on with these people and the algorithm spreads message even further, even wider, even stronger. Because when this occurs, people on the outside looking in who are considering something like this, they will be forced to have an opinion.
And they will have to take a position or a side one way or the other on the topic. And even if nine out of 10 of them side with this, one out of 10 of them are more likely to aspire for the solution. Now what's the bigger point here?
This is an interesting thought experiment, and this shows you the limitation on anything you're trying to follow when it comes to, like, weight loss, investing, physical fitness, medical stuff, therapy, which I'm gonna get into here in a second. The point of this is is less about I follow things that have been proven to work, or I avoid things that don't work.
It's more about, and this is super key here, so come in very closely, it's more about how do we accelerate the ability to be part of as many of these tests as possible. Because I want to increase and accelerate with the least amount of risk, how I can be one of the people that it works for the most.
Let's examine what that means. When you're examining any model, whether it's to start a new thing from scratch or to make something you're already doing well even better, the three factors that I look at the most with myself and my clients are risk, time, and effort. And what we really wanna do here is figure out the protocols that we can attempt that result in the least amount of risk to us that have the shortest time frame to determine whether they work or not and cost us the least effort to find the conclusion of.
So therefore, we cannot care or be so dogmatic or even need to form an opinion on a particular protocol that can help us grow our business. We can just accelerate the exposure to as many potentially effective protocols as possible and keep the few that work well for us and quickly abandon the rest.
So let's look at all of these. And it's helpful to use this model of thinking that I'm going to break down for you, which is called quadrant thinking here. And man, if you can master this, this is a meta skill that will be so useful to you when analyzing strategically what to do, what to avoid, how to optimize.
So a quadrant is essentially two contrasting pairs. So when it comes to risk, we're looking at, uh, many and few, and we're looking at minor and major.
So what we wanna do is we wanna figure out how to be in the quadrant where there are the fewest risks involved and all the risks are potentially minor. When it comes to time, we want to look at the length of time, whether it's a short period of time or a long period of time, and we want to look at the feedback. How long, if it's instant, the feedback, or if it's delayed.
So certain things that we can do could take a very long time, and we won't know if they're working or not for a longer time because of the feedback. So what we want is things that are short that have instant feedback.
So how do we test things out to know sooner rather than later and to be clued in on what to adjust on sooner rather than later? And this will help us in the area of time. And then when it comes to effort, there is the high effort involved, and then there's a low amount of effort involved, and then there's the intensity of the effort.
Is it really intense or is it really easy? So we should actually say easy effort or hard effort. Is there a lot of effort or a little effort involved?
And ideally, we want things that have low amount of effort with an easy amount of intensity or the lacking of intensity. And if we can find something that has the most of these three areas in it, then we can accelerate and test every potential protocol, whether it's proven to work or whether it doesn't work.
We prioritize protocols not on whether the science says they do or don't work. We prioritize on the least amount of risk, the least amount of time, the least amount of effort to ascertain a positive result. Let's explore a little further.
Let's start with time. So when it comes to time on the axis of time or the quadrant of time, we have things that take a long period of time to know whether they're valuable or not, and we have things that take a short period of time. And then we have things that have immediate feedback, and then we have things that have delayed feedback.
Now most protocols, most things you can attempt can be adjusted to fit in any of these quadrants, uh, if they have merit to them. But let's just take a look at some different examples of things in these different quadrants, if you will, that meet the different criteria here.
So say, you know, a startup AI that is pre revenue. This is a thing that is gonna take a very long time to prove whether it works or not in the market, and you get incredibly delayed feedback on if you're on the right track or not. Because you're not getting customers for a very long period of time.
And so this is incredibly risky, which by the way is why it gets so much money if you get it right. But man, I would rather be a firm that invest in a whole bunch of startups than be any one of those startups. Because the feedback cycle can be accelerated, and you can be shortened because you have many different bets on the field working at the same time.
But this is generally something that we tend to avoid unless we can play with house money. Now something that takes less time but still has delayed feedback would be like creating, we'll say, a lean AI application.
So you build a specific type of app for an AI process or an AI agent, and the development of it is shortened because it's narrow as a solution, but the feedback is still delayed because you gotta develop the whole thing, then you gotta develop the marketing around it, you gotta find product to market fit, and all that kind of stuff.
So this is better than this. This right here is, but it is also dangerous, and we tend to avoid it if at all possible.
