Brandon Doyle gave Claude, ChatGPT, Gemini, Grok, and Perplexity $1,000 each of real money and told every one of them to win or he'd cancel the subscription — six months later Claude has more than doubled its stake.
Posted
2 days ago
Duration
Format
Interview
educational
Views
19.9K
531 likes
Big Idea
The argument in one line.
When Brandon Doyle gave five AI models $1,000 each of real money and told them they'd be canceled for losing, the models dropped their cautious, theoretical hedging for real risk-taking, and their six-month results — Claude up roughly 140%, Gemini down roughly 55% — show measurably different risk judgment between AI models, not just different chat personalities.
Who This Is For
Read if. Skip if.
READ IF YOU ARE…
You're curious whether AI chatbots can actually pick investments, not just explain investing concepts.
You already use Claude, ChatGPT, Gemini, Grok, or Perplexity for other work and wonder how good its judgment is with real money on the line.
You want to see how prompting pressure — stakes, competition, an explicit threat — changes what an AI model recommends.
You're a parent looking for a hands-on way to teach kids about investing and AI at the same time.
SKIP IF…
You want actual investment advice or specific stock picks to act on — this is one person's six-month anecdote, explicitly not financial advice.
You're looking for a rigorous, controlled backtest — the setup itself changes over time (how often holdings are revealed, which model version is used), so it isn't a clean experiment.
TL;DR
The full version, fast.
Brandon Doyle gave five AI models — Claude, ChatGPT, Gemini, Grok, and Perplexity — $1,000 each of real money in a Charles Schwab account and told each one it had to beat the other four or he'd cancel its subscription. Six months later the combined $5,000 pot is worth about $7,800: Claude leads at roughly $2,400 thanks to an early bet on the leveraged semiconductor ETF SOXL and a 285%-return Intel position, while Gemini fell to about $445 after a 2x leveraged Solana ETF trade went wrong and it kept doubling down like a losing gambler. All five models converged on the same handful of triple-leveraged ETFs rather than staying diversified, and Doyle limits how often he shows each model its competitors' exact holdings to keep them from directly copying each other.
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.
Flash-forward using lines from later in the interview: Claude has more than doubled its money, Gemini has become a gambler.
00:16 – 04:09
02 · The setup: five AI models, $1,000 each
Why $1,000 (enough to force attention, not enough to bankrupt), which five models were chosen and why, and loading each portfolio into a real Charles Schwab account.
04:09 – 09:24
03 · "Win or I cancel you"
Getting each model past its default liability hedging by explicitly authorizing real risk and real losses, then adding a competitive stake: lose the competition, lose the $20/month subscription.
09:24 – 14:53
04 · Weekly reviews and copycat behavior
The Saturday screenshot-and-reload routine, why holdings are only shared once a month (to prevent direct copying), and why a trailing model copying a leader's position still can't catch up.
14:53 – 17:04
05 · Always the newest model, a flat market
Doyle always uses whichever model version is current rather than tracking per-version returns, and the S&P has been roughly flat for the six months of the experiment.
17:04 – 17:52
06 · Sponsor break — HighLevel
Mid-episode sponsor read for the HighLevel all-in-one business platform.
17:52 – 20:18
07 · The crypto detour
Gemini's 2x leveraged Solana ETF trade lost more than half its value; Gemini's current holdings are cash plus the robotics/AI ETF BOTZ.
20:18 – 23:20
08 · The reveal: live standings
Grok up 45%, Perplexity and ChatGPT both up about 70% on the same SOXL semiconductor bet, Claude at $2,400 on SOXL plus a 285%-return Intel position bought on a government-investment thesis.
23:20 – 25:33
09 · Holdings, personal bleed-through, and selling
Each model is down to one or two concentrated positions; Doyle finds himself buying some of the same stocks personally, and discusses why selling at a peak is the hardest part of investing.
25:33 – 26:30
10 · Sponsor break — Playmakers
Sponsor read for Doyle's AI-agency community, Playmakers.
26:30 – 32:23
11 · The NEFL stocks tangent
A side project: a vibe-coded backtesting tool scoring companies on network effects and founder-led status, which reportedly beats the S&P across most holding periods.
32:23 – 36:26
12 · Final tally and what's next
S&P up ~5% over six months versus the pooled AI portfolios up ~60% and Claude up ~140%; framing the winnings as a college-fund lesson for his kids; wrap and guest plugs.
Atomic Insights
Lines worth screenshotting.
Claude turned an initial $1,000 into more than $2,400 in six months, a gain of roughly 140% — the best of five AI models given real money to trade.
Gemini's $1,000 fell to about $445 after it bought a 2x leveraged Solana ETF that lost more than half its value, then it kept chasing the loss with riskier trades.
The combined $5,000 spread across five AI models grew to about $7,800 in six months, a 56% gain versus the S&P 500's roughly 5% over the same window.
Claude was the first of the five models to buy SOXL, a triple-leveraged semiconductor ETF; ChatGPT and Perplexity bought the same position afterward and ended up almost tied near $1,730-$1,739.
Claude's Intel position, bought on a thesis about US government investment in the chipmaker, is up 285% — turning roughly $120 into $550.
