Offshore
Video
EndGame Macro
Warren Buffet’s Quiet Framework for Not Waking Up With Regret
What Buffett is really passing along is a way to stop drifting. Not in some dramatic, motivational way, but in a very practical one. Instead of asking, “What should I do next?” he’s saying to ask, “How would I want to be remembered if this all stopped tomorrow?” Then work backward from that. Not the polished obituary. The real one. The one written by the people who actually knew you.
Once you do that, a lot of choices get clearer. Who you spend time with. What kind of work you tolerate or refuse to tolerate. How you treat your family when you’re tired. Whether you’re building something meaningful or just staying busy. Charlie’s point was that life compounds in the direction of your daily habits, not your stated intentions. If you want to be seen as generous, steady, curious, dependable then you have to practice those things now, quietly, when no one’s watching.
The career advice fits the same frame. Buffett isn’t saying chase your passion and everything works out. He’s saying pay attention to what holds your interest without forcing it. If you haven’t found that yet, keep looking but don’t forget that the goal isn’t status or money for its own sake. It’s work that fits who you are, done alongside people you respect. You’ll still make mistakes. You’ll still hit rough patches. But if you’re aiming at the right destination, the setbacks don’t feel like failure, they feel like part of the path.
The deeper message is simple and uncomfortable: no one fixes this later for you. There’s no separate chapter where you suddenly become the person you admire. You become that person by how you choose today and who you learn from, who you marry, how you work, how you show up. Do that long enough, and the ending tends to take care of itself.
tweet
Warren Buffet’s Quiet Framework for Not Waking Up With Regret
What Buffett is really passing along is a way to stop drifting. Not in some dramatic, motivational way, but in a very practical one. Instead of asking, “What should I do next?” he’s saying to ask, “How would I want to be remembered if this all stopped tomorrow?” Then work backward from that. Not the polished obituary. The real one. The one written by the people who actually knew you.
Once you do that, a lot of choices get clearer. Who you spend time with. What kind of work you tolerate or refuse to tolerate. How you treat your family when you’re tired. Whether you’re building something meaningful or just staying busy. Charlie’s point was that life compounds in the direction of your daily habits, not your stated intentions. If you want to be seen as generous, steady, curious, dependable then you have to practice those things now, quietly, when no one’s watching.
The career advice fits the same frame. Buffett isn’t saying chase your passion and everything works out. He’s saying pay attention to what holds your interest without forcing it. If you haven’t found that yet, keep looking but don’t forget that the goal isn’t status or money for its own sake. It’s work that fits who you are, done alongside people you respect. You’ll still make mistakes. You’ll still hit rough patches. But if you’re aiming at the right destination, the setbacks don’t feel like failure, they feel like part of the path.
The deeper message is simple and uncomfortable: no one fixes this later for you. There’s no separate chapter where you suddenly become the person you admire. You become that person by how you choose today and who you learn from, who you marry, how you work, how you show up. Do that long enough, and the ending tends to take care of itself.
tweet
Offshore
Photo
EndGame Macro
The Two Economy Consumer And Why U.S. Spending Looks Strong While Most Households Feel Broke
This chart is about five very different consumers living in the same economy and only one of them is keeping the numbers looking okay. That’s why you can have spending that doesn’t collapse while confidence is sliding. The Conference Board’s Consumer Confidence Index is down to 89.1 in December 2025, and sentiment has fallen for five straight months, even as top end spending props up the aggregates. Roughly 67% of Americans still say they’re living paycheck to paycheck, which tells you the floor is thin even if the ceiling is doing fine.
Start at the bottom 20%. This group is basically out of the discretionary economy. They account for under 10% of total spending and only a tiny slice of new vehicle purchases. That’s affordability problem. Essentials have eaten the budget: groceries are up nearly 25% since 2020, and cumulative inflation plus new tariffs have kept pressure on day to day goods. Add the credit stress and it’s easy to see why they can’t participate in big purchases: with student loan delinquency reporting resumed and roughly 9.4%–10.2% of total student debt is now 90+ days delinquent or in default, and subprime auto delinquencies 60+ days are around 6.65%.