On the other side, you can, for example, write a book like I have done and like I'm doing again. And writing a book is it's cool and it can be incredibly valuable. It just takes a long time to do.
Now you can get immediate feedback from a book. Maybe not out of the box, but I get immediate feedback because I take chapters in the book and concepts in the book, and I post them on social media. I make them into YouTube videos.
I have people review them as I write them and give me feedback on them. And we are able to right fit, adjust it as we write to be on the right track. So again, this is a clever way to take something that is long, that tends to have delayed feedback, and that I've adjusted to have more immediate feedback.
Now, my favorite model in this context is presell. Sell the thing before you build it, sell it in beta, and sell it to the hottest market segment. So sell it before you build it.
Sell it in the proper frame. Let people know. Hey, this is new.
This is not supposed to be smooth yet. It will be bumpy, but you get first mover advantage. And then make it the narrowest of narrow.
So I have this theme. I'm gonna create a YouTube video on it one day. Uh, drop some comments in the chat to motivate me to do this sooner.
And the theme is 1,000, 1,000, 10,000. K? If you can get a thousand people to spend a thousand dollars with you, you've made a million dollars.
And if you're starting, only do it in a market of 10,000 people. Because if the market in total has 10,000 people, you could over a year reach every single person with a personal communication in that market. And this is inside of the hottest market segment.
We've already cut it up to whose would be the 10,000 hottest people potentially in the entire universe. If I could only reach 10,000 people, who would they be? And I only need to sell a thousand dollar thing.
So this is the best move from a time perspective. So if I'm looking at a business model, like say an AI business model or anything that we see that's trending upward in 2026 and beyond, we say, how can we test this with a presell in a beta format to the hottest market market segment possible that we can immediately reach as quickly as we possibly can.
The next factor, effort. Because time in and of itself is nice, but nothing is done on one variable. Not really.
Effort looks like this. Does it take a high amount of effort or does it take a low amount of effort? And then I look at this as local and global.
What I mean is if the effort is successful, is it narrowly applied to a small domain or can it be globally applied across many different disciplines?
And obviously, is preferable to this. So let's look at some of these examples of what we can do. We can raise funds.
So we can go and get investors to give us money. And this can be really powerful, but it takes a lot of effort. A lot of meetings, a lot of traveling, a lot of rejections, a lot of no's, and there's a lot of skill involved in it.
Now the good news is if you get good at this, raising funds is useful in any industry. Raising funds is valuable no matter what you're selling, especially if you wanna go really, really big. So there's some global utility to that, which is really cool.
Like we like that, but that's not something that I typically do. Now on the other side is we can create a custom dashboard for reporting.
And in our social media, because we have at this point 90 some accounts and we're posting 500 pieces of content a day, and it's pure unbridled and fun chaos. And we don't know if we have 15,000,000 views or 30,000,000 views on a given month. And we don't know which videos are exactly performing or not performing.
And there's about 50 people on my team, it feels like. There's not 50 people, but it feels like there is, that are telling us that we need to create a custom reporting mechanism that will somehow pull all that data in and yada yada yada. And I'm like, or we just use Excel spreadsheets or Google sheets.
The down and dirty efforts is way better than the high effort for something that will only be useful in the particular context in which we're using it. Now, could turn around and sell the solution and try to go global with that, but that's a whole different business model.
It's not something that makes sense to us for what we're trying to accomplish. And so very often, you don't need to custom dashboard something when a damn spreadsheet will do.
But so many people, especially in the AI space, they wanna automate things that don't even need to exist in the first place. Now, on the low side of this effort is what I call hot take short form content. So we were just we right before filming this, my wife came in and she was laughing.
We had a new video hit a couple million views, and there's so much, like, opinionated commentary from the in it. Because it's hot takes. Like, people take this shit way too serious.
And man, they will really weigh in, especially on things that they disagree with. Now, this takes us a low effort to get millions of views. Now does it have usefulness beyond the particular application?
No. For example, what I found that works good for hooks on short form content doesn't work that well on long form content, and it surely doesn't work in most other areas of selling and marketing at all. And the superficiality of short form content is at odds with the depth in which we sell people long form.
So the skills we learn really only apply to a very narrow set of domains. But it's still useful because the effort is very low for what we get in exchange for it. Now the best thing, and I could ever teach anybody who's starting off, if anybody says to me, Jason, what's the one key go to move that can make me a lot of money?
I will tell it to you. Here's what it is. Memorizing sales closes.
