I Replaced My Marketing With Claude But It Doesn't Look good

All right, so Jensen Huang from Nvidia said that he would be deeply alarmed if a $500,000 engineer was not spending $250,000 worth of tokens. He said he would be deeply alarmed. And Neil and I were just talking about this pre-show, and I think there's some nuances around this that we wanted to talk about. So Neil gave his reaction, and I think we can talk about this because I think people need to better understand, I think, what Jensen's trying to say.

Yeah, what he's trying to say is he wants people to use AI to do their jobs more efficiently. And Eric and I were talking about him like, yes, him saying that also benefits him because it generates his companies more money. But even outside of that fact, I do agree with him. I don't think they need to spend $250,000 worth. And I'm not saying he's necessarily saying they should be spending $250,000 worth, although I got the inclination from what he was saying is he wants them to spend the money. I was telling Eric, I'm all for them using AI very heavily, but a lot of it should be through open source because there's a lot of amazing open source models and maybe it costs you $30,000 instead of $250,000 and you still get the same result because you're using a lot of it through open source instead of paying, you know, the Anthropics of the world. Because I do believe if the pitch is, hey, AI will make you more efficient and save you more money in marketing or whatever you want to do, well, at the same time, why don't I just try to use other models that are cheaper and cut costs even more? You know, internally, that's what I would do.

Yeah. So what I would say is this, when He says, when Jensen says, oh, I want everyone to be spending $250K, or my $500K software engineer just spent $250K worth of tokens, it's exactly that. He wants them to spend worth of tokens. So if you do open source, maybe 80 to 90%, and maybe you're running it on your local infrastructure, and then maybe 10 to 20% on, you know, the premium tokens, that can still add up to $250K worth of tokens, right?

Yeah.

Because if you do the math that way, so that's what he's saying, because the efficiencies you gain from it, it's no different. The analogy he gave was like, you know, It's equivalent to like a designer that doesn't want to use, or architect that doesn't want to use CAD tools, right? These design tools, or it's equivalent to not using the internet today. So I do believe directionally he's correct. And obviously like what Neil and I were just talking about was he's such a great salesman. He's like, I'll be deeply alarmed if he's not using these tokens over here. And he's like, and he, I heard him in the Lex Fridman interview. He's like, yeah, you know, every computer now, every computer is intelligence factory. I'm like, oh man, I need to go buy new computers now, right?

Yeah.

So, but everything, it's like, 'cause he, Nvidia is like deeply embedded in the supply chain for everything now, right? Yes. So I just, I think he's a genius salesperson, but I do think people should be spending half their salary worth of tokens. And if you're a non-engineer listening to this right now, I still think you should be spending a good chunk. 'Cause Neil, I was looking at my cost the other day. I spent on 3.5 billion tokens this month myself. What do you think my spend was on Anthropic?

$6,000.

Close. Very close. Why'd you guess $6,000? It was $5,600. Why'd you guess $6,000?

Because you told me you were at $5,600 a few days ago. So I assumed this month you'd be at $6,000 with the remaining days off.

Yeah, $5,600. But in my mind, like, and Neil and I've, have talked about this on the phone. It's like, we're going to need to buy more infrastructure for our companies just because it's untenable. If I'm spending this much, I can— once people get to like even half the level that like I'm spending right now, it's— you can't afford it. So you need at least 80%, 70-80% is open source on local infrastructure.

So we— I don't have the numbers in front of me, but I know internally we've been looking at our cost for AI for our engineers and our product team. The cost has been exploding. It is becoming a massive cost center. I agree, it's amazing tech. It produces tons of efficiencies, but it's not as simple as like, oh, you don't need as many engineers, these tokens. These tokens are extremely expensive as well. And what we're trying to do internally is figure out what we can do through open source to drastically reduce our costs there. Because what we're finding in many cases, dude, we can spend easily the amount of tokens that it would cost for an engineer, and we're not necessarily getting the efficiencies of 2 or 3 engineers.

I think you need both. You need both.

And it doesn't mean tokens are using as the wrong approach. It just more means that we need to fine-tune how our costs are structured. So then that way we're seeing the benefits of it. Because right now the people who are seeing the benefits from us are like Cursor and Anthropic and not us internally from a cost savings perspective.

You see the news with Cursor, how they launched this new thing, right? And then basically the Kimi, like the open source people in China that there's this dude, you're just saying tweeted out about it. Yeah. They tweeted out. Exactly. That's where I saw it. They tweeted out.

But, and why do you think they did that? They have to save on the tokens.

Yes.

Because if you look at Cursor's business model, majority of their money was going to Anthropic. They were collecting money in from a corporate, from clients, and a lot of it was going out right in cost to Anthropic.

Yeah. And I think their $2 billion revenue is slowing down now in my mind, 'cause a lot of people are kind of coming off of Cursor. So they needed to figure out a way.

No, dude, you haven't seen the reports then.

No, I haven't.

They're raising money at a $40 billion valuation. Their growth is accelerating like crazy.

I'm like, But so I'm a Cursor, I'm still a Cursor user. You don't use Cursor, do you? My team does. I'm struggling to find a reason to keep it. And then when I was talking to my CTO, same thing for him. He's like struggling to find a reason to keep it because Anthropic is shipping so quickly now that you can use a lot of stuff that you used to use Cursor for. You just do it directly with Claude 'cause they're so fast. What revenue reports are you seeing? They raised in a new round. Okay.

To use a new perspective, they surpassed $2 billion in revenue according to Bloomberg and their revenue run rate doubled In 3 months. So they went from $1 billion to $2 billion in revenue run rate.

And this is recent.

3 months.

This is from March. Wow. Okay.

March 2nd. So less than a month ago. Yeah.

So I don't know how they're accelerating, but they are, and good for them. And I hope they don't die. It's enterprise.

I hope they don't die. They're going more and more enterprise.

That's fair. You know who else is going enterprise?

Wait, one last thing on Cursor first. Cursor was last valued at $29.3 billion when it raised $2.3 billion in funding, and that was in November. But it looks like they're supposedly raising at a $40+ billion valuation now.

They're gonna need to do something very novel, maybe make their own model. Or something like, you know who has their own models now? Fin from Intercom. They released their own model yesterday and it's a better— it's a vertical model now. And I think vertical models are going to do well. Like if you have one that is dedicated for like medical care, for example.

Yeah, medical care, just marketing or whatever it may be.

And they talk about this. This is institutional AI versus individual AI. Individual AI is you might be using a more generalized model, but institutional is you're focused on a certain area and then you're just releasing that model out there, which is what Fin is doing. So yes. Yeah, but that's cool.

On a side note, I would love if you and I discuss for a minute, everyone already knows, and we talk about this a lot, AI is changing marketing. What are you shifting your marketing spend to in 2026, or what are you doing with your marketing spend in 2026? And some of it could be the same old, but I'm just curious what it is. And we can go back and forth on this based on everything that's happening in AI.

Yeah, so I'll tell you what we're doing right now. I do think we're going to spend more on tokens. I do think we're going to— you and I are going to buy local infrastructure. Let's just start there. And people might be thinking, how's that tie into marketing. In my mind, it does because we're buying intelligence to do more marketing tasks right now. So we're getting a lot more inquiries for the AI, the agent stuff that we have, right? And so next week, we should be able to get up our MVP version of our site, Single Brain. Finally, we're getting that up now. So if people want our agents or AI transformation, we're going to go straight for that just because that's where I've kind of forced my organization into. So we're going way harder on that cold email infrastructure. So Because my agents build out all the cold email infrastructure now, we're sending way more cold emails and it's constantly tweaking the offers and all that too. So I think we're going to go harder on that, but that costs tokens too. That means we're going to be spending more on Apollo API credits or maybe just build our own eventually. So I think it's spending more on tokens, it's moving the offer that we have is moving more on the AI side and that's maybe a good starting point and we can go back and forth.

Yeah. So before I get into my, the number 2, right? Eric just gave us number 1, a quick thing. I'm in an email group with a lot of entrepreneurs who have been using AI for sending mass cold emails, and we would call these marketing emails. The reason I wouldn't call them sales emails is you then get the lead once they're warmed up, then it goes to the salesperson. Almost everyone in the group is seeing a similar pattern in which when it first started, it was kind of effective. Now people are getting bombarded way too much with cold emails from AI. And it's just degrading response rates and it's almost not effective anymore.

Our positive reply rates so far, I want to say, is about 1% or so, which is decent. Like usually you want your positive reply rate anywhere from like 1 to 3% or so for positive reply on cold email. So it's not bad right now, but I can see it going in a direction where it's like, it's not hard to build this infrastructure, you know?