Move up to the 20–40% bracket. This group still participates, but cautiously. They’re the ones cutting frequency, trading down, delaying replacements, and leaning on used goods. And it shows up in the quiet stress metrics where credit card delinquencies may look stable at the 30 day level 2.98% in Q3 2025 but severe delinquencies 90+ days are still elevated around 12%, which is usually the cleaner sign that households are losing flexibility, not just reshuffling bills.
The 40–60% bracket is where the story gets deceptive. On paper, it looks like the middle is holding up. In reality, a lot of that spending is inertia, maintaining existing lifestyles rather than upgrading. Higher rates changed what’s possible. The average new vehicle price is now over $50,000 versus roughly $37,401 in 2019, and the average monthly payment has moved from about $550 to roughly $748–$754. Nearly 20% of new car buyers are carrying payments of $1,000+ a month. That’s a financing strain signal wearing a shiny new car sticker.
The 60–80% bracket is still functional, but behavior has changed. Even here, people are more value conscious and more sensitive to job security and asset prices. They can spend, but they don’t feel as good doing it which is why sentiment keeps slipping even when the data doesn’t look catastrophic.
Then there’s the top 20%. This is the support beam. They account for roughly 39% of all goods and services spending, and about 55% of new vehicle purchases. In other words, the consumer holding up is mostly the top end holding up. That’s the heart of the two economy problem where aggregate consumption can look fine because the top end is fine, while the broad base is squeezed, anxious, and increasingly locked out of the durable goods economy.
The risk is that the system is becoming narrow and fragile. When spending depends heavily on the top 20–40%, it looks stable right up until markets wobble, bonuses shrink, layoffs hit higher income sectors, or credit tightens again. And when that support beam cracks, the slowdown doesn’t arrive gently, it accelerates.
US consumer spending is extremely concentrated among high-income households:
Households in the top 20% of the income distribution now reflect ~39% of all spending on goods and services.
Their concentration is even higher in new vehicle purchases, where they make up ~55% of total spend.
This is followed by households in the 60% to 80% income bracket, which represent ~23% of total spending and ~21% of new vehicle purchases.
As a result, nearly 8 out of 10 new vehicles are bought by the top 40% of households.
Meanwhile, the bottom 20% contribute just 9% of total spending and a mere 4% of new vehicle demand.
The [...]
The Two Economy Consumer And Why U.S. Spending Looks Strong While Most Households Feel Broke
This chart is about five very different consumers living in the same economy and only one of them is keeping the numbers looking okay. That’s why you can have spending that doesn’t collapse while confidence is sliding. The Conference Board’s Consumer Confidence Index is down to 89.1 in December 2025, and sentiment has fallen for five straight months, even as top end spending props up the aggregates. Roughly 67% of Americans still say they’re living paycheck to paycheck, which tells you the floor is thin even if the ceiling is doing fine.
Start at the bottom 20%. This group is basically out of the discretionary economy. They account for under 10% of total spending and only a tiny slice of new vehicle purchases. That’s affordability problem. Essentials have eaten the budget: groceries are up nearly 25% since 2020, and cumulative inflation plus new tariffs have kept pressure on day to day goods. Add the credit stress and it’s easy to see why they can’t participate in big purchases: with student loan delinquency reporting resumed and roughly 9.4%–10.2% of total student debt is now 90+ days delinquent or in default, and subprime auto delinquencies 60+ days are around 6.65%.
Move up to the 20–40% bracket. This group still participates, but cautiously. They’re the ones cutting frequency, trading down, delaying replacements, and leaning on used goods. And it shows up in the quiet stress metrics where credit card delinquencies may look stable at the 30 day level 2.98% in Q3 2025 but severe delinquencies 90+ days are still elevated around 12%, which is usually the cleaner sign that households are losing flexibility, not just reshuffling bills.