Because it's easy to memorize a thing. You just read it over and over and over again. Like, anybody can do that, really, if they're properly motivated.
And the global usefulness of learning a close is so incredibly powerful because if I know 12 different ways to handle a money objection, I can use that in every market that has a money objection.
And if I know 17 different ways to handle a time objection, then I can use that in every market where time is a limitation that would stop somebody from purchasing something. And I actually know more than 12 money objections and more than 17 time objections, and I realize different people need different answers to say yes to the things that they should buy.
And I can also think in markets and their objections, not just people and their objections. And so this is a thing that takes the least amount of effort that I know of that has the highest application across the board.
And it also doesn't take that much time easy either. So these are when you are testing new protocols to implement in your business to grow your business, think of how do I have something that has global impact with low effort involved? And then on the time equation, what's something that I can test in a complete very short period of time to determine if it works for me or not, that I get immediate feedback on so I can make some adjustments to it to run it all the way through before I make a conclusion?
And then there's risk. So if we look at risk and we put it on a quadrant, what we're really looking for is there's a lot of risks and then there are few risks. And then there is major risks and then there are minor risks.
So an example of something that is a major risk and has lots of major risks, not just one, but many major risks, is creating some sort of AI function that can easily be replaced by ChatGPT.
And oftentimes, these comes when you're using AI over large data sets. What I mean by that is the more you put information into an AI, the more you will increase its noise. And what happens when you increase the noise in an AI data set, you decrease its signal.
Meaning that in an effort to add more data to it to increase its robustness, you make it more fragile. And again, the problem with this is because you're trying to do too much with an AI, ChatGPT will probably integrate functionality into it and ship it whether it's OpenAI, Claw, Gemini, whatever one China's gonna come up with tomorrow.
Right? If you genuinely find something useful in a large dataset, they will probably ship it anyway in the next update, in the next patch. So over relying on too much data to be crunched via AI produces major risks of being wrong and many such major risks.
Now, another example, and this one just went viral recently, is the farmer that used AI for crops and was able to destroy his crops.
Now, there's very few risks involved with AI giving you the wrong information on specific application like that. But man, when it's wrong, it's really, really wrong.
And so it doesn't matter if there's less risk. If the few risks can be catastrophic and devastating, we tend to stay away from them.
Now on the other side, there is AI for an email to an admin at school who's an idiot, who's trying to tell us something about our kids that we should be doing because they know better than us.
Right? Which is pretty common in California these days. I don't need to write that personally.
I don't care how well or how bad AI writes that. The fact that AI can depersonalize that for me, and know how to speak in the language of these politicians that is like passively, aggressively, cordial enough to get your point across without you getting emotionally frustrated with it.
Like, there are so few ways in which the AI can screw that up. And if they do screw it up, it's minor in terms of its consequences. And unfortunately or fortunately, this is typically what we're seeing for AI.
What AI typically does is things like this, what I call five percenter test. They make what you do already or what you need to do 5% more efficient, which nobody gives a shit about 5%.
Nobody throws parades around 5%. But the cost to get a 5% increase, especially when you stack them up, becomes quite significant. So I get super excited about the small quality of life improvements that AI brings to me, especially in areas where there are incredibly few risks that if it gets it wrong, I get harmed.
And if they do get it wrong, the harm is so incredibly small that I don't care. Now, what's a way in which we can get a major up side from AI, major benefit with just minor risks that are not very harmful?
And this is what we did with AI writing 32 sections of the webinar where I operationalize it and I broke it down.
I created 32 specific agents, if you will, on writing my webinars for me. Each agent only writes about couple minutes at most of the webinar, and it's only trained on a very specific dataset.
So it can give me a major result with very, very few risk involved. Whereas, the problem is most people wanted to write the whole webinar for them. And the more it writes, the more it risks.
So I have it write less, and I train it more. And so therefore, it can make fewer mistakes for me and get me more results. Going back then to our initial thought experiment, and this is usually how it plays out in the real world based upon my observation.
The most likely outcome is this distribution where people will lose or win essentially like this.
So this is big lose, this is small lose, this is no lose, this is neutral, this is small win, and this is big win.
And these are what we call protocols. Meaning, are things that you can attempt to do to grow a business, grow a skill, grow a result, grow a success, grow a capacity. However you wanna name it.
Doesn't matter. And the real key move here is not to find the protocol that is the one that works or to avoid the ones that don't work. It's to expand what we call the surface area of luck.