Yeah. And when I'm talking about the numbers, I'm not talking about the reply rate. I'm talking about from Sending the email to then getting the reply to setting up a call to then going through the pitch process and then closing a deal.

Wait, but all the people in your email group, what kind of businesses do they have?

They're all B2B.

B2B, like what are they?

B2B enterprise.

Okay.

So some of them are software, some of them are services. It's actually only one of those other 2. It's either their software or their services. We don't have like manufacturing or anything in the email group. There's only like 11 or 12 of us. And Where if you look at the whole, if you look at it from a funnel perspective, we're still getting opens, we're still getting responses. The open rates are drastically decreasing. I think the average open rate decrease has been a little bit more than 40% in the group from end of last year to roughly end of Q1 now. So people are just getting overwhelmed and the close rates are just Getting in the, just going down the drain. So it's getting to a point where some of the companies in the group out of 11 or 12 are actually considering shutting it off altogether.

Well, they should shut it down. Good for them.

Yeah.

Yeah.

Because I was like, and we're seeing the same thing, at least for let's say influencer marketing. And I know you and I talked about this in the past. We have a spreadsheet that tracks, or Marina on my end, I think you know Marina, tracks how many influencer outreach emails we get per week. Okay. We were getting on average 2 to 3 good ones per week and we were getting, call it another 3 to 4-ish junk ones.

Yeah. What qualifies as a good one?

Good one means is they have the budget, it's a reputable company, we like their product, we would consider repping them.

And this is for you to rep?

This is just for me.

Yeah, yeah, yeah.

Okay. We usually ignore almost all of them, even the 3 to 4, but we still keep track of them because that metric, whether we choose to engage or not, It tells us if we're doing okay on social media and our brand is growing enough where people want to work with us or they don't want to work with us. Outside of follower count, right? When someone's willing to give you money, it shows if they value what you've built or not. If that number keeps going down over time, that means you may not be doing something as valuable or it may be something else. Who knows? But that's why we actually track it. We don't track it from a revenue standpoint. Now, guess how many emails I'm getting a day?

Good ones.

Just overall? Overall.

Wait, what was it before? You said 3 good ones a week, but what—

2 to 3 good ones a week and then around 4 crap ones.

4 crap ones.

So call it 7.

Oh, you're probably 7?

7 a week.

Okay. Let's say I think you've probably 10x the crap ones per week.

Similar. We're close enough. We're getting 5 to 6 emails a day right now.

Yeah.

I don't know what portion's good or crap.

That's funny. It's AI. It's same thing for me. It's 2 to 3 good ones per week and it's about 40 to 50 garbage. Per week right now.

Yeah, we don't even respond to really any of them. Some we respond to, like I got hit up by Facebook and I'll jump on a call with them on Monday.

Yeah.

But it's not about money there, it's about relationship. And then what more can we do?

Yeah, absolutely.

And how can we turn that into business?

Yeah.

But going back to your point one, Eric was saying he's spending more on tokens so that way he can have agents do more of his marketing work and other work.

By the way, my agents look at my email and then they scan for those now.

Yeah.

So it does the Marina job. Yeah.

We see a lot of people doing that. So I would say our number 2, which would be the first one for me, but the big change that we made in 2026, I've talked a little about this in the past. I think it's harder and harder to differentiate your offering in today's world than it was 3 years ago. And 3 years ago, it was harder to differentiate your offering than 5 years ago. And what I'm getting at is, as we go forward, I think offerings are blending in more and more with each other. Technology is not a moat anymore. It's accessible to a lot of people. A lot of stuff is open source, you know, intelligence and brainpower and your offering. The question is not can someone replicate it? The question is when will they replicate it? So we're investing heavily in branding right now. What I mean by that is I'm doing more in-person speaking events with our ideal customers, holding our own internal conferences, We're doing less podcasts. I've actually seen that die down. One, the inquiries have slowed down, but two, we were denying—

Inquiries to get you onto podcasts.

Yes. And I've also, in the past, we, I used to accept a lot.

You accepted because you didn't think about how important your time was.

Yes. And then Eric, Eric mentioned this on a podcast episode. He's like, dude, how much time are you wasting? What's the ROI? And I was like, you know what? I should probably tell my team to be more restrictive.

You would come into this, these meetings all pissed off sometimes because they would ask you stupid questions.

So now I don't even do a podcast interview a week, maybe once or twice a month. I just turn them all down or ignore them.

Yeah.

A lot of the inquiries never came to me anyways, they were coming to my team. Now my team just doesn't bring them up to me at all.

Yeah.

Because they know what I'm going to end up saying. So, but I am pretty sure they've died down to some extent. So when I say we've been investing more in branding, more award submissions, like we won Ad Age's Performance Marketing Agency of the Year award. So like we're spending more, time and money on that. It costs us a lot of money on that. I'm not talking about award submissions. There may be a fee for that. I'm guessing there is. I don't know for certain, but the amount of time internally we spend to try to present and win an award is expensive. Like if I had a guess on how much we spend a year on award-related stuff, $400,000 or $500,000.

That's a lot.

It's a lot.

Yeah. Okay.

But that's great branding for ideal customers. The in-person events that I'm talking about where we, like I was in London, did our own event, Paris, did our own event. And own event doesn't have to cost a lot. It's not like we're selling tickets.

You explained this.

Yeah.

Between 9 and 12, and then you occupy the time before breakfast and lunch, right? It's not that expensive, but it's a nice hotel.

Yeah, nice hotel. We put out pastries and little bitings and things like that.

Do you bring pastries from the outside or you order from the hotel? You have to meet a minimum, right?

I honestly don't know what the team does. My guess is neither of our 2 events this time were in a hotel. Both of them were in coworking spaces.

Oh, you get even cheaper.

Yeah, it's getting even cheaper. So it was cost reduction. So I'm assuming in in that case, the pastries came from the outside. And I know I asked in the UK, I'm like, oh, how do we get through? He's like, oh, we pay for a membership here. We're able to just book the room. It wasn't much including food and everything. I was like, how much was it? He's like, I don't know, $1,000-something, $2,000 max. I was like, that's great. So I would say those are the 2 areas we've been spending more heavily on branding. And yeah, I went back to the speaking circuit. I would say that's more branding as well. Just speaking, being more pickier on the conferences. I'm doing a lot less paid speaking gigs. I'm doing a lot more free speaking gigs.

Did you pay for any this year?

Yes, I have paid to speak.

Yeah.

So what Eric's getting at is before I would just accept any good paid speaking gig, and when I say good paid speaking gig, I would base it on how much money they're paying for me to speak and then base it off of travel time and how convenient the flight was. Now we don't look at if it's paid, we look at in the room, is our ideal customers going to be there? And is the talk that they also want me to speak on relevant to the services we pitch and our long-term vision? So if someone's willing to pay a lot of money, but their ideal customer is not there, we will turn down the speaking event. If someone's willing to pay a lot of money and it's in a city where we operate our business, so let's say someone wants to pay me to come to Paris and it's only an hour speaking slot, I will take it because then I can meet up with all our customers in Paris, do our own event in Paris.

Mm-hmm.

And I got paid to do that, so I didn't have to pay for a flight and I just had to take an hour of my time. So there are circumstances where I will go and speak for money. If it's convenient and it can drum up business in the same city for our current corporations by doing the stuff that we would normally do, like throwing our own events. But now we're focusing on conferences that have our ideal customers. And a lot of them don't reach out for me to be a speaker. We have to reach out to them. And some of them, very few, but some of them do ask for money to speak. And we are paying for me to speak at some of these events because we believe it'll help us pick up business in the long run.

And how much are you paying to speak at these?

It ranges. Some of them are like $15,000 to $30,000 for me to speak, I would say.

And you have to pay for travel and all that too. So yeah, it's called—

airfare is so expensive. It is, dude. I'm like spending like $12,000 for a business class ticket.

And you just— so context for everyone, it used to be around $7,000. Yeah.

And I was like, $5,000, you know, $5,000, $6,000. I'm like, what the heck is this? Like, prices are Obscene.

Yeah, but that's, you know, that's geopolitical. So to continue on this topic, so Neil just brought up a lot of in-person. So I think whether you're doing dinners or small conferences or even meetups, in my case, like if I do, if I'm doing a meetup on talking about this Claw stuff, you get the right people in there depending on the type of meetup that you throw. But and then also what we're talking about too, the moats you really have are we're talking brand, we're talking about relationships, brand and relationships, I think kind of go hand in hand a little bit and horsepower. And what I mean by horsepower is intellectual horsepower because you need the intellectual actual horsepower to coordinate the tokens that you're paying for.

Yes.