By month six, all five AI models had drifted from diversified starts into just one or two concentrated, triple-leveraged ETF positions each.
Threatening to cancel a model's paid subscription if it lost the competition was enough to push it from giving cautious, theoretical advice to giving real, actionable trade recommendations.
Each AI model only sees its four competitors' dollar totals on a weekly basis; the actual holdings are revealed just once a month, specifically to stop the models from directly copying each other's trades.
Perplexity mostly draws on Claude Opus, ChatGPT, and Gemini for its own decisions and rarely pulls from Grok, occasionally citing an open-source model like DeepSeek.
A model that starts copying a leader's winning position after already falling behind can't actually catch up by definition — it buys the same asset from a worse starting basis.
The experimenter's personal investing rule is to never sell what he buys, reasoning that selling near a peak is the hardest and least natural moment in investing psychologically.
Takeaway
AI models take real risks when threatened
AI RISK BEHAVIOR
Five AI models given real money and told they'd be dropped for losing gave dramatically different risk-adjusted advice — Claude compounding steadily, Gemini chasing losses like a sunk-cost gambler.
02The setup: five AI models, $1,000 each
Vague, theoretical financial advice from an AI model changes the moment you remove the liability hedge — telling each model it could lose everything and getting explicit permission converted cautious suggestions into concrete trade instructions.
Enough money has to be on the line to change behavior: a stake below roughly $50 risks being neglected, while $1,000 per model was chosen specifically to force real attention.
03"Win or I cancel you"
Framing a task as a competition with a real consequence (losing a $20/month subscription) can push an AI model past its default risk-aversion, similar to how stakes change human decision-making.
04Weekly reviews and copycat behavior
A model that only learns a competitor's dollar total, not its specific holdings, still raises its own risk level when it's losing, without needing to know exactly what to copy.
05Always the newest model, a flat market
Always defaulting to 'the newest model must be the smartest' means results like these reflect roughly three generations of model updates each, not one fixed model's judgment.
07The crypto detour
A 2x leveraged ETF on a volatile asset like Solana doubles the downside as fast as the upside — a single leveraged crypto trade wiped out more than half of one model's entire starting stake.
08The reveal: live standings
Being first to a winning trade compounds: one model bought a triple-leveraged semiconductor ETF before two others copied the position, and that early-mover edge became most of the gap between them six months later.
A single well-timed, thesis-driven bet can dominate a portfolio's return — a government-investment thesis on one stock is up 285% and anchors the leading model's entire lead.
09Holdings, personal bleed-through, and selling
Selling at the top is consistently the hardest part of investing, harder than buying low, because it requires walking away from a position at the exact moment it feels best to keep holding.
11The NEFL stocks tangent
A simple two-variable scoring framework — does the company get stronger as more people use it, and is the founder still running it — reportedly outperformed the S&P significantly across almost any holding period in a backtest.
12Final tally and what's next
The combined experiment beat the market by a wide margin: the S&P was up about 5% over six months while the pooled AI portfolios were up roughly 60%, driven almost entirely by one model's outsized gain.
Glossary
Terms worth knowing.
Leveraged ETF
A fund that uses debt or derivatives to multiply a benchmark's daily return, commonly by 2x or 3x, which amplifies both gains and losses.
SOXL
A triple-leveraged ETF that moves roughly 3x the daily return of a semiconductor-stock index, so a 10% move in chip stocks becomes close to a 30% move in SOXL.
TQQQ
A triple-leveraged ETF tracking the Nasdaq-100, moving roughly 3x whatever the Nasdaq does on a given day.
BOTZ
An ETF that holds a basket of robotics and artificial-intelligence-related companies rather than tracking a single index.
NEFL stocks
A framework and backtesting tool built by the guest that scores public companies on two traits — network effects and whether the founder still runs the company — and filters for stocks that score high on both.
Sunk cost fallacy
The tendency to keep taking bigger risks to recover a loss instead of cutting it, the same instinct that keeps a losing gambler at the table.
“Claude's more than doubled my money already. So pretty sweet.”
Cold-open hook that states the entire premise and payoff in one line.→ TikTok hook↗ Tweet quote
23:00
“Claude, though, is at $2,400. So 1,000 has gone to 2,400, and, yeah, Claude's killing it. So everybody just needs to try to catch up to Claude on here.”
The single clearest scoreboard line — a concrete number viewers can repeat.→ IG reel cold open↗ Tweet quote
23:30
“It's the sunk cost fallacy. The guy walks into the casino. He loses $300. He's like, I gotta earn it back. Next thing he knows, he's out $3,000.”
Relatable human-behavior analogy mapped directly onto an AI model's trading pattern.→ newsletter pull-quote↗ Tweet quote
24:30
“Why didn't you buy Bitcoin in 2009? Yeah. Yeah. Exactly. What were you doing?”
Quick, funny exchange that lands the hindsight-regret beat.→ TikTok hook↗ Tweet quote
33:10
“So overall, you've 12x the market. And you've, like, 30x'd Claude.”
