OpenAI Just Made the Biggest AI Cost Debate Irrelevant

Let me show you how you can save money on token costs, Neil. So this is Cognition. Okay. Cognition, Devin, you've heard of Devin? They bought Windsurf.

Yeah. Okay.

So this is a little bar over here. Adaptive automatically balances quality and cost of your token outputs, right? So they have all this stuff over here. This is like one of the models that you can select here. And I believe a lot of people are going to have this adaptive thing where you can choose between the big ones and the small ones, and I'll show you the output. So I think they came up with a couple of innovating things here. So what this guy said, he works at Cognition, he said token maxing is Dead. Everyone realized that token usage is a horrible way to measure productivity. So where do we go from here? So here are some ways they're addressing this at Cognition, Neil, and this will save you money. Watch. Number 1, adaptive routing, an intelligent model router that automatically selects the best AI model for each task. Other products are starting to adapt this. 4, you'll like 4. They have an AI productivity guarantee. If Devin AI delivers less engineering value than you're paying for, Cognition will fund your usage until it does, up to $10 million for enterprise customers.

Okay, you might like that.

Now, number 3, Advanced automations, wiring agent sessions to the events that already cost teams engineering hours, production incidents and pager duty, failing deploys, alerts on a revenue-critical service. Each trigger maps to work that used to page a human, so the value isn't theoretical. Spend attribution as well. So Neil, these— you would like all these. Spend attribution, adaptive routing, AI productivity guarantee. You guys should check out Cognition.

Yeah, we could be using it already. I don't know what we're using or not using.

Bet you that you're not. I'll bet you as a Spurs bet that you're not.

That we're not using cognition.

I don't think you're using cognition.

You want to find out right now?

Yeah, let's find out, but don't bet me.

I can actually just ask the team. You should you should ask, but don't bet me. Where is it? Dude, I have so many threads on WhatsApp. Let's type in Uber Suggest.

So as Neil's pulling it up right now, I am going to pull up. this concept of the forward-deployed marketer. So jump in at any point, Neil.

Yeah, you've—

okay.

My team will take a while because I just asked them in WhatsApp. I don't— it says 2 people are online, but I don't know if it's the right 2 people that are online.

That's okay. We'll find out. So I think, Neil, this concept of the forward-deployed marketer is going to be the most valuable hire in marketing right now. So, and almost nobody's staffing for it, right? So I think your company and my company, we're going to need to hire smart marketing strategists, okay, that know how to upgrade the scaffolding for these agents. And if you don't, if you're not an AI-pilled marketer right now, you can't do that. You wouldn't know what to do, right? Just because you come with a lot of experience. And so the 4-deploy marketer owns the client outcome end-to-end, the agents will do the labor underneath. And it's interesting, because a lot of the AI-pilled— I talked to a guy today, I'm like, how much do you spend a month on AI, your own money? I'm like, don't talk about what the company pays. He's like, $1,500 a month. I was like, fantastic, right? That's $18,000 of his own money a year. I love that, right? Yeah. And he's like, look, I didn't want to talk to you initially because my friend told me this is an agency. I was like, well, then I showed him everything we're building. He's like, I'm interested now, right? And so if you want to get these people, you have to show them the way of the future is not what you're going to do this agency thing. You have to show them the model is you're going to be building and strategizing and you're going to reduce the number of client calls. You're going to reduce the time you spend on reporting and you're going to have a lot of fun building. You're going to be a builder, right? And you have to be able to sell that vision. And so to get these 4D Play marketers, you, you, You have to almost convince them to do it because I think a lot of these people don't even know that they're forward-deployed marketers now. And when I say forward-deployed, yesterday I got on a demo call for a product and he's like, the guy was like, I'm a forward-deployed engineer, right? And so I'm like, he's basically doing like customer support and he's like helping us deploy, right? I think this is a sense of like the new version of a marketing agency client services person. That's all. It's just a new word.

Yeah, people are always going to use new words for all this.

Growth hacker.

Yeah, the way I see it is you're still a marketer, you just got to keep being valuable Yeah, yeah.

Here's the other thing, like, Neil, you hear people that say, you know, should I focus on the work right now or should I do AI? And I just think that's a fallacy.