The 40–60% bracket is where the story gets deceptive. On paper, it looks like the middle is holding up. In reality, a lot of that spending is inertia, maintaining existing lifestyles rather than upgrading. Higher rates changed what’s possible. The average new vehicle price is now over $50,000 versus roughly $37,401 in 2019, and the average monthly payment has moved from about $550 to roughly $748–$754. Nearly 20% of new car buyers are carrying payments of $1,000+ a month. That’s a financing strain signal wearing a shiny new car sticker.
The 60–80% bracket is still functional, but behavior has changed. Even here, people are more value conscious and more sensitive to job security and asset prices. They can spend, but they don’t feel as good doing it which is why sentiment keeps slipping even when the data doesn’t look catastrophic.
Then there’s the top 20%. This is the support beam. They account for roughly 39% of all goods and services spending, and about 55% of new vehicle purchases. In other words, the consumer holding up is mostly the top end holding up. That’s the heart of the two economy problem where aggregate consumption can look fine because the top end is fine, while the broad base is squeezed, anxious, and increasingly locked out of the durable goods economy.
The risk is that the system is becoming narrow and fragile. When spending depends heavily on the top 20–40%, it looks stable right up until markets wobble, bonuses shrink, layoffs hit higher income sectors, or credit tightens again. And when that support beam cracks, the slowdown doesn’t arrive gently, it accelerates.
US consumer spending is extremely concentrated among high-income households:
Households in the top 20% of the income distribution now reflect ~39% of all spending on goods and services.
Their concentration is even higher in new vehicle purchases, where they make up ~55% of total spend.
This is followed by households in the 60% to 80% income bracket, which represent ~23% of total spending and ~21% of new vehicle purchases.
As a result, nearly 8 out of 10 new vehicles are bought by the top 40% of households.
Meanwhile, the bottom 20% contribute just 9% of total spending and a mere 4% of new vehicle demand.
The [...]
Offshore
EndGame Macro The Two Economy Consumer And Why U.S. Spending Looks Strong While Most Households Feel Broke This chart is about five very different consumers living in the same economy and only one of them is keeping the numbers looking okay. That’s why you…
US wealth divide has never been larger. - The Kobeissi Letter tweet
Offshore
Photo
EndGame Macro
The Big Stay Is the Loudest Signal in the Job Market Right Now
The quits rate is a confidence gauge. When people feel good about the job market, they quit because they believe something better is waiting. When quits fall, it means workers are staying put not because they’re happy, but because the outside options feel uncertain or risky. A private sector quits rate around 2% and total quits near 1.8% is a big psychological shift. That’s not a healthy, dynamic labor market, it’s a cautious one. People stop moving first. Everything else follows later.
Low Hire, Low Fire Is How Slowdowns Actually Start
What we’re in now is a quiet freeze. Hiring slows. Job openings fade. Wage leverage disappears. Employers don’t need to fire aggressively to tighten conditions; they just stop adding headcount and let time do the work. That’s why unemployment can stay okay on the surface while the labor market underneath weakens. Historically, this is how downturns incubate…not with a bang, but with fewer opportunities and less mobility.
The Mirage of Job Openings
The rise of fake or ghost job listings fits perfectly into this phase. Companies keep postings up to look healthy, calm overworked employees, and preserve optionality and not because they’re actively hiring. To job seekers, it feels like plenty of demand. In reality, the funnel narrows fast. Fewer real offers, longer searches, and more people deciding it’s safer to stay where they are. That’s exactly what falling quits reflect.
My View
This is a confidence collapse in slow motion. Workers sense the shift before the data turns ugly. They stop quitting. Employers quietly pull back. The illusion holds for a while, but the flexibility is gone. Once quits fall this far, history says the next phase is weaker wage growth, softer consumption, and eventually higher unemployment.
tweet
The Big Stay Is the Loudest Signal in the Job Market Right Now
The quits rate is a confidence gauge. When people feel good about the job market, they quit because they believe something better is waiting. When quits fall, it means workers are staying put not because they’re happy, but because the outside options feel uncertain or risky. A private sector quits rate around 2% and total quits near 1.8% is a big psychological shift. That’s not a healthy, dynamic labor market, it’s a cautious one. People stop moving first. Everything else follows later.