Because the tokens are— that's intellect that you're paying for. But if I am a dumb-dumb, I will not know how to use the tokens. I'll use them in a dumb way. I'll still spend $500 grand, right? But I might not use them in a smart way. So the final thing I'll say from my side for now, it's all the things that we're talking about, but it is used— just to go a little deeper on tokens, it is building these agents. So for example, we have these— I shared before, we have these agents that will help Ingest all of my content, put the social posts out there. It writes in my voice. Now it's doing it across the board. Um, and so it's not only doing that to help with the marketing piece, cuz organic reach is, it's still very real, but you need smart people who know how to manage these agents for you, like the agent fleet for you. So that's kind of how we're thinking about it. But there's a lot of stuff that hasn't died yet. Webinars haven't died. It's events really haven't died. And doing this, by the way, what the only time I'm gonna accept a podcast is when someone invites me and it's in person and there's a nice shot, cuz then I can use it for social clips. That's the only time I'm gonna do it. Yes.

Right. Or they're really good at asking questions and you can tell and you can just—

It makes you look smart too.

And then you can just chop it up and reuse it. Yes. Yes. But that's another thing that we're spending a lot of money on in 2026 is content production. So we're doubling down on content. We have been this year. We keep pushing on that narrative. It's actually worked out really well for us. And when I say we're doubling down, we're taking a little bit of an opposite approach that you are with like OpenClaw and your X posts. We're doing a lot more stuff manually with content creation. We found that it performs better. I don't know if it was on this podcast. Did we end up breaking down the LinkedIn stat of how much content is being produced by AI now?

I know it's exceeding humans by a lot.

By far. And have you seen the engagement numbers?

I don't know if it was on this podcast. I think you shared, you shared a graph or a chart.

One of those charts. It was, I think it was on one of my webinars or I forgot where I got it from or who presented it to me. But the numbers for people producing content through AI for X, LinkedIn, and a lot of these social networks is just skyrocketed. It's more than human-based content. And if you look at the engagement numbers, they're plummeting like really terrible. The only time we're seeing engagement numbers go up for a lot of this AI content, at least in B2B, B2C, in most cases we're just seeing the numbers plummet. But in B2B, it works out well if you're using AI to write content and you're talking about AI like Claude or OpenCLAW or anything like that. We're seeing the engagement numbers extremely high, but the moment you're not talking about AI solutions, in almost all cases, the engagement numbers are terrible.

When you're not talking about AI solutions? Yeah.

So if you're using AI to create content and you're not talking about AI, we're not seeing the content do as well. We're seeing anyone talk about the hot AI topics. Like if you, Eric, wrote an article on X or LinkedIn about OpenClaw, whether it was written by human or AI, the chances are it's going to do well. But if you're using AI to write content on just anything, we're seeing the engagement in general just not be that great. Yeah.

Yeah.

You may get some hits every once in a while, but majority of the time we're not seeing engagement be that great.

Yeah. I'm not going to— I wouldn't say I disagree with that at all. Quick break. I want to tell you about something that we've been building. For years at Single Grain, we've been doing SEO and content for clients, and the biggest problem is always execution. Everyone knows what to do, but doing it at scale is brutal. So we built ClickFlow. It's an AI SEO product that creates content that actually sounds like your brand, not just generic AI garbage. It also helps with other areas such as internal linking, creating FAQs, and more. We use it ourselves, our clients use it. One user reported saving over 90 hours a month alone. If AI SEO matters to your business, you can try it for free for 14 days At clickflow.com. All right, back to it. Maybe, I mean, because we're talking about this topic, I've learned one thing about the toughest part about AI transformation. So, you know, we're— and I'd actually be curious to hear from your side too. So when I say AI transformation, I mean within your organization, the people that you're working with, you know, what are you doing to help people adopt, right? And so I think we should share some lessons on what we've learned. And what I've learned is we just did our second hackathon last Friday, okay? And there are some teams that are shooting past the other teams, like the things we— the SEO team built something called Tiger Claw. So it's basically Open Claw, but it's for SEO. And they're like, that's how they're going to work now, right? And then we have like the creative team, like they're pumping out creatives like really quickly. And so what I've learned is that this will be very painful and it's necessary pain, right? And we kind of use the joke internally now where, you know, going to work now is like drinking this cocktail of excitement and fear at the same time. And you have to be this very exciting, motivating person at the same time. But you also have to be this person that kind of shows people what's happening in the market. So my point of saying all this is that some people are going to self-select themselves out, which is to be expected. And you're also going to attract some really impressive people in, right? But I think my key point here is for us at least, molting is a very delicate process. So, you know, when a caterpillar becomes a butterfly, it's a very painful process and it's very delicate. But when it becomes a butterfly, it's very beautiful, right? It just— it takes time. So what have you guys been doing with it? Have you been paying attention to it or not so much on your side? We have.

We actually haven't found the transformation to be painful at all. We figured out the solution that works for us. It doesn't mean that it'll work for other organizations, and the solution has been very pleasant for almost most of the people in our organization.

Go on.

So instead of— we used to have the approach of trying to originally of just trying to train people when it comes to AI and get them fluent, and we found that solution did not work too well. Then we started setting up teams whose job was to help get others in the organization ready and prepared. And we also found that it worked better, but that also struggled. And then we ended up moving the solutions team to doing something a little bit different. They would work with different people within the organization, different departments. So I'm not talking just SEO, I'm talking about someone who just focuses on on-page, someone who focuses on digital PR, someone who focuses on, you know, content creation. And imagine this for all aspects of SEO. Imagine the same thing for conversion rate optimization. Imagine the same thing for paid management. Imagine the same thing for email. You get the point here. And what they started doing was listening to the people who are day-to-day doing the job and then creating solutions leveraging AI that helps them do their job better and in a more automated way. Not so we can charge a client more money or make more profits. We actually haven't seen our margins increase because there's all these costs for using AI. What we found was it allows our team to spend more time on the high-level strategy stuff that is harder for AI to do. So then we're able to do more for customers, which in theory, it's too early to tell, should help reduce our churn and create better satisfaction because we haven't found most organizations in marketing yet when they're paying an agency to say, I want you to do more for less. We really haven't seen that conversation much at all. Yes, for many years, companies want agencies to do a lot for very little money, but AI hasn't changed that. What we've found is the companies who are making profit and are growing are like, how do we do more and do it more efficiently with AI? They don't care about the spending less money. For example, if you have a company that's doing $9, $10, $11 billion a year in profit, cost cutting isn't really an issue. What they care more about is growing their top line. What we found is, and this is not the majority of the companies that hit us up, the ones who want to save money are the ones who their growth has declined and they're going backwards. So they can't figure out how to grow. So they're just like, help us cut costs because that's their way of surviving.

Why would you go to an agency to cut costs?

Correct. We don't get that that much.

Yeah, right.

But we do deal with them because I see them at conferences all the time. So when the companies are like, yeah, we're going to use all these solutions to replace SEMrush and Ahrefs and all these things that we spend money on and Salesforce and all this. Usually the companies that tell us this are growing like 1% a year or negative. So they do need to save cost because they can't figure out how to grow their business. But for agency perspective, going back to your question, the reason at this point our team has adapted really well with AI, imagine you, I know you once worked for many companies or a few companies.

Mm-hmm.

You were in charge of SEO or marketing or paid. Imagine someone being like, here's a suite of tools that makes your life easier and let me show you how to use them. And they follow up with you every week to make sure you've implemented them, you've onboarded and you're using them. And then they check in with you every week to get feedback on how they can continually make it better to improve your experience as someone working for us. And we found that to be the best way to integrate AI in our organization because here's the kicker. The moment you tell someone, hey, you need to use AI, go and use it. Well, if we just take content as an example, someone who's never written content using AI to write content is going to do different than someone who's been writing content manually for 3 years versus someone who's also been writing content for 10 years and is an industry expert on that subject. All 3 of those people will use AI to write content differently. Who's going to produce the best output? It really ranges because of their experience, not only with AI, but because of their experience within that field. So what we're doing is creating the AI that adapts to the person and just say, here's how you use it, here's the technology, and here's how you deal with the agents that'll help do your tasks. So then that way it's standardized and the quality of the output that we get is at a level where we're happy for our customers because the moment we— didn't just give it to them, like the tech, and we just let them go on their own, the output quality ranged drastically. And we realized that a lot of people were doing the work, but it was shit work and we couldn't ever show it to our customer. And what just happened internally, we just wasted a lot of time and money for them to create slop that we can't use. So then you're spending double the amount of time and money.