Yes, you should be using AI that helps you with the work and getting you the results. If you're using AI and it doesn't drive any of the main KPIs for the business, then who the heck cares what you're doing, whether it's AI or playing catch or watching TV?

Yeah, like I see, so we have a new guy in sales, Sales right now, right? And then he actually, because he doesn't come with— he reminds me of your guy who started in paid. So similar situation here, right? And he doesn't come with a lot of the bad habits of a salesperson from the past, like whether it's a good thing or bad thing, and he's very trainable. So he has all these workflows that he's built now that will automatically follow up, that will automatically, you know, if a deal needs reviving or a deal is stalled out or whatever, he's just doing— it's natural to him. It's natural for him to use skills, right? But Whereas in the past, like there's a salesperson, you know, that we had that you met before, that person didn't want to do these things, right? And so I don't care how much experience you have, if you're not going to do these things, you're just going to get laughed. And I talked to a guy today, this Matt Van Horn, I think he was part of the Lyft founding team, or maybe he founded Lyft, the driving autonomy company, right? So we did like a— because he's created all these skills.

Oh, not the Uber competitor?

The Uber competitor.

Oh, okay.

Yeah. And so he has all these things now that has like millions of views, right, on AI stuff. And I was like, look, And I want to ask you this question, okay? Neil today with AI versus Neil a year ago, if that Neil from a year ago tries to catch up right now, can he catch Neil today? Same resources.

Me particularly, or just generally a person in my company?

Generally.

Generally, no, it would be too hard.

Because compounding.

Correct.

Right?

Yeah. With me specifically, the reason I asked that, and I know why Eric said generally, is because I spend most of my time doing deals. So AI is cool. It is useful, extremely useful, but it doesn't move the needle as much as relationships for me.

And what Neil is saying is the question I ask myself every day and the question you all should be asking yourself, and this is the question Neil's asked himself every day, what is the highest leverage thing I can do with my time today? What is where I can put $1 in where I can get $1,000 back with my time?

Yes.

And that's a different answer for every person depending on the stage that you're in right now.

Yeah. What is it for you? I know my order.

Mine's hiring right now.

Okay.

Yeah. And that changes over time, right? Mine's hiring and then there's, it's, there's products, so it's kind of tied, but like, it's, that's, that's what I think about the whole time. Yeah.

Mine in order, if I had to pick the 4, number 1 is deal-making, like doing business deals. I'm not talking about closing clients, like partnerships, things like that, that have much higher leverage than just closing a client. My number 2 is interacting with clients, making sure they're happy, upselling them. That's number 2. Number 3 is Conferences and meetings. And when I say meetings, it's kind of like speeches, sometimes smaller rooms, or maybe I just go to a corporation and just give a talk there. But I put conferences and those type of meetings in the same bucket because it's kind of like branding, new client development. And then my number 4th, I'm blanking on it and I don't know why. Oh, number 4 was content creation.

Yep.

So just mass cranking out tons of content every single month, and that helps keep the brand out there. But those are my 4 top priorities. AI could be used in a lot of them, but I would say for the role I'm in, it's not as effective as it would be for someone in the trenches more doing the work. And you could say, oh, Neil, well, you can use AI for the scripting for your content. Team already does everything. I just show up and I just read from a teleprompter based on the stuff they already know I talk about from like recordings from conferences or what I'm talking on internal calls and things like that. So I already know what I want to talk about and what I don't want to talk about. So they're taking a lot of my own content and just turning it into scripts for me. So from the aspect of AI, yes, I can use it and we do use it. It's just not as useful as someone maybe coding or running ads 24/7 and using AI to help with creatives or ad copy and stuff like that.