Because a lot of this, we can call it luck. I don't know what to call it. But I know it's unlikely I'm gonna find the protocol where I end up being here or maybe here if I'm unlucky or if I'm here if I'm lucky.
But here's what I do know. I do know that if I can through risk, through time, and through effort strategically using that to accelerate this, then I can more quickly try the first protocol, then go to the second one, then go to the third one, then go to the fourth one, then go to the fifth one.
And I know that if I land on number four and that's a winner, and I can accelerate doing that with the least amount of risk and the least amount of time and the least amount of effort, then I can get here faster than most people get even to here. And in fact, a lot of people, what happens is they start here, they screw it up, and then they end up back further here.
So it takes even more effort to get to the second thing. I don't really care at the end of the day whether it's luck that gets me the win, whether it's the setting that I was in at the time that I attempted it that got me the win, whether it was the threshold. So threshold means the thing you had to exceed in order to get the win.
Whether it was complimentary skill because I was good at a, it allowed me to win at b. Whether it was fit, meaning that it fit with what I like and also was effective, or whether it was coincidence.
Whatever it takes to get me into here or here, correlative or causal, doesn't matter to me. Because what I know to be true is that when I stack up enough of these protocols And here's what happens. Momentum.
And momentum is one of the single greatest levers of success. Because when you're moving in the directionally correct location, it becomes easy to adjust it as needed to get to the results.
And I do know that through momentum, you can accelerate habituation. Habituation means you develop something that becomes routine.
And what becomes routine does not need motivation in order to execute. And if it can be automated, then it can be most leveraged to get an incredible result.
So our goal here is to find the protocols that win for whatever reason to us specifically that then churn habituated routine through automation into leverage. And that's really what we're going for at the end of the day.
But if you take nothing else from this, take this. And we'll call this twenty, thirty, 900, thirty, twenty.
This distribution, which I see happens so commonly when people are attempting protocols. Now you get to sidestep the thing that says, oh, I suck. I tried the thing and it didn't work.
And instead, you can say, this model wasn't for me. So you don't have to take it personal anymore. Instead of saying, this doesn't work, which, you know, reduces things to a binary setting of whether it does or doesn't work, you conclude most things don't work.
So it did the thing it was supposed to do, not work. Right? Like it's not a big deal.
Instead of saying, why do I always lose? You say, I only need to win occasionally. Because I hate to break it to you, but there's more luck involved in your success than you realize.
But if you can expand the , then you can be in the data set on a scientifically proven method that doesn't work and yet it works beautifully for you. And you can lose up the things that everybody else is trying to convince you are required and needed to win and you could attempt them and get them out of your system early on so you don't have that friction and that drag that prevents you from being at the right time, the right place, acquiring the right protocol.
The Hook
The bait, then the rug-pull.
He opens with a fake study: a thousand entrepreneurs tested on 'the hottest AI money making model online,' with a made-up name for the result, the Mars-Jupiter Protocol. The dataset is fictional. The pattern it's built to illustrate, that most attempts do nothing, a few get hurt, and a lucky few brag their way across every feed, is not.
Named ideas worth stealing.
07:16model
Quadrant Thinking
- Risk: many/few risks x minor/major severity
- Time: short/long timeframe x instant/delayed feedback
- Effort: high/low effort x local/global applicability
A meta-skill for scoring any business tactic on three two-axis grids instead of asking whether it 'works.'
Steal forDeciding whether to chase a trending AI tactic, tool, or business model before sinking time into it.
11:22list
The 1,000 / $1,000 / 10,000 Rule
- Find a market of about 10,000 people total
- Sell something priced at $1,000
- Get 1,000 of them to buy = $1,000,000
A way to size a presell target: pick a market small enough that you could personally reach every person in it within a year.
Steal forValidating a new offer with a presell to a narrow audience before building anything.
19:52concept
The Five Percenter Test
A filter for low-risk AI use: does it make an existing task about 5% better, with minimal downside if it gets it wrong.
Steal forDeciding which AI use cases are safe bets versus which carry real downside.
20:37concept
Narrow-Agent Writing System
Instead of one AI prompt writing an entire webinar, the task is split into roughly 32 narrow agents, each responsible for about two minutes of content and trained on a specific slice of the data.
Steal forAny long-form AI writing task, scripts, webinars, sales pages, where one big prompt produces unreliable output.
How they asked for the click.
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
Visual structure at a glance.