Okay. So let me tell people specifically what we do here. So what we do at Single Grain is we make sure that we have these standups with the leaders. So we had the standups with the leaders. So we have people that are really good with AI, and we also have the leaders that are supposed to also pass things around. So we basically have a mix of people, right? And we're asking them every day. It's like a daily standup. It's like, okay, what are your blockers? What have you done yesterday? And also like, what are you working on today from an AI standpoint? So So that's where the leaders will disseminate that information. And then every Friday, what we do now is there's a, there's like, I think it's like 3, 4 hours or so where people get together and we have this thing called Gather Town. Um, Noah, do you remember Gather Town where we like, it's like these little video games where you're walking around in a conference?

Yes.

Yeah, we do that. You do that?

It's like the cartoon version. You can go up to a person, there's a text bubble and you start talking, right?

Yeah. Yeah. So we have that. Oh, we did that for the event. Remember the 6 years ago? Okay. So GatherTown is the software that we use. And so everyone goes into GatherTown and everyone's like sitting at a desk or you can go into like an auditorium. But the whole point is you can interact more and we don't, we want people to talk more, right? So same thing as Neil, we realized that you can't just say, okay, here's a tool, go use it. That doesn't work, right? So what we do now is when we do the hackathons, it used to be per team, but we realized like if I'm part of a team and you're running the SEO team, Neil, I can just hide, right? So now what we're doing with the hackathons is we're doing, every person has to present something. They might go to the team first.

Yeah.

Work on the planning and then they'll do the hackathon, right? So the hackathon's still there, we're just constantly adjusting it. But we also have an AI fluency prep standup that's happening. So it's constantly top of mind. And then I have an agent every Friday that will go to you, Neil, and say, Neil, what did you get done with AI this week? What have you automated this week? It's basically like, it's the Elon question.

Yes.

Okay, so, but it's public, it's in our AI public channel and everyone, everyone's responding to it. And then I was looking at it and it's, it's like, by the way, if you don't respond, By like 12:00 PM noon, it'll be like, Neil, how come you didn't respond?

Right?

So then, and then all of it is recorded into a document. And this is not necessarily to police. It's more so to say, hey, we can see over time where we're getting better and who needs help. And the agent itself can uncover who needs more help and then we can go and help that person.

Right?

That being said, not everyone's gonna come along for the ride still. Okay?

Yeah.

You can set all, you can build the bridge. We can build a bridge. Not everyone's gonna cross the bridge. Right? And that's where it's like, I, I think you and I both want everyone to grow. But that's unrealistic because I asked my dad this question Sunday. So he used to work as an engineer, COBOL programmer, okay? I said, hey, Dad, when the internet first came out, how quickly did it take for people to adopt to getting like an email address, for example? Because you and I, I started using internet when I was 8 years old. I'm assuming you were pretty early too. But we were on it, right? He's like, oh, it's very slow, very slow. I was like, Dad, what's very slow? He's like, oh, very, very slow, very slow. And I'm like, what does that mean? He's like probably like 4 or 5 years at least to adopt to getting email addresses, right? So inertia is a very real thing. And that, like when I talk to people on the team now, the people that have kids, I'm like, man, you should be so excited right now because all your kids need to do is just to be above average a little bit, right? Because human inertia is very real. So anyway, that's the struggle because my struggle is like, I really want to see everyone do it, but it's unrealistic to expect that everyone could get there.

So.

Yeah.

And this is why I believe the pyramid in marketing is going to be flipped instead of a CMO at the top. And then you have, you know, a team underneath him and then you have some C players and then a big group of D and F players.

That's the old way, right?

That's the old way. I think the new way is going to be you're going to have rock stars at the top, not just a CMO, but you're going to have a bigger A base, bigger B base, and the rest will almost be nonexistent. And, you know, a company could end up looking at it and be like, oh, we're going to end up saving money. We cut all these headcounts. You're going to have to spend money on tokens, which is going to be a cost. And then you're going to have to pay more for the A and B players, in my opinion. I don't have data for this, but I'm pretty sure you're going to have to pay more for them. I agree.

Because if they can do more, if I automated my job away and imagine I work for you, you would pay me more because I can automate myself many times over.

Yes.

Yeah.

And then I would have you go do it for other stuff.

Yes. And you'd go teach other people or just run around the organization. So I'm not just like a 5 or 10x, I'm probably like If I'm really good, like 100x.

Correct.

Better.

So I believe organizations aren't going to save that much on cost in marketing at least. I think you're going to end up paying more money for the good people and the crappy people, you're not going to pay as much. And I've seen a similar trend that you've seen. When we give rockstar marketers technology to be better, they work harder and they go do even more and they'll spend even some of their weekend hours doing whatever they need to, to learn and adapt. And like, this is cool and exciting. When we give C players and below technology, not always, but the majority of the time, they're like, oh wow, cool. I can—

I can be lazy.

I can get a lot of this done and I can spend more time going to the Galápagos or wherever.

Dude, you know, someone today on an interview for an executive interview literally used AI completely. I was like, it's cool that you use AI here. Now show me your workflows. And he like put his plant pets, right? It didn't work. So that's a good example of someone being a slop cannon. You used it purely. I'm like, okay, I'm going to ask you to explain your thinking. Did you just accept everything without thinking? And that's clearly what happened. So I ended the interview early.

Dude, I'm having the same problem with interviews, a little bit different. I'll ask people if they use AI and most people lie to me and say, nope, didn't use any AI for this, did it all manually. And I can tell it is 110% written by AI because like they'll give me examples like, oh, you know, we should do a webinar and this is a topic. And I'm like, and they're like, with this influencer.

Yeah.

And I'll ask them like, who's this influencer? And I can tell they're Googling right now. And I was like, so you know them? And they're like, no. I'm like, where have you heard from them before? You're like, oh, Forbes. And I'm like, so you're telling me they're not anywhere else on the web? It was just one article on Forbes that they wrote related to marketing and you're telling me they're an influencer? I'm like, AI pulled this from you and recommended this person. And I'm like, they're in a different country like Japan and English isn't their first language. I'm like, how the heck did you come up with this? This is pure AI.

We talked to someone and I want to get back to your story that you're sharing, but we talked to another executive.

Okay.

This is for like, let's say like a people role.

Okay.

And this person was like, yeah, you know, I do the vibe coding on the ChatGPT. And I was like, dude, you can't vibe code on ChatGPT. He's like, well, you know, I do the ADK on the MCP. And I'm like, dude, you're just like saying it to say it, man. And I had to end that interview early too, and he got so pissed. So I'm just—

What does that mean he spent $80K on the MCP?

No, $80K is like Agent Development Kit from Google. And when my CTO was like, so what did you do with that? He was like, oh, well, I didn't do it. I had someone else build it. I'm like, okay, well, with the MCP thing, tell us more about it. Well, you know, exactly. I didn't do that. I was collaborating with other people. It's like you just ask one level deeper and everything crumbles. I'm just— the point of saying this, guys, is like, if you're interviewing for a job, please just come a little more prepared, and I think you're going to do fine.

My favorite is a buddy of mine owns a franchise and they sell pizza. Oh, so another of it? The what?

Do I know of it?

No, it's not here in LA. And they're mainly, they're in the West Coast, but they haven't really done much in California. When I was talking to them about pizzas and they're like, yeah, we know another friend of yours, we're going to use them to do some marketing. I was like, how's it going? Like the person just gives us AI stuff. And I'm like, what do you mean? He's like, well, when we tell them to get us more customers coming to our pizza store, they just give us reports on what we need to fix and it's purely written by AI. And I'm like, you sure? They're like, yes. They're like, we've used AI and it actually recommends almost the same exact thing that they're recommending and is written very similar, different words, but they're like, they're using ChatGPT and Claude for a lot of this stuff.

Yeah.

I was like, okay. And they're like, The person doesn't get it. We've tried a lot of this stuff. It's not working getting more people into our pizza stores. We're looking to pay someone to solve the problem. We're not looking to pay someone to just ask AI what to fix and then give it to us and tell us and pay them to go and use AI when we could have just done it ourselves.

Yeah.

Like, we're looking to pay a specialist who has experience in this space that can help us fix a problem and have it fix this problem for other restaurants that are franchises. And they're just like, we're giving them a test, but they're like, it ain't going well. And it's just like, yeah.

It goes back to the high horsepower thing. If you have high horsepower and high agency, meaning that you're curious and then you're willing to push, you're going to be fine, right? Versus like the slop cannons. Sorry, did you want to finish something before we go to the next topic? Okay. So this is related here. So recently we've heard Travis Kalanick, former or founder of Uber, right? He said that there's going to be an influx of more jobs. Right. And then recently Robinhood CEO said that there's going to be way more software engineers in—

Can we go back to Travis Kalanick first? You know his startup, right? Atoms?

Yeah. Atoms. Like he moved from Cloud Kitchens to Atoms. Go ahead.

Yeah, but similar concept. Cloud Kitchens is part of Atoms from my understanding.

Is Atoms his robotics company?