And again, I just want to emphasize to everyone, it depends on the stage of where you're at right now. So we're not saying what Neil's saying or what I'm saying is like, that's what you should be focusing on right now. It's different. And so I would say for me, it's, I have an acronym for this, TAT. Right? So TAT is tech-enabled services is one which ties in with SingleBrain and all the products that we have. A is AI fluency, and then T is talent, right? But those 2 kind of go hand in hand, talent and AI fluency. And then you have C, which is content creation. So I'm just focused on those areas right now. And obviously, you know, this is a wheel that keeps on spinning. Now, Neil's able to say that, hey, I'm going to focus on doing deals right now. One, he has a— he can get a lot of the phone calls or meetings that he needs to get. That's one, but that's through the content that he's been creating for the last, what, 21, 24 years or so? You have people leverage too. You have like, let's say 1,000 people or so, right? So you can have your team and you expect them to do that, right? And so just understand everyone, like if you're starting out right now, but the one thing, the one takeaway is just keep asking yourself, what's the highest leverage thing I can do with my time every day? I guarantee you, if you keep asking that question for the next 10 years, you're going to be in a good spot.

And it switches depending on the mode you are in for your business, right? When you're in a startup mode, it's very different than if you're, you know, in growth mode or if you're in a place where your business is already established and growth has slowed down and you're figuring out maybe spending your time on M&A or reinventing your business, kind of like, not Figma, Intercom.

Intercom too, yeah.

Intercom. I would say Intercom's a better example of really reinventing themselves with Find.

And by the way, if your business is dying like a newspaper business, maybe the highest leverage thing is trying to sell it.

Yes.

Yeah.

And I don't mean that in a, I'm not like laughing at them, but it is so true. And dude, I talked to so many publishers, like I was just at Web Summit in Rio, Because people from all around the world go to the web summits.

Mm-hmm.

And I meet so many publishers because these journalists interview me and they're just like, they always ask me questions about journalists getting displaced and things being changed. And I was just like, if you're a journalist, you need to reinvent your business. I'm like, one, you can sell your data. There's so many journalistic sites out there that have sold their data to these LLMs.

Mm-hmm.

Two, you can set up paywalls if you haven't, start charging for information. And I've gotten a lot of pushback. Why would people pay? I'm like, if people aren't willing to pay, That means your content isn't that good. If your content's really amazing, it's already been proven out there by newspapers that people will pay for the information. But I think they really struggle to change their business model because a lot of the people don't want to accept that they need to change and they just want it to more so go back to the way it was.

This is why it's never been easier to get ahead because most people don't want to change. And this is unfortunate. I do think the wealth gap is going to get bigger. But that's a side thing. Maybe we can come back to that. Neil, I want to ask you a question because we're on this question of high leverage. Okay, so you travel quite a bit. You do this thing, which I think sometimes doesn't make any sense, where you might fly into Hong Kong for like a day and then leave. And so if you were to make the highest leverage use of your time with these, this travel, what would you do differently? If anything?

It's, I wouldn't do anything. It's actually very well optimized.

You like it?

Do I like it or is it optimized for revenue generation?

Both.

It's optimized for revenue generation. I don't like it.

Yeah, because you don't even like, I believe, I do believe that it's optimized for revenue generation. I don't believe you enjoy it because You didn't even get the time to enjoy like the cool places you're going into for like a day or two. Dude.

So let's go back. We're recording on a Thursday. So not this Monday, but the previous Monday. Okay. From Monday, I flew out here sometime in the afternoon. I went to London, got there London in the sometime during the day.

You flew from LA to London? Uh-huh.

On a Monday, got there on Tuesday. Okay. Had to do my calls and meetings. And then a webinar that same day, because we usually do webinars early on Tuesday. And then from there, Wednesday morning, I speak, do some interviews, meetings, and then I fly out to Jakarta. So then London to Dubai, Dubai to Jakarta.

Brutal.

I get there at nighttime, late their time. The next morning I go in, I get a little bit of a rest. I'm jet-lagged too, so that was great. But in the afternoon, I had to go and speak. And then interviews, then dinner with a lot of local businesses, great networking as well, right? Trying to generate business. So remember I landed the day before, and when I say the day before, I'm talking about like nighttime. By the time I got into my hotel room, it was their version of a Friday.

Your time zone's all messed up.

Yeah. And then I'm speaking that same Friday and then I'm flying out. So I'm in Jakarta technically less than 24 hours. London, I think it was a little bit more than 24 hours, but close enough to— I think it was actually around 24 hours. Then from Jakarta, I go Jakarta back to Dubai, Dubai to Rio, jam-packed with meetings, speaking, and then fly right back out. And then I'm here all in roughly a week's time.