Low Hire, Low Fire Is How Slowdowns Actually Start
What we’re in now is a quiet freeze. Hiring slows. Job openings fade. Wage leverage disappears. Employers don’t need to fire aggressively to tighten conditions; they just stop adding headcount and let time do the work. That’s why unemployment can stay okay on the surface while the labor market underneath weakens. Historically, this is how downturns incubate…not with a bang, but with fewer opportunities and less mobility.
The Mirage of Job Openings
The rise of fake or ghost job listings fits perfectly into this phase. Companies keep postings up to look healthy, calm overworked employees, and preserve optionality and not because they’re actively hiring. To job seekers, it feels like plenty of demand. In reality, the funnel narrows fast. Fewer real offers, longer searches, and more people deciding it’s safer to stay where they are. That’s exactly what falling quits reflect.
My View
This is a confidence collapse in slow motion. Workers sense the shift before the data turns ugly. They stop quitting. Employers quietly pull back. The illusion holds for a while, but the flexibility is gone. Once quits fall this far, history says the next phase is weaker wage growth, softer consumption, and eventually higher unemployment.
More evidence of a weakening US labor market:
The quits rate in the private sector declined to 2.0% in October, the lowest since the 2020 pandemic.
This measures the proportion of private workers voluntarily quitting their jobs.
At the same time, the total nonfarm quits rate fell to 1.8%, the lowest since May 2020.
Both metrics are now consistent with levels seen during in midst of the 2008 Financial Crisis.
For perspective, private and total quits rates during the 2001 recession were higher at 2.5% and 2.2%, respectively.
As hiring slows, fewer job opportunities are present, leading to a reduced number of workers quitting.
The US job market needs rate cuts. tweet
Offshore
Photo
EndGame Macro
Truck Sales Are Quietly Warning the Market
Truck sales are weak because fleets don’t trust demand or credit conditions, but labor supply is quietly tightening at the same time. The recent rule changes around non domiciled CDLs matter here. By narrowing eligibility and pushing enforcement even amid legal uncertainty, thousands of drivers and potentially far more over time are being forced out of the industry. That doesn’t show up immediately as higher sales or higher wages. It shows up as hesitation. Carriers delay decisions because the operating environment just got less predictable.
Cyclical Weakness Meets Policy Driven Friction
On the cyclical side, freight demand cooled after the post COVID boom, rates fell, and credit tightened. That alone explains a lot of the collapse in truck orders. But policy is now adding friction at exactly the wrong moment. Pulling drivers out of the labor pool doesn’t revive freight demand; it raises uncertainty around capacity, compliance, and costs. Fleets are left asking whether they’ll have the drivers they need 6 months from now, what wages will look like, and whether financing even makes sense if utilization stays low. When business confidence is already fragile, that kind of uncertainty freezes capital spending.
Where This Puts Us In The Chain
Freight usually weakens first, then credit, then labor. We’re in the uncomfortable middle. Demand is soft, credit is restrictive, and labor policy is adding stress instead of relief. The result isn’t a clean recession signal yet, services are still carrying the broader economy but it is a setup for delayed pain. If demand doesn’t recover, carriers won’t order trucks. If carriers don’t order, manufacturers cut back. If labor availability tightens while volumes stay weak, margins get squeezed and layoffs follow. That’s how a slow grind turns into a sharper downturn.
My View
This isn’t just about trucks. It’s about confidence. The freight economy runs on long dated decisions in financing, hiring, equipment and routes. When demand weakens and policy introduces uncertainty around labor at the same time, businesses choose caution. That’s why this signal matters. Not because it guarantees a recession tomorrow, but because it shows an economy where growth is losing its footing and policy choices are amplifying stress instead of cushioning it.
tweet
Truck Sales Are Quietly Warning the Market
Truck sales are weak because fleets don’t trust demand or credit conditions, but labor supply is quietly tightening at the same time. The recent rule changes around non domiciled CDLs matter here. By narrowing eligibility and pushing enforcement even amid legal uncertainty, thousands of drivers and potentially far more over time are being forced out of the industry. That doesn’t show up immediately as higher sales or higher wages. It shows up as hesitation. Carriers delay decisions because the operating environment just got less predictable.