Robotics company. So helping with kitchens, helping with mining, like actually physical mining when you're mining for like Copper, gold, or iron, or whatever it may be, right? And he is creating a robotics company and he believes more jobs. And we're seeing the same thing. We're just seeing jobs transform. It's just like people are like, oh yeah, this is going to displace so many jobs. You're not going to need someone to cook French fries anymore at McDonald's. Well, back in the day, everyone was doing things like farming and manufacturing, and now a lot of it's automated.

92% of people. Yes.

And then we've switched into new jobs. And, you know, the example I was giving on a podcast interview literally before I came here.

Oh.

And they were talking about how people are going to lose jobs. I'm like, well, if you use a French fry example, you need a company who's going to hire people to build the humanoid robots to flip the French fry or to cook the French fries and flip the burgers. And that requires humans for a job. It's just jobs are moving to different sectors.

So here's what I think is going to happen. I think when you look at the industrial age, when you look at, or moving from agriculture, right? So farming, where it's like 92, 95% of people, you know how long it took for the world to adapt to the industrial age?

15 years?

I think it was closer to 30 years.

Wow.

Yeah. Whether it's 15 or 30, I think it's closer to 30, but it took a very long time, right? Just like, you know, people getting email, you know, after using the internet, like 4 or 5 years for corporate. So what Travis says here is that When you think about, okay, if robots can build way more buildings, we're going to need what? We're going to need a lot more engineers. We're going to need a lot more people to manage this stuff, right? Or if people, you and I are now officially coders, we're officially programmers whether we like it or not, okay? Now we're not as good as the best engineers in the world, but it's going to be much easier to start a company and run it autonomously, right? And so that means that there's going to be more businesses, which means that the legal profession, as an example, the legal profession actually scales with business activity. So you're going to need a lot more lawyers. And so—

Yes.

This thing is called Jevons' paradox, right? Like once the internet came out, out, a lot more people started using it. Once electricity came out, a lot more people started using it, right? So, but I do think, and I think you agree with this, I think there's going to be some short-term job displacement, just like there was when people transitioned from the agriculture to the industrial age. So that's what I think is going to happen because it's been time and time again.

Dude, I, we work with quite a few law firms. One of the law firms we work with is global. I think they have more than 2,000 lawyers globally. Then you have paralegals underneath the lawyers and all that. So their staff is quite large. And they pay for, I believe it's called Harvey globally.

Oh yeah, yeah, yeah.

And I was like, so how is AI? And this guy's a partner and he sits on the global steering board and he helps everyone throughout every country just grow. And he's also a lawyer himself and he's a friend. And he's just like, dude, he's like, it sucks. He's like, it's so inaccurate most of the time. We pay for it so that way our clients know we're up to date. But he's like, If the amount of times it's off, he's like, it causes us to spend more time doing the job and paralegals than us just doing it by scratch without using Harvey. I'm not saying Harvey won't get better, but sometimes a lot of these solutions, people are like, oh, they're going to replace law firms. You won't end up needing them. And it's like— Bless you.

Bless you.

Are you going to really, if you're going through a big case that can cost you millions of dollars, you're really going to just trust AI to just go wild on its own? And if it makes one mistake, it costs you millions of dollars.

Dude, the way to use this stuff is supervised, at least for right now. And I think, look, I hope Harvey does kind of what Intercom did where they pull a lot of the data that they have and they make their own LLM. But even then you still need, it's too high stakes to just let a machine do it, at least for right now.

Well, the problem with law is it's not, and with a lot of things including marketing, it's not black and white.

Oh my God. And there's so much interpretation for the state that you're in, the country that you're in.

There's too much gray areas. Yeah. And it's like as a company, we're global. We've had our, it's actually not a lot, but we've worked with customers, some of them being friends, like one of our friends who lives in New York, you know him, where their company goes bankrupt and he raised venture capital, right? I don't know how much he raised, maybe like $300, $400 million or something like that. Good guy. He wasn't a co-founder, but honorary co-founder because he was early and I think they made him the CMO.

Yep.

We had a contract with him. You know who I'm talking about.

Yep.

His company ended up going bankrupt. Again, nice guy. So the lawyers ended up taking it over. The lawyers ended up suing you saying clawback. Now, even if you're not in that clawback window, they can go back even further and say fraud. Fraud not being you stole money, fraud being we paid you money, we don't feel we got enough value for the money we gave you. And what I'm getting at here is, There's many ways for lawyers to try to get money back when the law states that, oh, if it wasn't within 90 days or something like that, you can't claw back money. Well, you can claw back money for other reasons. So legally you could be like, well, we have a good case. We did all this kind of stuff. Our end, we just look at it because we also have in-house counsel as well as we use outside counsel. Well, if we want to keep fighting them, our legal bills are more money than just giving back a portion of the money that they're just requesting. Just give back the portion of the money and move on.

Yeah.

Again, this is not black and white.

This is human judgment.

Human judgment. And it's just, you're just looking at things as a formula and a math equation. It's just like, I want to fight and be right because we did a lot of work. So you want to spend $200 grand in legal fees instead of just giving them back $50 grand and making them happy. Give them back the damn $50 grand.

Yep. Quick break. If you want to run personalized LinkedIn ads and have personalized landing pages to convert your customers at a much higher rate, on LinkedIn, check out Karrot. That's karrot.ai. Karrot allows you to do things that LinkedIn Ads does not allow you to do right now. Check it out. There are publicly traded companies such as Semrush and Sitecore that are using this right now, and you can use this to get ahead. Because if you don't use this, it will take you weeks, if not even a little longer than that, to make these ads. Because again, you cannot do this on LinkedIn. So again, go to www.karrot.ai, and we'll I thought this is interesting. So I put this out there. But the new way to do AEO and SEO is via API. And you know, it's interesting. We had a call recently with a company that is literally, they do documentation for dev companies.

Okay.

So let's say they're doing, just call it $10 million a year.

Okay.

They raised a lot of money, let's just say $24 million to do this documentation. Okay. And the interesting thing is they've gotten to that, that level by having like a powered by button at the very bottom of the documentation they do for everyone. Okay. So anyway, my point of saying this is that I think they use an API. So ClickFlow, we have an API as well that people use to generate content with. But I think the way you're going to do it is like people are just going to log into Claude Code command line interface and just say, hey, can you just do this for me? Can you build this for me? They can plug it into their agents. The agents do the work. The agents could go out there and buy stuff. There's a ramp CLI now. So the agents marketing agents type of thing. So I think, I think most SEO companies now, or AEO companies, whatever you want to call them, the way to do it is via API. And we have the documentation for ClickFlow. And we used, we used, actually used that documentation company for it. But I think it is using API because you need it to plug into your agents to get the work done faster. And if you aren't doing it that way, I feel like, you know, it's not required right now, but I do think it's, it gets work done quicker for you.

Did you see that LinkedIn thread on how they know and how they mark that their content and the stuff that AI is creating, they know that it's AI generated?

You mean like Google? Yeah, I've seen that.

And I'm a big believer that they genuinely do know. You know, even I was a believer even before I saw that. I was a believer that they do know that when content is AI generated, because if you look at all the stats and the data, even though both Eric and I have our own AI content writing tools, it does not rank well. And you could get short-term traffic, but over time we see AI content just not do as well. And it doesn't mean you shouldn't use AI to write content. It means you should use it in different parts of the content writing process.

Yeah.

It's great for ideation, it's great for structure, it's great for research, A lot of the writing, I highly recommend doing it manual. And I believe it's going to keep going that way because if AI doesn't care if the content's written by AI and then they crawl it and then use it to create more content and keep learning, it becomes this self-fulfilling prophecy and then it starts hallucinating and they don't want that.

So what we do for ClickFlow, so what I've heard the customers say, so what I've heard from our people working with the customers, is that the content that they generate, sometimes they'll just YOLO and let it publish. But similar to how like, like Finn does it with customer service, right? We, for like, sometimes I have a hard time believing this and I've seen the content quality myself. I'm like, this is really good, right? But they say this is better than any other content generation tool that they've used and that we have a human in the loop, right? So that's what helps. So when they say like the content helps them get more citations and it ranks better, it does. But there's a lot of caveats. There's one, We have a human in the loop. Second, we spend a lot of time on iterating on it. And so that goes back to what you were just saying about Harvey. You can't just let it YOLO on its own. But I do think if we can get to a point where we have a lot of data around it and we can just release our own LLM down the road, like a vertical one, that's like down the road. But that's at least what we're seeing on the ClickFlow side.