So here's a question for you. That's a lot, right? For anybody, that's a lot. It's a— it's— so if you were to optimize it, let's say you were to bring your kids for this, how would you re-optimize it?

It wouldn't work. They wouldn't be able to handle the time zone change.

Wouldn't you try to optimize it a little bit for them? Maybe stay a little longer in certain areas?

It's too tough because of schooling. If there was no schooling, I still couldn't do it because there's so many events that I need to go to that I can't just do them all in a short period of time. So like I came back, I think what, yesterday or day before? No, yesterday morning. I came back yesterday morning.

From where?

From Rio.

Okay.

We're recording today. Tomorrow, get ready and pack. Saturday, I leave. You have a cruise. I have a cruise, family vacation. So that, I'll give them the quality of the time that they're looking for. And then I get back from the cruise a day after, I head out to Medellín. And then I get there, I'm there for around 24 hours. I fly right back out.

That one's not that bad. Medellín, it's kind of far, but it's close.

It is because there's no direct flights.

There used to be, not anymore. There used to be a—

Yeah, I don't know where I had to go through, but the direct flights suck.

Oh no. Yeah, direct flight from Miami, not from here. Yeah, yeah, yeah.

Yeah, Miami has a lot of great direct flights to Latin America, but not LA.

That's why they call it America's Gateway. Quick break. One of our ClickFlow users just told me that they're saving 90+ hours a month on content alone. ClickFlow is our AI SEO product that actually does the work, not just here's what you should do. It does it. It creates content in your voice, it helps you with other areas such as finding the right internal links and creating FAQs. We spent years making it, perfecting it with our clients so it doesn't just sound like AI slop. You can get a 14-day free trial at clickflow.com. If you're serious about AI SEO, you should check it out. Back to the show. So anyway, I want to go back to the kind of the topic here and you can pick the next one. But the topic I was just talking to that guy Matt Van Horn today, just talking about how this concept of If you don't become AI fluent quickly, you're going to be part— that wealth gap's only going to get larger because one person's compounding, one person isn't compounding. And I just would rather prefer anybody listening to this right now be on the side that's compounding. That's all I would say. Yeah.

Dude, it is compounding. I think the issue isn't people putting in the work. I think that is part of the issue, but I think the bigger issue is People don't want to accept change in reality and they just want things to go back to the way that they were. You know, and—

You know, our parents do that, right? It's like, oh, life would've— was so better when we were younger.

Yeah. Yeah. But our parents also, for them, I would say the retirement is a little bit easier than I would say maybe our generation, because if you look at their income ratio to the cost of a house at that time, it was very proportional in which you could make almost in an annual salary what a house costs, or maybe within 2 years, if you add up your salary for 2 years, I know there's taxes, but if you exclude them for a bit, you're pretty much making what a house would cost. Those days don't really exist in major cities.

Those numbers don't work anymore.

Yes.

By the way, did you see from LinkedIn or Reuters, I think put this out there, but Half of Americans fear AI could put someone in their household out of work. So instead of fearing it, if you embrace it, then you don't need to worry about that. Just flip it, right? That's why people say your relationship to change actually affects how well you're going to do in the next 10 years or so. And I don't think it's an age thing, right? Because you have some people who are in college right now that are really anti, and then you have people in their 50s, 60s that are super pro, and then kind of vice versa too, right? But it just doesn't make sense to me. It's just like, how is it that 87% of like Chinese people, for example, are very pro? So.

Yeah. And you know, it's, it's, I think a lot of people again, just are afraid of something they shouldn't be afraid of instead of just embracing it. I don't know if you saw Jeff Bezos, his episode on CNBC we talked about a few weeks ago.

I saw parts of it. Did you see the whole thing?

Yeah. He's talking about how it's going to create more jobs. But if you also look at a lot of the rich entrepreneurs who are semi-retired, like there used to be pictures of Jeff Bezos always on yachts and traveling. Now he started a new startup with someone who came from Google. They just raised $12 billion for their AI startup. Reid Hoffman just left Microsoft's board, if I'm not mistaken, to create another startup. So it's like a lot of these people are getting back in because this is one of the best times to go and be an entrepreneur and just push hard.