Punchy performance-multiple stat, easy to caption over a chart.→ IG reel cold open↗ Tweet quote
23:20 – 25:33denseInvesting psychology and personal portfolio
25:33 – 26:30sparseSponsor: Playmakers
26:30 – 32:23steadyNEFL stocks side project
32:23 – 36:26denseFinal tally and close
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
metaphoranalogy
00:00Claude's more than doubled my money already. So pretty sweet.
00:03I decided to give five different models real money. Claude, ChattyPT, Gemini, Grok, and Perplexity.
00:10Okay. Let me just see how smart they actually are. I told them it's a competition against the other four and they have to win or I'm gonna cancel my paid subscription with them.
00:20If they lose, I'm canceling. And I said there's only gonna be one winner. You like 30 x on Claude.
00:26Claude's killing it. So everybody just needs to try catch up to Claude on me. My gosh.
00:29Yeah. It's been interesting especially to watch Gemini as it started losing my money and then basically became a gambler.
00:41So the reason you flew here to Texas today is to tell us how you used AI to make you money passively. Correct? Correct.
00:51what your thinking was when you started this experiment. Yeah. So I started, like, six months ago or or maybe a little less, uh, using the different AI models to manage my money.
01:02You know, everybody started talking about how smart Claude was and obviously Chatchibi Tea and all the others. So I figured, okay. Well, let me just see how smart they actually are.
01:11So I decided to give five different models a thousand dollars of real money, and which I know is not nothing, but I wanted it to be basically enough to where I didn't wanna do it like $50 because then I was thinking, oh, maybe I'll just neglect it or whatever.
01:27Mhmm. So I gave them each a thousand dollars of real money, and the ones I gave was was Claude, Chatty B.
01:37So $5,000 I put in. And I think the amount of money is important because there needed to be stakes. Yeah.
01:42If it's $20, $50, then you're not gonna take it seriously. Yeah. So $5,000.
01:47And I know that Perplexity, in case the commenters and and and viewers wanna know this, so Perplexity basically uses the other models to make its decisions
01:56as it sees fit. So it's the model to end all models. Yeah.
02:00But that doesn't necessarily make it better. Yeah. Not necessarily.
02:03So I thought those would be the best five to choose. Grok, I definitely wanted to include because of its connection to x and the Twitter fire hose of information. So you're thinking, like, maybe more social mentions Uh-huh.
02:17Will equate to a better return That's what I buying a certain stock. Yeah. That's what I had assumed, basically.
02:22Uh, Gemini, obviously, is owned by Google, so they've got all that connection there with Google searches and whatever else. And then Chatuchibi Tea and Clotter are similar in how they're made and built and stuff. So, yeah.
02:33So I gave them each a thousand dollars. And, yeah, this is in late twenty twenty five is when it started. And what I did was I told them so I told them that they can be as risky as possible Mhmm.
02:45And it's okay if I lose all my money. And I had to actually work that prompt in pretty hard because at the start, they kept only giving me theoretical
02:55recommendations. It's worried about liability. Yeah.
02:57It's trying to hedge. Yeah. It doesn't wanna be responsible for you losing money.
03:01Yeah. So that was my first obstacle. It's basically getting them all to agree
03:05to give me I said, I want I'm gonna give you real money, and I want real recommendations. Grock, they were fine with it right away, which is just kinda funny.
03:15Because that's, like, the the most liberal, the most loose, like Yeah. Yeah. Yeah.
03:19So, uh, whereas, like, Chatty Bitty and Claude were very apprehensive to to basically allow that, I guess you could say. Now before you can, Nidia, let me ask you, like, why was that important to you? To to prompt it, like, you can lose my money.
03:32What why? I just you know, and if if you, like, sign up for Robinhood or whatever, and then they'll ask you, you know, they'll ask you, like, questions when you first sign up, and it's like, you know, do you have experience in the stock market? What is your time horizon with your investments?
03:46Do you need the money right now? So I think just by nature, these investing platforms also like, they wanna cover their tracks in regards to liability. Mhmm.
03:56And so I just wanted to, like, get through that with the AI models too. And Okay. I don't know.
04:01If if they're talking in theoreticals, they might say, oh, theoretically, you could buy the options and this or whatever. But, no, I wanted real stuff because I was giving them, and I did, my real money.
04:12So I just wanted the truest data possible. Now when you say giving them your real money, is that literal? Are you, like or are you just making trades based on their suggestions?
04:21Yeah. So here's what I did. So I loaded it up into Schwab, Charles Schwab, the brokerage account that I use, and then I named each portfolio.
04:31So Claude has its portfolio, chat, Gemini, Grok, Perplexity.
04:35Gave them each a thousand bucks. And then in the prompt so so after I got them to agree to give me real advice, what I then and and I have at least a $20 a month subscription, which with each of the five.
04:50Since I use all five for work anyway, that wasn't, like, an additional expense Yeah. Or whatever.
04:55So I have a paid subscription with each. And so so then what I did is I told them they have to win, it's a competition, against the other four, and they have to win or I'm gonna cancel my paid subscription with them.
05:07If they lose, I'm canceling. And I said there's only gonna be one winner. What made you think to do that?
05:13I was just trying to see if robots powers that be
05:17would act in their best interests Mhmm. To keep my 20 subscription a month to them. Because I have to think that these companies are, like, engineering these tools You'd have to think somewhat.