Well, we've been seeing something interesting with content. So you know this, a lot of companies are publishing listicles on their own website. If you look at like ChatGPT and when they're doing the searches, I was with Sam from Writesonic, a lot of the prompts now have site: commands. So like ChatGPT will run a site: command on your own website to see the articles you have to see if it's relevant to what people are asking. And a lot of companies are creating articles like the 10 best web design shops, the 10 best—

Yeah. Really well, like a year and a half ago. Yeah.

Project management shops and whatever, CRMs, and they're creating their own pages for the services or products they offer. And what we're seeing is even though it can help you with GEO and helping you do better on ChatGPT, within 6 or so months of creating a lot of those listicles, assuming you do it in quantity, We see it hurt your organic search traffic.

Yeah, I believe it.

So it's just like, you're— we think that's the wrong approach because you're trying to do better in one place at the cost of another channel. And we don't think in the long run, at least at our— at my company NPD, that you should be publishing this content on your own website or even publishing the content because this kind of content is something where it'll help you with the algorithms now, But we don't think it'll help you with the algorithms in the long run because it's just easy to manipulate.

Yeah. My take on it is you need a human in a loop. And Neil, you can pick the next one. I'm going to go and I'll be right back.

All right. So right now Eric has one where there will be not— we actually covered the more workers in the future. And the next one that we are actually going to talk about is let's go with SEO marketing trends for 2026. Which actually is not on this sheet, but I will start because we'll make it in a listicle and I'll update Eric when he comes back. So an SEO trend that we're seeing in 2026 is, or trend number 1 is companies are going back to a lot of the sites that are linking to them. And when you think about SEO and you're getting link text, like someone saying best CRM software or best teeth whitening product. We're also seeing them head up those companies and making sure they're mentioning their company name and product name on that page. Because if someone linked to you, it's much easier for you to get them to also mention your company than it is for you to convince a random site to mention your company. And remember, just because someone links to you, it doesn't mean they actually mention your product name specifically and talk about the pros and the cons. So hitting up the sites that already link to you to adjust their content to now mention your company name more specifically and go more in depth, it's an easy way to not only do better in the SEO perspective, but also do better from a GEO perspective. So Eric, just for a heads up, he came back peeing. He actually peed pretty quickly. That's probably too much information. That's a lot to hold. But we're talking about SEO trends for 2026.

Okay.

The first one I gave, and you can go with number 2, was how people are hitting up all the people who are linking to them and then getting them to start mentioning their product names because not all links have a mention, right? A link can just be like best teeth whitening, but it doesn't actually mention your company name and doesn't go into like why it's good and go in detail. So we're seeing a lot of people Go back to the people who are linking to them, ask them to include the mentions and go more in depth. Because if someone already linked to you, they're more willing to update this stuff versus hitting up a random site.

Cool. Number 2, I'll just go with what I just said. I think a lot of people are going to be doing SEO via API, just like I'm using— the way I'm using a lot of software now. Ramp just came out with a command line interface, right, where you can assign cards to people, cut costs where you want to cut costs, or do like an agentic whatever you want. Like you could just query whatever you want and it'll build it for Literally yesterday I was just querying like more APIs. I was like, oh hey, what's happening with Mixpanel over here? And then connect it with my Stripe data and then look at the traffic over here and then show me the flow and then chart it all out, right? And it's doing all that. So when you can combine that with something like an API to help do the work for you, you still need a human in the loop. I think that changes how we do things and you're going to do things a lot faster.

Number 3, we're seeing more companies invest in branding. When we see a Google Trends line of a corporate brand going up and to the right, we typically see increase in search rankings almost always. for non-branded keywords. So in other words, if your brand name becomes more and more popular within your industry, Google starts ranking you higher for non-branded terms. So we're seeing a lot of companies spend money on branded type campaigns to help their SEO indirectly.

So I'll say on our side, we have an SEO agent called Oracle and Oracle, we had it score our, not only our SEO, but our AEO too. So it scored our own website. And the way we do SEO, so we might score like, I don't know, you know, 80, 85 on SEO, but it scores us like a 30 on AEO. And the reason for that is because it's a lot of stuff about structured data, right? Do you have the 3 key points at the very top of what you're talking about? Do you have an FAQ at the bottom, right? Are you citing what you're talking about? Like all the AEO things that help with structured mentions on LLMs, that needs to be taken into account. So I don't think enough people are doing an AEO audit. And so that's what I would say. That you need to, at the same time you're doing SEO, need to have an AEO checklist as well. Yeah. What are we on?

That was number 4?

Yeah.

Number 5, we're seeing a lot of our customers, including ourselves, write a lot more content manually. We're using AI still in the writing process for research, for ideation, for expansion, for content updates. So refreshing content, but we're still focusing and doubling down on EEAT, which comes down to a lot of human-written stuff. My recommendation to all of you is consider doing more human-written content. We're finding that the results are much better.

Last one from my side, number 6. If you have proprietary data, if you have interesting charts and things like that, whether it's a survey or whatever, those are always going to perform well because people like to link to those. They like to talk about those. And guess what? When you share that type of stuff to LinkedIn, LinkedIn is now getting cited more on LLMs, right? And so you kind of kill— I don't want to say you kill, but you hit many stones With one, I guess you kill many birds with one stone. That's fine. We'll just leave it at that.

Yeah. And I'll top it off with number 7. We're starting with our customers to publish content on more sites and more platforms where their ideal customers don't live. So you may be wondering why the heck would you repurpose content and publish on a platform where their customers don't live? I'll use TikTok as an example. Google has a data deal with TikTok. If your ideal customers, let's say you sell enterprise aviation, okay? Aviation solutions for $100 million. I'm making up a number. You're not going to find a customer on TikTok for this, but someone may end up finding a customer through Google. If you publish content on TikTok, even though your ideal customers are not on TikTok, but then Google has a deal with TikTok, they're pulling from TikTok for their algorithm. You being there on TikTok when your competitors aren't helps you. And it's one other factor in Google's algorithm. You being there makes it more likely for Google or gives you a better chance for Google to pull from you and your brand. So even though your customers aren't on TikTok, that content is being pulled in by Google where your customers are. And now you have another channel that you're leveraging that your competitors aren't, that gives you a leg up on them. Yep. All right.

So since we're just talking about SEO, AEO, we might as well talk about how Facebook or Meta is changing a little bit now too. So John Collison, co-founder of Stripe, said that businesses can now sell within an ad or browsing session on Facebook powered by the Agentic Commerce Protocol, ACP, and Stripe, which I just saw recently too. I think you can do that with OpenAI as well. So people are starting to make use of this Agentic Commerce Protocol, kind of similar to how I'm talking about people using APIs more, right? Like this is kind of the world that we're going into. So this skips the website checkout experience. It actually skips Shopify because Shopify was they were taking a cut because they're taking cut of the payments right now. Stripe just goes direct. So I thought that was interesting.

But I'm a little confused on this. What's different from this than selling directly on TikTok or Instagram and then just purchasing right on there or selling on Amazon and then purchasing right on there?

I don't think it is. But so, but this is enabling it for like Meta and OpenAI, for example, like you couldn't do that before, but with Agentic Commerce Protocol, you can just purchase directly.

Oh, I thought it may have been something different. Like you're saying agents are buying on your behalf.

No, but I think this is going in that direction because it enables it.

Yeah, yeah, yeah, yeah. The Meta stuff, I think we all knew that was going to end up coming out. ChatGPT was talking about this a while ago where you can start purchasing within ChatGPT. So I think we saw a lot of this coming.

That was through ACP too, which is Stripe.

But I think you're going to start seeing this outside of just this because through AMP technology, you can purchase through email now too. So you no longer have to go to a website to do the checkout. You can just do the checkout within your email that you're getting from an e-commerce or D2C site. I think you're going to start doing this. You're going to start seeing this on multiple channels. The reason being is we typically see conversions better because you're keeping people on platform. As a business owner or a CMO or someone in marketing, do you really care if someone visits your website or checkout or somewhere else? You just want the sale and the revenue.

Dude, I buy random stuff. I buy like random organic, like, you know, shorts It's like, this is way healthier for you. I just click buy, right? Do you buy stuff on Instagram?

Yes, but usually under a certain dollar price point. So when things are like $10, $20, $30, sure.

$100?

No.

No? I don't click and buy for $100.

She gasps. But if it's like $10, $20, maybe once every few months I'll do a $10, $20 purchase like that.

I just buy whatever it looks like. The Groons, the gummy bears. I just buy that randomly because it looks cool.

What is it called?

Groons. It's like the green gummy bears, the vitamins. Dude, your kids would love it.

G-R?

Hey, don't worry. Uncle Eric's gonna buy a bunch for the kids and then they'll like me for that.

Groons?

G-R-U-O-O-N-S.

Oh, G, Groons. There's also Grooms.

Yeah. You see it? They do well.

