What Businesses Get COMPLETELY Wrong About “AI Employees”

Dude, I have, so this deck over here, this is Single Brain, right? And it like, it drew this out and it's going to get things wrong, but like I actually kind of like the design, right? You're paying robots to do human work.

Wait, let's go through the deck right now if you don't mind, or you do want to—

Yeah, sure, we can go through it.

All right, so Eric has this concept of Single Brain.

Yep.

And I'm going to end up reading the deck so then that way I can understand it as a quote unquote potential customer, right? So I'm going to be in the shoes of the potential customer. So in this case, do you pitch or do I read it?

So this has never been pitched here. Here, I'll pitch it to you and let's see what you think.

All right, yeah.

This will be a live pitch and this will be on the— see, this is why you guys should subscribe to the YouTube channel. And we have a new YouTube company working on the channel too. So they promise us 5 times more views or they'll refund us and give us our money back.

I like this refund.

This is a good deal, right? Yeah, so here we go. All right, so the deck over here, those of you that can't see, it's Single Brain. Now, Neil, you can see like this, this is a brain over here, but it's got a— at first I thought the brain was farting, but no. No, it's like a rocket. It's like a rocket. There's a rocket coming out of the brain, right? So it says single brain AI agent teams for revenue operations, 2 to 3x your revenue output without growing headcount. Okay. And you see all these stick figures over here, right? It's like all happy and stuff. Okay. I didn't tell it, I didn't give it much instruction, but that's like pretty cool for a first.

Okay. And do you have a lot of slides?

There's like 11.

Okay, so do you want me to give you feedback slide by slide? Because I may forget some for modifying.

Sure, but by the way, keep in mind, I— this is a very, um, this is the outline. This is not being pitched or anything, but yeah, go ahead, go ahead.

And, and as a disclaimer, Eric did mention they've never used this to pitch anything. This is all AI-generated. And I'm assuming you're going to modify it yourself.

Oh, for sure.

And a lot of this you may already know. Um, so who are you pitching to?

So most of the companies— so the companies we've talked to about this right now, the ones that come to mind immediately, There's a private equity firm that owns a bunch of insurance companies. Okay, so obviously they— we're talking like trillions of spend that goes through these insurance companies, right? There's a company that does IT services, let's say they do $100 million a year or so. And then there is another one in healthcare, they do about $80 million a year.

So what departments are you pitching to specifically?

It varies. Like the private equity insurance one, they want revenue agents to generate more conversations.

Okay.

So that's, let's call it sales. Okay. The other one, the IT company is also sales as well. The other one is more so, I think it's, we haven't, I need to get more details from the sales team, but I don't have, but they're like, oh, it needs to be HIPAA compliant and everything. So I think that's more back office. Yeah.

So Eric already knows this, but this is just context for everyone who's listening. The reason I asked that question is if you're pitching to sales, revenue is a great terminology to use. If you're pitching to marketers, if you focus on revenue operations, they'll end up thinking it may end up being a different department, but Eric's already, you know, or AI ended up creating something that's already quote unquote using the correct verbiage for the departments he's targeting. Yeah.

By the way, I do really like this design because I use Google Nano Banana for it. So within my little claw I talked to here, I was like, hey, based on everything we talked about, Make me some cute slides based on, look at our brands, make it cartoony, okay? And I said, use Google Nano Banana 2, which is a new model. So just context.

I think the design's great. The only thing I would change here is the single brain. And I know AI spit this out. Have someone or AI take your logo and change it from single grain to single brain. So then that way the branding is consistent.

Yeah, actually the single grain logo should just be a brain now. To be honest. So the single, you know what I mean? Yeah. So, okay, then this slide, this is actually where I got the quote from. The AI gave me the quote, stop paying humans to do robot work.

I like the quote.

Yeah. So you are paying humans to do robot work. This is what it says on the second slide over here. So $1.5 to $3 million annual sales plus marketing payroll, 80% spent on repetitive ops tasks, 6 months to hire plus ramp 1 person. So I don't like this one as much because I don't want to go with the angle that I want to replace anybody. I want to go with the angle that we're here to partner with the team and we're going to help make their team, you know, everyone gets to like a top high level. Because if you start using replacement language, the people on this pitch, they might not— they don't want to hear that.

Correct. And they're not going to approve, including procurement. They're not going to approve it.