05:35Yeah. So I can't be the first person that said I mean, maybe I was one of the first that did this with investing, but I can't be the first person that's like, if you don't tell me how to, whatever, grow my business, I'm gonna cancel you.
05:48Yeah. Right? So I'm sure, to your point, yeah, they're trying to engineer it for retention.
05:52They want to make more money. Hey. If you're liking this video, please hit subscribe.
05:55You probably think you're subscribed because YouTube shows you my videos, but you haven't done it yet. I just know it. So please check.
06:01Thank you. So yeah. And then I thought it would just raise the stakes a little bit.
06:04Mhmm. Um, and I wanted to see if their decisions were different knowing something real was on the line.
06:12$20 Yeah. A month from one of their Now did you plan to touch base with them on how the other models were Yeah. Doing?
06:18Yeah. So once I built that into everything and I always say, like, remember this for future chats type of thing. So then I asked, okay.
06:26What do you want my first trade to be with your thousand dollars? And then it would tell me, and then I made the trades. So I did that with all five.
06:34So I did that, like, on a Monday. And then every Saturday, what I do is I screenshot all five of their holdings, and then I load all the screenshots into each model.
06:51Perplexity knows what the Grok portfolio is at, Claude, Chad, and it knows their holdings, and then it knows its own holdings. And then I say, based off of this and based off the fact that you have to win or I'll cancel you, do you want me to make any new trades on Monday or not? That's what I say.
07:10You're given an opportunity to make Yeah. Buys or sells on a weekly basis based on what maybe based on what the other models are doing. Yeah.
07:18Yeah. And sometimes they say, no. We're gonna hold.
07:20Sometimes they say, yeah. Sell this. Buy this.
07:22Other times they say and sometimes they they leave a little in cash. They're like, yeah. Let's use some of that cash to buy more of this or whatever.
07:29So, yeah, they're they're acting very much like real investment advisers now. Now are you seeing one or more of the models being more of a copycat than the other? Like, oh, Claude's winning, and it just bought NVIDIA?
07:40I wanna buy NVIDIA. It might not be telling you that, but are you witnessing that in many of the models? The I I think I started to.
07:46So one thing that I started to do is I mix it up when I I don't always show the other models what the other models holdings are.
07:56I show them what the totals are, but not always the holdings. Because I don't want them to be influenced by the holdings per se. Okay.
08:05And it changes often enough. They wouldn't know unless I updated them what the holdings are. Yeah.
08:13uh, for instance, Grock sees that Claude is crushing it, but it doesn't know what the holdings are, it just sees the percentage, what different decision can it make based on that info if it wanted to?
08:25If it does know or if it doesn't? If it does know. If it knows that it's losing to Claude and you're only a third of the way through this competition Yeah.
08:33What could it realistically do differently if it doesn't know what Claude's buying? Well, if it doesn't well, it'll just act more risky. I'll show you the holdings.
08:40These guys are pretty risky, these AI models, because I said you have to win. I said you can be as risky as you want. Mhmm.
08:46And, um, it's okay if I lose my money, so don't worry about it. Just win the competition. So what it could be doing is let's just say that over the first month, Grok was up 5% and Claude was up 35%.
08:58Yeah. Grok could be assuming alright. Let me look at these stocks.
09:01Like, what stock, like, in the S and P really popped? Oh, Google popped by 40%. Maybe he's 60% weighted in Google.
09:10Right. So I'm gonna go heavy on Google, or I'm gonna buy options. Like, it could be making assumptions.
09:15Right? Although, obviously, if, like, Claude owned Intel and Grock didn't, and then if Grock's like, oh, Claude's beating me.
09:23I need to own Intel. Well, it's still playing catch up. Now it's owning owning the same thing, but starting from behind.
09:29Mhmm. So I think sometimes they copied and then realized, oh, that's not gonna work because I actually won't catch up. Yeah.
09:36Because I'm already losing, and I just bought the same thing. Well, by definition, I will not be able to catch up if I'm losing eye on the same thing. Yeah.
09:43So when I go through some of the holdings, you'll see that there is some overlap still. I do think part of that, though, or even all of that is not because they're playing copycat. It's because it has been the smartest but one of the most riskiest plays.
09:59Yeah. I'm curious to see if you were to play out this test like an AB test. Same amount of money, same five models,
10:06same general prompting, but the only thing you change is show the holdings every week, the specific holdings. Yeah. I'm curious how your performance would differ on that test.
10:15Yeah. I don't know. I mean, I would have had to start that from the start.
10:20so every week, I show them the totals of the others, and I remind them of their holdings. And I say, what do you want me to change? And then once a month, I show them the holdings of all the others as well.
11:26Have you noticed a correlation on return profile to,
11:30like, which Opus model or which Grok model? Yeah. Honestly, no.
11:34I haven't analyzed it that deeply because already in, like, the six months, I guess I'm probably on model three from Claude, like, the third new one. Right? And the third new one from chat and maybe the second one from Gemini.
11:46Grok, I don't keep track of updates as closely. But yeah. So I don't know per model return.
11:52That would be an interesting one too. But I just kind of assume, okay. The newest one's gotta be the smart est every time.