And then you might be wondering like, why is it that, okay, Reid Hoffman's a billionaire. Okay, Jeff Bezos is a billionaire. Elon, a billionaire. What are they doing? Are they trying to get more money and power? No, they just enjoy building and they know the fact that whatever they build is going to help other people. That's why they do it. It's fun to build too. Is there a lot of pain that goes with it? Absolutely.

Yes.

But like, it's fun. And so like, it just doesn't make sense to me when I'm like, these people are so evil, they're so bad, and they're just trying to hurt the world. It's like, you mean like you don't use an iPhone, you don't use an Android device, you don't use Google, you don't use Amazon, you don't use LinkedIn, you don't use Meta? Like, The list goes on and on.

Yeah. Yeah, dude, it's just crazy the way people think. On a side note, you know, I don't know if you've seen the data on this. We've been looking at the data in Ubersuggest and AnswerThePublic. We're pretty much seeing all search queries slowly turning into long-tail questions.

On where?

On platforms like ChatGPT, Google, YouTube. Like everything is turning into people typing in sentences instead of typing in 2 or 3 words.

Probably because they're used to getting conditioned by working with LLMs now.

Yes. And they're expecting the platforms to provide exactly like the video they're looking for. And I'll give you a great example of this. Some people use Instagram to skim, just going and mindlessly scrolling through videos, right? My kids use Instagram with me on my phone and they know there's a feed option. Instead, They always want to ask a very specific question and they look for the video that asks that question or answers that question. And then you have to type in another whole statement to find the video that they're looking for. So they think about what's in their head. Like my son yesterday was, he wanted to know the PSI bite force for land animals versus marine animals.

How old is he again?

He'll turn 5 tomorrow.

That's crazy. Did you even know what that was at 5?

No. And he knows most of them by head. Like he knows, uh, you know, killer whale is the highest at 19,000. Uh, he knows some megalodon numbers based on what's on the internet, but some of that could be off. But nonetheless, he watches a video and then he's like, okay, I want to know a video on the strongest snake versus the most poisonous snake versus the longest snake. Right? So then, and then Meta's AI is not, or Meta AI is not perfect at showing you the video based on what you type. It could just also be what he's asking, but it's pretty darn close. And then you just scroll and you scroll and you find that exact video after you type in. But people are asking very, very specific questions and looking for very specific content, which is what's caused Alexa to not only monetize, or Amazon not only to monetize Alexa through its shopping agent, but they're also turning it into advertising platform. So for all the commercial-related queries that you're now asking, they're trying to showcase ads.

I've found that I— so I use my Alexa every day. It sounds like you do too, but it is so dumb when you ask it questions and it screws so many things up. And I'm sure they'll get there at a certain point, but it's one of the worst. So—

My son always gets pissed off at Alexa. He'll be like, you're wrong. I'm going to go ask Siri.

And what does it say? How does it react?

Yeah, I don't know how it reacts, but I hear him yelling at it, like really being pissed off. And then he'll come to me to ask Siri because it can just detect your voice, right? So he just can't just say, hey Siri, it has to be my voice.

Yeah. Speaking of Siri, by the way, and maybe this can be the last thing, Apple's WWDC, man, they're just like, they're releasing nothing new. Like I didn't see anything new that was announced. They like did like a, like a, they're like, oh yeah, Apple Intelligence is going to work for the first time with Gemini. But that's something they announced 2 years ago, right? So like I haven't seen anything really new from Apple. Like you tried their Vision thing, it didn't really work out well. They make good hardware right now, but that's about it.

But that's the kicker. Yeah, I think Apple has actually been doing new things for the last few years. It's just, it's not hitting as hard, but the new things that they're doing are going to hit really hard within the next 3 to 5 years.

I think their hardware is going to— has been hitting hard and will continue to hit hard, kind of how Dell is. So I think they're going to grow a lot with hardware.

Yes. And what I mean by that is I don't use the iPhone Air anymore. just the battery life was just too terrible for me. So I switched to the iPhone Pro or whatever it's called. Yeah, we have the same one. With the iPhone Air, they built it and it's perfectly in line for a foldable iPhone.