Cyclical Weakness Meets Policy Driven Friction
On the cyclical side, freight demand cooled after the post COVID boom, rates fell, and credit tightened. That alone explains a lot of the collapse in truck orders. But policy is now adding friction at exactly the wrong moment. Pulling drivers out of the labor pool doesn’t revive freight demand; it raises uncertainty around capacity, compliance, and costs. Fleets are left asking whether they’ll have the drivers they need 6 months from now, what wages will look like, and whether financing even makes sense if utilization stays low. When business confidence is already fragile, that kind of uncertainty freezes capital spending.
Where This Puts Us In The Chain
Freight usually weakens first, then credit, then labor. We’re in the uncomfortable middle. Demand is soft, credit is restrictive, and labor policy is adding stress instead of relief. The result isn’t a clean recession signal yet, services are still carrying the broader economy but it is a setup for delayed pain. If demand doesn’t recover, carriers won’t order trucks. If carriers don’t order, manufacturers cut back. If labor availability tightens while volumes stay weak, margins get squeezed and layoffs follow. That’s how a slow grind turns into a sharper downturn.
My View
This isn’t just about trucks. It’s about confidence. The freight economy runs on long dated decisions in financing, hiring, equipment and routes. When demand weakens and policy introduces uncertainty around labor at the same time, businesses choose caution. That’s why this signal matters. Not because it guarantees a recession tomorrow, but because it shows an economy where growth is losing its footing and policy choices are amplifying stress instead of cushioning it.
⚠️Plummeting US heavy truck sales point to a spike in the unemployment rate:
Heavy truck sales dropped -119,000 from June to November, to a 336,000 annual rate, the lowest since the 2020 Crisis.
Historically, such a rapid decline has rarely occurred outside of a recession.
It has also served as a leading indicator for US unemployment.
This now signals at least a +1.0 percentage point increase in the unemployment rate, to ~5.5%, over the next 6–9 months.
Is the US economy in a recession? tweet
AkhenOsiris
RT @buccocapital: Absolutely fascinating and wonderful response from @benedictevans about where work will be done: ChatGPT/Gemini/etc or SaaS
From @BenBajarin’s pod. Long but VERY worth it!
“Well, there's maybe two answers to this. One of them is, you know, think of a use case like, you know, take a photograph of a recipe and then take a photograph of your fridge and say, hey, use Instacart to order whatever I don't have. Now, these systems can kind of do that now.
They might get something wrong. You might have to check, but they'll basically be able to do that. Is that something that you should do in ChatGPT or in Instacart?
Where should you, what should the entry path to that be? How should that work? Anyhow, again, it kind of comes back to this phrase, the thin wrapper.
So, that's its scenario. Instacart would be like a thin wrapper. Of course, the real thin wrappers are all the apps that the model apps put out.
ChatGPT is a thin wrapper. It's an input box and an output box and nothing else. But I remember, as both of you know, Stephen Sanofsky, I worked with at A16Z, and he was sort of pointing out that in the mid to late 90s, people were saying this about Windows, that the operating system is...
Because before we're operating “before GUIs, like word processor, I had to do printing and graphics and file storage and everything else. And now with Windows, the operating system was doing all of that. So what was left for the application?
Well, it was just like a thin wrapper on some Win32 APIs. Basically, all the hard stuff was being done by the operating system, and Microsoft Office is just like a thin Win32 wrapper. And which is kind of technically true if you're Bill Gates, but kind of not true if you're a user.
And so you've got this kind of question of like, how generalized and how specific do the use cases become? How much is it that you ask this thing to do anything for you? And I think there's a kind of fascinating kind of challenge in here, because, actually going back to Bill Gates, he was talking about this like two, three years ago.
He was saying this is the biggest thing since the GUI. Because with the GUI, you didn't need to memorize keyboard commands, so you have this massive expansion in what software can be. But someone still needs to have made that piece of software.