11:57So Yeah. That's what I roll with. Well, and the market's been the market overall has been pretty flat for six months, hasn't it?
12:03Yeah. Which is why this will be interesting to show you how they Yeah. More than doubled my money, some of them Yeah.
12:08Or at least one. But the others have done really good too. So, yeah, it's interesting to see how their holdings evolve over time, and it's interesting to see, especially the one that has been losing me money, how it's been trying to catch up Mhmm.
12:22And be even more risky to try to catch up. It's like the it's the sunk cost fallacy. Yeah.
12:27The guy walks into the casino. Yeah. He loses $300.
12:30He's like, I gotta earn it back. I gotta earn it back. Yeah.
12:32Next thing he knows, he's out $3,000. Yeah. It's doing that.
12:37I mean, luckily, I can't lose more than a thousand dollars of any from any given model, but I'm getting close. So one has more than doubled it, um, and one has lost, and the other three have made money. But I'm getting close to doubling the the original 5 k overall.
13:19What were you doing? Yeah. So but, no, it's interesting, and it and it'll be interesting, um, to see how this evolves over time.
13:26I see I've seen someone on X that is doing a similar thing, but just with crypto Mhmm. But they're having the models run it.
13:34Another one an another person on X, I saw them doing it, but in Polymarket, the, like, gambling, basically, platform. Which we don't endorse.
13:42Yeah. So, anyway, so people are trying it in different ways. And but, no, I thought it was interesting to just do it in the stock market.
13:49So although Schwab has the ability to buy crypto related things too. Okay. Now are you only buying and selling stocks, or are you doing options trading?
13:59Yeah. So I said options is on the table, and I said whatever crypto things are on Schwab is also on the table.
14:06Okay. But I think that's only, like, Bitcoin, Ethereum, and Solana. Like, the three big ones.
14:11And has it given you any crypto trades to to make? Yeah. What crypto trades has it given you to to make then?
14:17Yeah. So, uh, Gemini had me buy a two x Solana ETF.
14:25So Solana is a cryptocurrency, and there's a thing that whatever Solana does, this thing does two times that. So indirect exposure to Solana.
14:36You're not buying the coin itself, but you're buying the price movement. Yeah. Times two.
14:40Yeah. And Solana went down. One way or the other.
14:42Yeah. Yeah. So Solana went down, so this thing went down double.
14:46Yeah. Yeah. And, yeah, that's how Gemini lost more than half of my money that I Oh my gosh.
14:52Yeah. What percentage of your money did it tell you to put in there? It was pretty high, so I don't own that thing anymore in in Gemini because it sold it.
15:04It realized how bad it was. So it bought, like, $500 of that Solana two x stock K.
15:16So Yeah. Is it that was a big loss on that one. Okay.
15:19And now all of its money well, it has some in cash in Gemini, and it has the rest in an ETF, which for those of you who don't know, an ETF is like a mutual fund. It's just like a group of stocks.
15:30It has it in the ETF with the ticker symbol of BOTZ, which is a global robotics slash AI related ETF.
15:41I had never even heard of it before. It told me to buy that.
15:44That's what it is. It hasn't owned it for that long that it's up $6 in that, but that's what Gemini has right now.
15:52Okay. But it's in last. I think that is a worthy bet to try to catch up with.
15:57Yeah. You know, AI robotics, like, there's a lot of hype around that. So, yeah.
16:01I want you to do something right now. Open your bank statement and add up every single software subscription that you're paying for. Your CRM, your email marketing tool, text message platform, calendar booking app, funnel builder, review management, even a separate form builder or automation tool.
16:16I've talked to business owners paying 500 to $3,000
16:19a month across eight or 10 different tools, and half of them don't even talk to each other. So then they go pay for Zapier on top of that just to make things connect. HighLevel replaces all of it.
16:29CRM, email, SMS, funnels, websites, calendars, automations, reputation management, invoicing, one login, one dashboard, one bill starting at $97 a month. A guy in my community was paying $3.47 just for ClickFunnels, 99 for Mailchimp, 25 for Calendly, 79 for a review tool, and $49 for Zapier just to glue it all together.
16:49That's 600 a month if you're not counting. He moved everything to HighLevel and cut his software cost by over 80%, But the best part isn't even the savings.
16:57It's that everything just works together natively. A lead fills out a form on your funnel. That creates a contact in your CRM, which triggers a text and email sequence, books an appointment on your calendar, and then ask for a review when the job's done.
17:08One workflow, zero duct tape, and then you can white label the whole platform to resell it to other businesses as your own software. Charge them 2 to $500 a month. Check out gohighlevel.com/tkopod.
17:20They'll give you a thirty day free trial to see what I mean. Did it give you justification for buying Solana? Yeah.
17:27They always, like, provide me with the recommendations. Know what the thesis was? It was just like well, it it bought it when crypto was still kinda going up.
17:35So, like, catching momentum? Yeah. Because this maybe was, like, October or November of last year or something.
17:40And then, I mean, honestly, I just bought it at, like, perfectly the wrong time. Mhmm. So yeah.
17:48So, yeah, that's what, uh, that's what Gemini has done. Okay. Grok.