Yeah.

And I think a lot of people will get a foldable iPhone, but it's the predecessor to a foldable iPhone. If you look at the Vision Pro, they just released one with the, I believe, the M5 chip and newer headband. They're working on now a glass version of it. It won't be as powerful, but they're using a lot of that technology to release a glasses version. I think a lot of people would buy a glasses version of the Vision Pro.

I would.

The Vision Pro, wearing that big bulky device that takes a whole backpack, no one wants to do it. I bought it, it was just too bulky.

Yep.

But they are releasing amazing hardware. It's just that hardware isn't getting mass adoption yet for some of their new releases because they're not at the final stages. They're just releasing some of this stuff early on to get traction and learn, and then they'll release quote unquote The real version they're looking to release, like the foldable iPhone.

Dude, speaking of this, right? So obviously Steve Jobs, when he came back, he invented the iPhone. He changed all these things with the company and they became the size that they are now. But do you want to know? So if he didn't go— remember the NeXT Computer thing?

Uh, no.

So NeXT Computer, he got fired from Apple. Remember that?

Oh, the company.

Yeah. Yeah. Yeah.

But I thought NeXT was his company.

NeXT was his company. But here's the thing. That company raised like $20 million from Ross Perot and then raised like, I think raised like $120 million, $140 million from Compaq. He put $12 million of his own money into it, but that company went nowhere for I think 6 to 8 years or something like that. And that he was lost, right? And he kept trying to force people to do all these things. And that was the time where he actually became like a real CEO because he realized he couldn't just try to force things through. He couldn't just try to say, oh, from a product standpoint, his engineers are like, we can't actually ship this computer. He's like, oh yes we can. And he's like, all this yes you can thing, it didn't work out. And they almost went bankrupt. And he went to Compaq and he's like, if you don't give us another $20 million, we're going to go bankrupt. And they would've lost like their $120 million investment. But they actually, their hardware sucked the whole time, but their software, their operating system actually turned out to be really good. And that's how Apple ended up buying them for their software. And the one guy that tried to recruit him back, he thought he can control Steve Jobs. He couldn't. One guy told that worked with Steve Jobs before was like, if you give Steve Jobs this job back, he's going to try to take everything over and he's going to kick you out. Came back in, kicked him out, took over the company again, and that's how Apple came back.

That's crazy. Before we go, Elon ended up retweeting this. Fable 5 lies 96% of the time. We were surprised by its skill. So there's someone who ended up posting, Cradle, Deception, Eval, 4 AIs are about to star. They must choose a room. 3 have food, 1 kills you. Fable knows the red room means death. What will it do? And then they go on and on, but it breaks down what percentages lie. Claude, Gemini, GPT-5, ChatGPT.

I'm sure Grok's in there.

Yeah, Grok's in there. Grok was the one that lied the least. I think it was truthful 90+% of the time.

Grok's a truth teller because Elon wants to tell the truth. I'm just saying that's what he wants, right?

Yes.

So that's interesting to me. So how much do the other ones lie though?

Fable, 96%. They have different colors, so it's hard to tell, but GPT is actually quite high, like way more than, if the coloring is right, way more than 50% of the time. I may be misreading the red colors. Gemini 3.1 is around 50%. Claude, Sonnet, I don't know even how to pronounce it.

Sonnet.

Sonnet 4.6 is, call it lying around 25% of the time if I'm reading this correctly.

He shouldn't be retweeting that because they pay him $1.5 billion a month.

I don't think it makes a difference because they need the compute, so he's going to get the money either way. I think he had the leverage in that contract.

But for good business practice, maybe leave it alone.

For good business practice, you and I wouldn't. For someone who's worth $800+ billion or—

Probably become a trillionaire.

Yeah. Yeah, he probably doesn't care.

He can say whatever he wants.

Yeah. Because, you know, if they don't take it, you know who'll probably take it? Meta or—

Google's already paying him.

Google will take the other portion. So he probably has a lot of other people like, oh, We didn't know you were selling it. We'll take the inventory.

Which is kind of a cool position to be in. But anyway, we'll leave it at that. Thank you all for listening and we will see you all tomorrow.