So “if you want to do your taxes on a computer, someone needs to have made a piece of tax software. Whereas in principle, I could just ask ChatGPT, hey, just go do my taxes for me. And so you could have this huge expansion in how much stuff could be done in a software without needing individual pieces of software.
The problem is, most of those use cases, the user doesn't think like that. So you kind of do need that piece of software still.
I mean, I had a conversation with the CIO at a big retailer a couple of weeks ago, and it kind of occurred to me that like, there's all this question like, should I do this in Oracle, or should I export it into a CSV and do it in Excel? Or should I do it in a SaaS app?
To solve this problem generically, is it best done using the general purpose tools in Oracle, SAP or Salesforce or something? Or in a vertical SaaS app that unbundles like graduate recruit hiring or something, or accounts payable reconciliation into some specific SaaS application, which is why every big company has four or five hundred SaaS apps. They're all unbundling SAP or Oracle.
Or the third case is, do I export a CSV and do it in Excel? And maybe now what ChatGPT does is it's like it's a false option on that scale. It will enable massively more SaaS apps because now there will be this whole class of stuff you can make a piece of software to do that you couldn't do before.
But there will also be this question of like, should I do that in Excel? Should I do that in ChatGPT? Should I [...]
RT @buccocapital: Absolutely fascinating and wonderful response from @benedictevans about where work will be done: ChatGPT/Gemini/etc or SaaS
From @BenBajarin’s pod. Long but VERY worth it!
“Well, there's maybe two answers to this. One of them is, you know, think of a use case like, you know, take a photograph of a recipe and then take a photograph of your fridge and say, hey, use Instacart to order whatever I don't have. Now, these systems can kind of do that now.
They might get something wrong. You might have to check, but they'll basically be able to do that. Is that something that you should do in ChatGPT or in Instacart?
Where should you, what should the entry path to that be? How should that work? Anyhow, again, it kind of comes back to this phrase, the thin wrapper.
So, that's its scenario. Instacart would be like a thin wrapper. Of course, the real thin wrappers are all the apps that the model apps put out.
ChatGPT is a thin wrapper. It's an input box and an output box and nothing else. But I remember, as both of you know, Stephen Sanofsky, I worked with at A16Z, and he was sort of pointing out that in the mid to late 90s, people were saying this about Windows, that the operating system is...
Because before we're operating “before GUIs, like word processor, I had to do printing and graphics and file storage and everything else. And now with Windows, the operating system was doing all of that. So what was left for the application?
Well, it was just like a thin wrapper on some Win32 APIs. Basically, all the hard stuff was being done by the operating system, and Microsoft Office is just like a thin Win32 wrapper. And which is kind of technically true if you're Bill Gates, but kind of not true if you're a user.
And so you've got this kind of question of like, how generalized and how specific do the use cases become? How much is it that you ask this thing to do anything for you? And I think there's a kind of fascinating kind of challenge in here, because, actually going back to Bill Gates, he was talking about this like two, three years ago.
He was saying this is the biggest thing since the GUI. Because with the GUI, you didn't need to memorize keyboard commands, so you have this massive expansion in what software can be. But someone still needs to have made that piece of software.
So “if you want to do your taxes on a computer, someone needs to have made a piece of tax software. Whereas in principle, I could just ask ChatGPT, hey, just go do my taxes for me. And so you could have this huge expansion in how much stuff could be done in a software without needing individual pieces of software.
The problem is, most of those use cases, the user doesn't think like that. So you kind of do need that piece of software still.
I mean, I had a conversation with the CIO at a big retailer a couple of weeks ago, and it kind of occurred to me that like, there's all this question like, should I do this in Oracle, or should I export it into a CSV and do it in Excel? Or should I do it in a SaaS app?
To solve this problem generically, is it best done using the general purpose tools in Oracle, SAP or Salesforce or something? Or in a vertical SaaS app that unbundles like graduate recruit hiring or something, or accounts payable reconciliation into some specific SaaS application, which is why every big company has four or five hundred SaaS apps. They're all unbundling SAP or Oracle.