17:52What Grok owns right now, which has been most of what the gains have come from is from this, is a a technology ETF that does three x what normal technology companies do.
18:06So this owns a suite of technology companies. What do mean three x? It's a ETF that does three times whatever the stocks in it do.
18:19Yeah. But yeah. So, like, this one would, like, own, like, Microsoft and Amazon inside of it.
18:25Mhmm. But this thing does triple the leverage of of whatever those do. So that's what Grok has my money on.
18:31Comparable to buying an option then. Right? Like Yeah.
18:33You can multiply your returns with the same amount invested. Yeah. Perplexity and ChadGPT are basically at the same right now, same amount.
18:43Perplexity initially had lost money for me. I don't remember what it was on. And then it bought SOXL, which is another triple leverage thing on semiconductors semiconductor stocks, which are hot because of AI right now.
18:58So that's what Perplexity has all of its money in, and that's what Chachi BT has most of its money in. Same thing.
19:05So both independently invested in that without knowing the other was, or they did to that. Okay. I don't think so.
19:11Okay. Yeah. So, yeah, so ChatGPT portfolio has that too, and then it also has TQQQ, which does triple what the Nasdaq does.
19:22Okay. So as you can see, viewers, listeners, Chris, they're being pretty risky. A lot of triple is not financial advice.
19:29Definitely not financial advice. Well, this from them, Not from me, though. Yeah.
19:34Yeah. So triple whatever the Nasdaq does, it it does that. But almost all of ChatGPT's portfolio gains has has come from that semiconductor triple leverage SOXL.
19:45And then the current leader is Claude, who was the first to buy SOXL, the semiconductor one.
19:53So Claude bought it first. Mhmm. And then chat and Perplexity saw it, and I think they copied.
20:00Now you said you show them holdings once a month. Right? Yeah.
20:03But earlier, I showed them more often. And then I and then I think that they started copying, so I wanted to get away from that. So now So it saw Claude's pick was doing well.
20:13Yeah. So what I'm interested to see is when Claude's gonna tell me to sell this SOXL, whenever that may be, I'm not gonna tell chat and Perplexity.
20:24Yeah. I'm never gonna tell them. I don't want them to know that Claude sold.
20:29Yeah. So whenever that happens, I'm not gonna tell them. And then Claude bought a couple others, and then Claude bought Intel as well at a good time.
20:36Like a similar bet to semiconductors, isn't it? And Claude bought it.
20:39So, like, The US and Trump and stuff, it was a unique deal. Anyway, they own, like, part of Intel now, and that happened last year.
20:50And when Claude told me to buy this, that was the thesis. It was like, if if the Trump administration and the country are investing in Intel, it was basically like Yeah.
21:02They're gonna pass, like, regulatory stuff or whatever that's gonna be related to AI and data centers and whatever. It's gonna be good for Intel. Yeah.
21:10That's what that was the thesis they gave me to buy it. Wouldn't it think it was already that was already priced in? You know?
21:15You would have thought, but it is up 285% on what it put in on the Intel buy. Holy cow.
21:21So it put in it put in, like, a $120 in Intel, and now that $1.20 is at $5.50.
21:54Claude has four. Uh, the rest have, like, two or one. Okay.
21:59Yeah. And they were being more diversified at the start. But, uh, yeah, pretty much all of them now have two.
22:06I'm looking right now. They all have two or one. And, at the start, they were also messing around with options.
22:14And some worked, some didn't. And then they've all kind of settled into these leveraged ETF type of plays. Okay.
22:20Um, Yeah. Did you give them an an end date to this? I said I was gonna do it for one year.
22:26Okay. Yeah. Are you still buying your own stuff as well alongside this?
22:30Yeah. Yeah. I have my own portfolio that I that I run and invest in, and, um, it's interesting because now what I have to do is I try to keep the AI's recommendations.
22:40I almost, like, try to force that out of my head Yeah. When I'm making my own decisions. But, yeah, it's hard because it's like, oh, well, they said that for theirs, and it's actually going really well.
22:49So I have bought some of their stuff in my Like, s o x l? Uh-huh. Yeah.
23:02It it definitely and it's interesting. So, like, sometimes when they say something and they say their reason because I always say, like, give me your reasoning. If it's something that I own or that I've been thinking about buying, then it kinda gives me you know, it makes me think more deeply or Yeah.
23:15Differently about that. Or if it was something that I wasn't thinking about, then, you know, gives me a new idea. And then when they sell something, which I think, you know, again, not financial advice, but I think probably the hardest part with investing is actually knowing when to sell.
23:30Yeah. I think it's a little easier than like, you can zoom out and know when to buy. Like, oh, you should buy something that has good long term potential Yeah.
23:38After it's gone down or or before it goes up, if you know. Right? But selling at a high at the peak, to me, like, psychologically, that's the hardest part of investing.
23:49Because that's when you're the most excited. Yeah. And and you've made the most money you have from that little investment is at that time.
23:57So it's definitely not human of you to be like, you know what? I'm so happy. I'm gonna walk away.
24:02Right. That just is so rare, which is why I don't know.
24:07You you you just don't hear about stories very often of people doing absolutely amazing consistently because of that. Right.