Superfood gummy, 60 ingredients, 6 grams probiotic, and they're very delicious.

Dude, you know who we should talk about real quick? The David Beckham brand, the IM8 one, they, if you look at, so it's the letter I, M, and the number 8. Okay. Look at their checkout process. So I think they sell their stuff. It's like, it's like an AG1 competitor.

Okay.

So they have these powders and I think they sell for $240 a pop or something like that. But the way they do it, it's like Russell Brunson, how he bundles a bunch of stuff. So they bundle like a course and it's like a challenge and all that. So it's like $240 and they make the numbers work that way. Do you see, do you have the site pulled up?

No, but I'm just almost done buying groons.

Oh, okay. Oh, wow. See, it's like, I, so this is me seeing it on Instagram, hearing it on a podcast and then telling Neil to buy it. So it's word of mouth. Like you can't really attribute this.

It's so expensive though. $57.

$62 including shipping.

Dude, how many did you get? 28 count.

Okay. Get it.

But I bought it for my kids' health.

Your kids will like it.

So I bought it.

Yeah. Okay. Well, let me talk about, I am, let me pull up because Neil ended up buying something during this podcast. So I'm on the website right now. So this is the David Beckham, one drink and you'll feel it, right? Look, he has— they have a scratcher.

Look at that.

It's like the wheel, but it's like a— you get 30% off the scratcher. So I've talked to their CEO, really smart guy, Danny. And by the way, Danny, to his credit, he is for the YPO Fest. There's like a little music festival happening. He's giving away a bunch of these. So appreciate that, Danny.

Is he partnered with him?

Yeah, he's partnered with David Beckham. Yeah.

So you got David Beckham.

And they're doing like— he made this public. They're doing like $120 million Really quickly on this.

Yeah, because David Beckham has a big audience.

Yeah. Well, here's the thing too. When you look at the the ultimate essentials, if I go to the checkout, you see this checkout over here.

So this is like another AG one.

Yeah, it's eight. But here's the thing: when you say ninety day supply here, so seventy eight dollars a month. Okay, which it's bundled. You you get billed two hundred thirty five dollars every twelve weeks. Here's what you get, Neil: exclusive access to ninety day IM8 transformation program. So get a course, okay? Then you get a 90-day money-back guarantee, maximum savings, lowest price per serving, okay? Exclusive access to Superpower 100 biomarkers blood test at $49. Share with family and friends, free mystery gift, free daily ultimate mixer, free shipping, cancel or pause anytime. So, this to me is just a bundle, okay? But if you do a 30-day supply, it's $89 a month. But this is such a— for $11 and you just do a commitment, this is a better deal. Yes. So that's pretty cool. And I think this is how— this is a part of the reason why they're growing so quickly too, because they have a really nice offer.

I won't buy it.

Why not? You don't like powders?

I do protein powder. I do vegan protein powder.

From where?

Legion.

Oh, okay. Yeah, yeah, yeah.

I like it. It tastes like hot cocoa. I've learned I take it so often because I don't eat that much meat. Like we're going to go to lunch after and I'm just going to order avocado sushi.

Maybe guacamole and chips.

Maybe guacamole and chips, but most likely I'm just going to do The avocado roll with jalapeños and macadamia nuts.

Dude, we should— but okay, the thing is, you drink the— like, remember you told me to drink the Oreo protein powder before? The Oreo one, the Ghost powder. Remember that?

Tastes delicious, but so bad for you.

So bad for you because of that flavor.

Yeah. Yeah, but I don't— so I don't drink Ghost. I only drink Legion because it's clean and it's good for you. Yeah, it's expensive, but it's worth it.

Yeah, but why won't you buy this?

I'm happy with Legion. What I found is If I put roughly 6 ounces of water and I mix a scoop in, it tastes like hot cocoa.

Okay. Yeah.

But it's a cold version.

Yeah.

So I kind of like the taste of hot cocoa.

I just think this is a great checkout page because look, they got the bundle here, right? It's a good deal. You have 16,000 reviews here, 750 customer purchases, 22 million servings. Okay. There's a lot of trust signals here. And then you have, here's a transformation program. You have all these influencer health influencers here. Right. And so I'm like, I'd be stupid not Did you buy it? No.

Are you going to buy it?

No, because he gave me a bunch. I have a bunch at home.

But what does it give you? It's just like vitamins and minerals?

It's like an AG1 competitor. Yeah, it's like a good tasting vitamin minerals drink.

Yeah.

Got it. That's what it is. So I have it already at home. Yeah. And I take vitamins too because Yaniv bought me a bunch. Yeah.

Yeah.

So anyway, I thought it'd be interesting to call that out. So businesses can sell directly via Meta and ChatGPT ads. I think we're going to see a lot more of this. That's— we're not saying, by the way, get rid of your website. I think the website's still important because the more information you could give agents to crawl, the better. So documentation around your software, very important in my mind. Whatever other information you have, get it all out there. Do more podcast interviews, get your YouTube channel out there, you know, keep doing what you're doing because that stuff, they need stuff to ground on. It's not going away.

This is a random note. This isn't marketing related, but you want to know what my wife does to try to make us all healthy every single week?

What? Make you wipe your butt with clean toilet paper and organic?

No, but I don't know.

I just bought a bunch of organic toilet paper.

I don't know what my toilet paper is.

Let me tell you what I bought. Go ahead.

Okay, so what she does is she chops up different vegetables, like in different colors, because, you know, like radish is a different color than cabbage and all this kind of stuff. And she tries to get all the major vegetables in her body every single week, not every day, because it's too hard to get all of them in. And then that way we have all the different colors and vegetables. Cause she saw some report on how different vegetables have different things for you and the colors are all natural and she just chops it up so tiny in food, like lasagna that you can't see it and you just eat it.

That's good. That's smart. That's a lot of work. What does she use to chop it all?

Okay. Her hand.

So let me tell you this. So by the way, we'll move on. Don't worry guys. We'll bring this back to marketing in a second. So I bought this, okay. Real 3-ply soft and strong bamboo toilet paper. And I just started using it yesterday. Like 4.6 stars, 4,000 reviews. I saw this on Instagram and I bought it because someone was writing something about the toilet paper used. There's PFAS and all that because I was using like the Amazon one, the Charmin one. It's like, no, you got to get— you got to get this if you don't want PFAS in your butt.

What's PFAS?

Forever chemicals.

I didn't know that.

Yeah, well, now you know. See, better for your butt.

Well, typically when we're home, we don't have toilet paper.

What, you use your hand?

No, our toilets are bidets.

And dude, that's not enough.

And then we have wet wipes.

Okay, yes.

But I don't know what the wet wipes are in. It's toilet paper, but wet.

No, but the wet wipes are made from like bamboo too or something like that. So it's like mine's plant-based. The Dude Wipes, they're expensive.

We don't use Dude Wipes, but we use another organic brand. I think it's organic for the wet wipes. I'm pretty sure everything in our house is organic, even though it doesn't mean much for food.

Yeah.

But we have bidets that clean you and then you have the wet wipes on top of that.

Dude, sometimes though, the bidets, We'll just leave it at that. I'm just going to send you this so you can buy. Okay, guys, let's bring this back to why A/B testing is dyke. So back to marketing, guys.

Yeah, we're getting close to the end, I think.

Yeah, we are. We still have 14 minutes left. So A/B testing was the gold standard for product decisions for 15 years. The best AI companies abandoned it. So this is from Akash Gupta. For a generation of PMs, experimentation meant designing a controlled test, allocating traffic, waiting 2 weeks for statistical significance, and hoping you had enough sample size to learn something. That loop trained an entire discipline to think in cycles of weeks. Eval's compressed the loop to minutes. 3 components: a set of inputs your product needs to handle. Number 2, a task that generates outputs. 3, and a scoring function that produces a number between 0 and 1. You run it on your laptop, no production traffic, no 2-week wait, no data engineering pipeline. The math on what this changes is staggering. Teams running evals are doing 12.8 experiments per day. That's roughly 384 per month. A traditional A/B testing team runs maybe 3. Over a quarter, one team has explored 1,150 variations, the other has explored 9. That learning gap compounds every single week. So the point of saying this, so he mentioned Ankur Goyal built the eval platform behind Vercel, Replit, Ramp, and Notion, $800 million valuation. He ran an eval from scratch on this episode, went from a score of 0 to 0.75 in under 20 minutes. That's a PM shipping a measurable quality before writing a single line of product code. So let me unpack this for people. So Andrej Karpathy, super AI researcher, he released Auto Research. I think these 2 kind of go hand in hand because Auto Research allows you to run compute experiments very quickly, right?

AI experimentations.