Or the third case is, do I export a CSV and do it in Excel? And maybe now what ChatGPT does is it's like it's a false option on that scale. It will enable massively more SaaS apps because now there will be this whole class of stuff you can make a piece of software to do that you couldn't do before.
But there will also be this question of like, should I do that in Excel? Should I do that in ChatGPT? Should I [...]
Offshore
AkhenOsiris RT @buccocapital: Absolutely fascinating and wonderful response from @benedictevans about where work will be done: ChatGPT/Gemini/etc or SaaS From @BenBajarin’s pod. Long but VERY worth it! “Well, there's maybe two answers to this. One of them…
do it in the dedicated vertical app?
Should I do it in the horizontal app? And I don't think the answer is you'll just do all of it in the dedicated, in ChatGPT. And I almost feel like it's like, I feel like it's weird to even have to say that.
And then there's the two hour podcast where we talk about how we'll have human-level PhD researchers next year. Yes. So there's a very weird split.
I mean, I've said this a bunch. It's like, the last OpenAI, one of the last OpenAI videos, they spend the first 20 minutes “saying we're going to have human-level AI researchers.” And then they say,” oh, also, we're going to have thousands of ISVs.”
Like, well, which is it? Either you can have the software as a person, or you're going to have loads of independent vertical pieces of software, but not both.
I think within OpenAI, that's the tensions really coming back to front-facing product plus APIs, right? Because I use a bunch of SaaS software that integrates ChatGPT as something that will just work with the data sets or help you build a model.
And it's just using, right? It's just a wrapper. It's just calling APIs for their tokens.
But then I also use GPT as a front-end for a handful of things, right? And so that's the continued tension. Gemini, Google will be in the same thing.
There's APIs. You can use it on your back-end SaaS, or you can use it as a front-end for X, Y, and Z, I guess, right? There's the two sides of their model.”
tweet
Should I do it in the horizontal app? And I don't think the answer is you'll just do all of it in the dedicated, in ChatGPT. And I almost feel like it's like, I feel like it's weird to even have to say that.
And then there's the two hour podcast where we talk about how we'll have human-level PhD researchers next year. Yes. So there's a very weird split.
I mean, I've said this a bunch. It's like, the last OpenAI, one of the last OpenAI videos, they spend the first 20 minutes “saying we're going to have human-level AI researchers.” And then they say,” oh, also, we're going to have thousands of ISVs.”
Like, well, which is it? Either you can have the software as a person, or you're going to have loads of independent vertical pieces of software, but not both.
I think within OpenAI, that's the tensions really coming back to front-facing product plus APIs, right? Because I use a bunch of SaaS software that integrates ChatGPT as something that will just work with the data sets or help you build a model.
And it's just using, right? It's just a wrapper. It's just calling APIs for their tokens.
But then I also use GPT as a front-end for a handful of things, right? And so that's the continued tension. Gemini, Google will be in the same thing.
There's APIs. You can use it on your back-end SaaS, or you can use it as a front-end for X, Y, and Z, I guess, right? There's the two sides of their model.”
tweet
Offshore
Photo
AkhenOsiris
RT @stockthoughts81: Sharing this for my extended family. A 15-year-old has been battling Acute Myeloid Leukemia (AML) since August and the road ahead is heavy. If you’re able to donate or even just RT, thank you. https://t.co/QPjXitKp0q https://t.co/Ylvediys3q
tweet
RT @stockthoughts81: Sharing this for my extended family. A 15-year-old has been battling Acute Myeloid Leukemia (AML) since August and the road ahead is heavy. If you’re able to donate or even just RT, thank you. https://t.co/QPjXitKp0q https://t.co/Ylvediys3q
tweet
Offshore
Video
memenodes
LeBron’s opening line being…”First of all, happy…happy International Women’s Day” cracks me up EVERY time https://t.co/MjjungCCbt
tweet
LeBron’s opening line being…”First of all, happy…happy International Women’s Day” cracks me up EVERY time https://t.co/MjjungCCbt
What’s the best LeBron Moment tweet