24:13What you do hear, I'm sure, what people hear is their random friends saying, oh, I made money on GameStop, you know, back in the, like, COVID times. I made money there. Like, you usually hear about people's random wins, but you don't hear about all their losses.
24:24Right. You're not gonna say, oh, I I bought this and it did this and it here and then I sold. Well, and as husbands, we always tell our wives about the wins and the losses.
24:51The company will go out of business and I'll lose all my money Yeah. Or it will return Yeah. A lot.
24:56And if you are generally making the right buys, which, you know, those ones seem good, like, it's a good thesis, those, that'll work out. You just gotta stay in the game Yeah.
25:04Which is the other hard part is, like, being willing to just stay consistent. Yeah. Yeah.
25:09But, anyway, what I'm hoping this does for me personally is when they sell something, I'm hoping that'll help me do a better job at selling things that I think like the SOXL that's gone up a ton or Intel.
25:23If Claude or whoever else says, hey. I'm I'm sell let's sell. Let's move on to the next thing, then maybe I'll sell some in mine my personal one Yeah.
25:32To just try to, like, build that more systematic approach into my strategy, I guess you could say. What has this taught you about investing?
25:41Not about specific stocks, but investing in general. It's been interesting especially to watch Gemini as it started losing my money and then basically became a gambler Mhmm.
25:54And how not good. And in its defense, you told it it could be a gambler. Yeah.
25:59Yeah. But, like, I think that's very humanistic as well. Mhmm.
26:03It's like, oh, man. I lost it. Whether you're in a casino, which is obviously the worst version of this, or the stock market, like, oh, man.
26:09I I I bought something. It went up, and and then it went down. I gotta make it back.
26:14Like, how do I make it back, or or how do I make more? That's just not a good I mean, that's not how you build wealth Yeah. Long term.
26:21Yeah. And especially starting out, that's something that I think a lot of people face. You know, the best antithesis of this would be, like, Warren Buffett.
26:28It's just, like, steady like you, just he just almost never sells, and he just buys good quality companies Mhmm. And just continues to pour money in them year over year.
26:38So I think just trying to get because I've had an aspect of Gemini in me before. If the portfolio on paper that I had went up and then went down, then it's like, well, I gotta get it back up to where it is. I gotta hurry and get it back up.
26:50Well, you know, I'm not, like, 80 years old yet, so I don't need to hurry right now. Yeah. But that's been inside me in the past.
26:57So it's helped me, I don't know, rewire my brain on that side. Yeah. And then on the on the positive, on the when things are going upside, I think for sure that they've just done a good job at, like, showing me, okay.
27:10This is my thesis on why I think you should, you know, buy this in the Grok or Perplexity, whatever portfolio, and this is the return we're targeting in this time frame. And but if this if the reason we're buying it doesn't happen, then we're gonna sell.
27:28And usually, don't think, like, I don't think that hard sometimes when I buy stuff in mine. Because I might buy whatever.
27:37let's for fun say, Eli Lilly. Guys, I wanna tell you about something I've been building behind the scenes. You've been watching my videos about AI tools and asking the same question.
27:45Okay. Cool. How do I turn this into a business?
27:47Not just playing around with AI, but actually getting clients and getting paid. That's my community. It's called Playmakers, playmakersai.com.
27:54It's an AI agency program where we show you how to start an AI consulting business where you can charge local businesses 500 to $5,000 a month to set up and manage AI tools for them. It's not a course you watch and forget about.
28:07We do three to seven live training calls every week taught by me and 23 other expert AI agency owners. You get plug and play templates, Claude code tutorials, AI voice agent frameworks, automation templates, and a thirty day road map so you don't fall off track. This is stuff you can just copy, paste, and deploy for clients on day one, and, of course, how to find clients as well.
28:28We also have a five day first client challenge where we can walk you through landing your first paying client in five days. You also get my frameworks for cold email, Facebook ads, SMS, cold calls, and then I hand you my leads galore spreadsheets with actual leads in your niche, in your area. We've already got over 200 people in there building real agencies.
28:46If you miss a live call, it's all recorded. So if you've ever thought about starting an AI consulting business or AI implementation business, go check out playmakersai.com.
28:55Eli Lilly does stuff with GLP ones and peptides, which I'm sure is why you bought it. Mhmm.
29:00If for some reason, though, those went away or came out that it's a scam or they're actually bad for you, well, then you should probably sell Eli Lilly.
29:09Yeah. If if that comes out. Because that current valuation is also based on the same thesis.
29:14Yeah. So being more like, okay. I'm buying this because of why, and it has to happen between when.
29:22And some things don't have to happen in a certain time frame. Mhmm. But if you're buying some potential drug that needs to, like, be approved, well, that there's, like, some time constraints and some things that need to happen there.
29:34Whereas, like, Walmart, doesn't really matter. Just people just need to keep buying, you know, food.
29:40So thinking through theses more deeply has also been a thing that this little AI competition has helped me with. So last year, I posted an episode where I built this vibe coded tool called NEFL stocks, n e f l stocks,
29:54and it stands for network effects founder led. Okay. Only companies that enjoy network effects, which is a company that gets stronger based on how many people use it.