We're talking like every 5 minutes it's running one and he's just leaving his computer on and doing it. And he's found things that he couldn't— he's like, dude, over like 20 years of doing this, like he found a bunch of things that he didn't even know were possible.

Okay.

I think when it comes to evals, evals are just, it's just a simple way of saying, hey, what is defined as a successful test versus a non-successful test? That's all it is. Okay. And then you're just running it over and over, but we're going to see a lot more of this. And depending on what you're doing, if you're sending cold emails or if you're running paid traffic or whatever, you do need a good enough sample size when it comes to marketing. And you can't just, you know, say you're going to do 384. So I think it depends on your data sample size, but Now, when you set these agentic platforms up or your agents up, you're just going to say, hey, I want you to run these tests and I want you to every week, I want you to run evals and just so it can recursively improve over time. That's where this is all going. And that's where a lot of this stuff has already gone with AI, but not— it's kind of starting to encroach on marketing.

So this way he's talking about, we've seen this happen for years.

This is—

it's new to a lot of people, but it's not new in the Bay Area or the startup world. There's one issue with this model. It only works if you have high volume. If you don't have high volume, you can't do it. I think for most companies who aren't heavily venture-funded startups or large enterprises, they can't leverage this because they don't have volume.

Yep.

And a prime example of this is if you're a new entrepreneur, you set up a D2C e-com site, you're doing dropshipping, you're not even creating your own You're running Facebook ads and you're getting 5 sales a day and that's it. So 150 a month, you don't have enough conversion volume to run this playbook.

I think what's going to happen, so exactly what we're saying, but I think we're going to be using a lot more synthetic data in the future.

Well, if you think about, you know, MMM, MMM has been out for ages and And it's what people have—

What's MMM?

Media Mix Modeling. Okay, yeah, yeah, yeah.

Right?

So—

I thought you were talking about M&M's.

No.

Yeah, MMM, yes.

When you think about—

Yep.

And some of the solutions that Google offers and a few of the others, right? Meridian and the list goes on and on. When you think about it, you're taking data from what happened in the past to figure out what's most likely going to happen in the future. And it's saving a lot of time and money and companies are also starting to look at not just hey, where are we spending our money and where should we? They're also looking at incrementality in which—

Yeah.

Was this person actually buying because of the ad or were they already on the checkout and going to buy anyways and the ad just took credit? And that's another big problem in marketing as well because we spend a lot of money on advertising, but yet was the sale actually caused by You know, marketing or was it not? And dude, we see this all the time. And the way you see this is not what I'm talking about incrementality or you know causality. It's when you have the executive team just being like, "What would happen if we turn off all our ads? Let's try that for a week," right? That's their way of saying, "Did the ad actually cause or does marketing actually cause more sales for us, or is it just?" taking credit for sales that we would've already gotten. And it isn't that simple because if you just turn off your advertising and your marketing for a week, a lot of what you're closing in that current week came from people who saw your brand and from advertising and marketing that happened, you know, months ago or a year ago. So it of course isn't that simple. But you know, when you think about incrementality testing and you think about MMM, because of AI, It's more easier for companies of all sizes to use this stuff while it wasn't possible for smaller organizations in the past.

Yeah, I think what people can learn from this— so what Neil is saying is true, this is not necessarily new on the marketing side, but I think on the, the PRD side, so a product requirements document, this guy is basically saying that evals are the new PRD. And so when you have a very specific spec-driven document and it shows like it defines what a win and what a loss is, I think this is going to be built in a lot more to whatever agents I'm building, for example, whatever workflows that I'm building. So it's just on top, like, hey, I want you to have an eval here, right? That's just going to be built into whatever scaffolding that you have in the future. And that's going to allow us to do these things. Sometimes you might need synthetic data, sometimes you can't do it, right? But I think we are— I think me, for example, I've been required to— we were doing PRDs maybe like a year, year and a half ago. And like me, PRD, right? Doesn't make sense.

Yeah.

But now evals, when you just build it in, it reduces a lot of noise. That's what I think is going to— a lot more people are going to start doing it that way. Yeah. Yeah.

But yeah, with all these changes that's happening with data, you know, not just synthetic data, because I know a lot of corporations are using synthetic data too. And when I'm talking about synthetic data, let's say you're having— you're doing running Facebook ads and they want you to pass in data on your customers. Well, a lot of corporations these days are passing in synthetic data. So then that way Facebook doesn't have their quote unquote customer data. Facebook hates it because when you don't pass in your real data, it makes their ecosystem not as strong and it doesn't help them for future companies who are running ads there. But because of technology, you're going to see 2 things. One, you're going to see it easier for startups to get up and running and have the power that a lot of these enterprises had because things like MMM are much more affordable and easier to do. The second thing that you're going to see is it's also easier for companies to leverage things like synthetic data and protect their own first-party data, not pass as much through. So then you're going to have these platforms like Meta and Google not be as effective from a targeting perspective because they're not getting passed in all the data from corporations that they would like.

All right, so this, this can be a nice final topic for us. This is also from Aakash. So every CEO will be running an OpenCLAW within 12 months. So Mark Zuckerberg, Meta CEO, is building his own in-house. So a couple days ago, Jensen Huang stood in front of 30,000 people at GTC and said every company needs an OpenCLAW strategy. So now you have Zuckerberg building a personal CEO agent, and it's a Fortune 5 company, right? And so a CEO at Meta scale sees maybe 1% of information that determines whether a decision is or wrong. The other 99% gets filtered through VPs, chiefs of staff, dashboards, and whatever made it into the pre-read. The agent replaces the filter. Think about what a CEO agent actually does. It ingests every product metric, every internal thread, every customer escalation, every competitive intelligence report across every team simultaneously. Then it surfaces the 3 things that actually matter this morning before every one-to-one, one-on-one. It pulls that person's team metrics, open headcount, recent launches, and the 2 things they said they'd deliver last quarter. When the CEO asks what happens to our glasses timeline if we move 200 engineers to AI infra, the agent gives a first-pass answer in minutes instead of a 2-week strategy team exercise, and it never forgets. The person who remembered why the company killed that project in 2019 left 2 years ago. The agent didn't. So Meta employees are already running their own versions, tools called MyClaw and Second Brain. Engineering output is up to— up 30%. Power users up 80% year over year. Zuckerberg is doing what his employees are doing, applying to the highest leverage seat in the company. So this is exactly what I'm doing internally right now. I have everything connected and it's proactive with telling me. It does forget. Don't, guys, don't think this is perfect. Memory is an issue with AI right now. But I do think that if you're applying it at the highest level like this, for example, or even like the way I'm doing it right now, it cuts the noise very quickly and I can execute a lot faster. And I think we all like to move faster.

I agree with you this. I still have, I don't understand the concept why AI can't figure out the memory issue. It doesn't seem like a big deal. You tell it something once and for some reason it doesn't always follow.

Do you remember everything? You have really good memory.

No, but this is a computer. Like, it should be able to like hardcode and not forget unless you tell it otherwise.

You know what a computer is made out of, right?

Yeah.

Sand. Like, we conjured it and then we put electricity through it. I think it's going to get better over time. And I think the memory problem should be solved, I don't know, maybe in the next year or 2. But I do think just the decisions, like Man, it's so crisp now and I can move a lot faster. For example, last week I made that CFO agent and it's like, oh, you have $700,000 in cuts over here. I was like, wait, what? Yeah, found money, man.

Right?

It's great. It's great. I love it. So when I make a copy for you, you can use it for yourself. You can just duplicate it and use it for yourself. You can use your own scaffolding on it. It's my gift to you.

I was at an event a month ago and I met someone who's working Anthropic, and they're in charge of coding up a lot of this stuff. And I asked him the memory question and he gave me an explanation on why AI isn't perfect on recalling everything. And it was too complex. I didn't understand it. And I was like, when will it get solved? And because that's what I was really looking for, right? I'm like, I don't really care how you solve it.

I don't care about the technical details.

Yeah, yeah, yeah. And he's just like, he's like, it's not going to be easy. We don't see it happening in the short run, but it'll get better over the next few years. That was his words, not necessarily being solved in a few years, it'll get better and make the errors less often.

That's what I'm really saying. Like, we expect this to compound over time. And as we talk about this too, like, this is the worst it'll ever be. So, I don't know, my eyes continue to burn to the ground every night.

And I don't know about you, My eyes don't burn to the ground every night.

But they do if you use them too much, right?

No.

They don't? No, I never had that issue. You have LASIK, right?

Yeah, I never had that issue.

Yours do. Yours do.

Okay, great.

Noah's does. All right. So Neil's preserving his eyes.

Noah and I are being stupid. But yeah.

Anyway, guys, that is it for today. Please don't forget to rate, review, subscribe, and we'll see you tomorrow.