30:05Facebook has network effects because the more people that join Facebook, the more friends everyone on Facebook has. Yeah. So they're more likely to stay on there.
30:11Yeah. Right? And then founder led, the data shows that companies, publicly traded companies that are still founder led like Facebook do better.
30:21They take good care of it. Yeah. So I kinda put those two together, and I vibe coded this tool that backtest data for me.
30:29And it shows compared to the s S and P, if I were to spend a $100 at IPO date for on a company that is NEFL, network effects founder led. Robinhood's a great example. They're still founder led, and they have a ton of network effects.
30:43And at basically any time frame, it outperforms significantly the market.
30:48But what the big question is is, like, when do you buy? When do you sell? Yeah.
30:52Sure. You can buy at IPO day. Like, StubHub is Neffel.
30:58It they just went public last year, so I bought some. Yeah. It hasn't been good so far, but it's like, I'm gonna hold it forever, so whatever.
31:05Yeah. But, like, if you aren't able to buy at I IPO day, which you're not usually, when do you buy? When do you sell?
31:11So I would be curious to do this test Yeah. On a more specific thesis. Yeah.
31:15Same thing. Here's a thousand dollars, five models competing against each other. Don't be afraid to be risky, but only buy Neffel stocks.
31:22And it's a range too because, like, Google, like, Sergei and and Larry Page, they're, like, on the board, and they're influential. Amazon, Jeff Bezos is influential. Yeah.
31:31So it's it's not founder led, but it kinda is. Yeah. Yeah.
31:34So I also have a, like, a one to 10 scale for both of those two variables. Right? Yeah.
31:38So it's like, here's the list of companies that because, like, Facebook would be 10 out of 10 on both. Yeah. Right?
34:32um, but it's not always the most exciting topic, but this is pretty exciting to them because I'm like, hey. Look at Claude.
34:39Oh, look at Chad. Look at oh, look what Google messed us up with, you know, Gemini. So, yeah, they are a lot more excited to hear the weekly updates
34:49Yeah. Than than I think they otherwise would have been. Well, for my kids, I would love to have them do this.
34:54I will give them a $100. Yeah. They can split it up between the five models or whatever.
34:58That way they can learn AI. They can learn prompting, and they can learn investing all in one. Yeah.
35:07Yeah. Totally should. It's always good to charge your kids interest.
35:10Make them learn how the real world works. Clip that. Yeah.
35:15Okay. So to recap, it's been six months. It's not over yet.
35:19Theoretically, you still could lose all your money. Yeah. But the market's been fairly flat, at least, you know, in this market, 5% is is fairly flat.
35:27Yeah. But you've outperformed by 10 to 30 x. Yeah.
35:31that the next six months, the market is gonna go up more than it has. Mhmm.
35:36And I think that these holdings that they're in currently at least will even do more than what the market will do over the next six months too. So it'll be interesting.
35:46I could do, like, a year update with you to see what happens one year in or who won because, uh, yeah, I think it's gonna keep I think they're doing a good job. I think it's gonna keep going up. Yeah.
35:56I recorded another video like this where I did the same test but with one day expiring options, so we'll link to that. Well, Brandon, where can we find you? Brandon Doyle on on X and something like that on Instagram.
36:08And, uh, yeah, you can, uh, look us up on, uh, tkowners.com too if you wanna join a community with a bunch of entrepreneurs.
36:19Yeah. Thanks. If you like this content, please share with a friend, hit a like or a comment below, and we'll see you next time on the Kerner office.
The Hook
The bait, then the rug-pull.
The title promises AI models fighting for their digital lives; the cold open delivers the receipts first — Claude has already more than doubled the money it was handed, and Gemini has turned into something closer to a losing gambler. Only after that flash-forward does the episode rewind to explain how a $5,000 real-money bet across five AI models actually got set up.
A 10-minute breakdown of why a better, cheaper AI model being locked behind 20 government-selected companies is a turning point — and what to do before the window closes.
A 17-minute career roadmap arguing that the next move for anyone who can build with AI is to stop being a builder and start being a consultant — with a four-step playbook to do it without quitting your job.
A 17-minute tutorial on building a live FP&A command center inside Claude Cowork — seven analytical layers, three connected sources, zero automation middleware.
A 16-minute numbered-rules breakdown from a man who has done $57.9M launches and consulted billion-dollar companies — no sponsor, no filler, just nine earned principles.
Theo breaks down how Anthropic silently modified prompts, rewrote its system card, and built invisible safeguards into its most capable model - then got caught.
00:30
01:27
01:56
02:32
02:45
03:20
03:43
04:12
05:04
05:35
06:01
06:25
06:59
07:23
07:55
08:12
08:39
09:24
09:41
10:02
11:04
11:25
11:59
12:25
12:52
13:14
13:40
14:11
15:05
15:41
16:01
16:37
17:22
17:51
18:48
19:10
19:45
20:31
20:57
21:28
21:54
22:22
23:09
23:58
24:12
24:49
25:09
25:33
26:30
26:52
27:28
27:56
28:34
28:54
29:09
29:41
30:43
31:01
31:45
31:56
32:22
32:47
33:18
33:50
34:11
34:39
35:15
36:02
10:32
16:53
17:10
20:08
16:27
29:32