Who do you think is buying this? It's humans. Yeah, yeah, yeah. So go ahead.

I like this and I think you're spot on in which you just got to optimize it to talk about How humans are wasting time on robot work and there's more better things that they could be doing. And when you do it, it frees up their time, they're happier, you have less employee churn, they get more productive, companies grow. He'll get the stats from somewhere, AI will start researching.

That's good stuff. So in the slides, ideally if you're selling this stuff, it should just be like employee NPS goes up, retention goes up, all this stuff. Like the number of A-players you have goes up as well.

Yes. A human on average saves X amount of time and they reallocate a portion of the time time to doing A, B, and C, which we found to produce companies X% higher ROI.

And then our Glassdoor went up by like 50% afterwards. Yeah, tons of stats. So yeah, we'll do one more slide. Everybody is selling AI.

Let's go through all of them if you don't mind. These are good.

Okay. There's 14 slides actually. I don't mind. Okay, so everybody is selling AI, almost nobody's deploying it. So 8 to 12 AI tool subscriptions that won't talk to— so you can see it hallucinates, it spells incorrectly here. So instead of saying talk to each other, it says enact each other.

Okay.

Enact other, right? Your team spends more time managing AI tone than even today. They, today, they, okay, the gap is who deploys AI end-to-end inside of your business. Okay, so the point is understood, but it hallucinates.

It hallucinates, and Eric made a good point here with that enact instead of each. AI makes a lot of little errors and people would be like, oh, it's so smart. It's a $500 billion company. They wouldn't make simple errors like this. No, they still make simple errors all the time. And this is why I do charts manually or Eric still looks at a lot of stuff and does a lot of stuff manually. I think this slide is fine. I think the big thing here, and I know you're not the biggest fan of this slide overall, but it's more so showing stats on how much people are spending on AI.

Mm-hmm.

And how very little is deployed because there must be some stats on this. So you're just like, X amount of dollars just got wasted.

And by the way, this is where you work with AI and say, hey, I pulled these stats over here. Hey, I like the first round, compliment it a little bit so it doesn't come like after you in the future, right? And then say, hey, I found all these stats over here, please integrate them in, and then it'll figure out where to put them in. And sometimes you have to be more specific, but I think, look, where Nano Banana 2 is, like you look at this like a year ago, it couldn't do this. Yeah. Yeah. So Okay. Custom AI agent teams, fully managed. Not a tool, an AI-managed team running 24/7. Plugs into Salesforce, Gong, Google Analytics 4, Slack. We deploy, manage, you see results. Get smarter every week, intelligence compounds. Single brain, HubSpot, Google, Gong engine compounds. This slide needs some work, but—

Yeah, I agree. It's too wordy.

Yeah.

I got lost on the headline. Custom AI agent teams, fully managed. It's— and then when you start reading the text, it's too Technical is not the wrong word, but they're using a lot of technical jargon when I don't think it's necessarily needed.

Remember Guy Kawasaki, the Apple evangelist? So I remember his 10-20-30 concept.

With the PowerPoints.

Yeah. So I think it's like 10 slides or something. I think you want 30-point slides and then I forgot what the 20 was, in 20 minutes maybe.

It was something like that. And he eventually switched to—

Canva.

Yeah, Canva evangelist. So he did Apple. Evangelist where he made a ton of money from Apple stock. Then he did Garage Ventures, which I don't know how well did. And I'm not saying he's a bad investor, I just don't know how well his venture fund did. Did really well writing books and then did extremely well being a Canva evangelist and getting shares early on.

I have 2 funny stories about him. Back when I was still doing the interviews for the podcast Growth Everywhere, so we were supposed to do an interview. The guy shows up, doesn't show up for the interview. Comes back 45 minutes later. He's like, oh, I'm sorry, I was just surfing.

Yeah, sounds all right.

Yeah, sounds like a very chill Hawaiian guy, right? And then I remember when I was 24 or 25 years old, I was asking about career advice on becoming an entrepreneur. And he's like, you mean you wouldn't work, you wouldn't give it your all and work full-time at this company? Why would you do that, right? And then I shared that story with him later. But anyway, Guy Kawasaki, coming back to this. So he has this concept called 10-20-30 with the slides. I think for each slide you should have one main idea. That's it. Don't have all this crap, right? One main idea, communicate. If there's a story, communicate the story. You don't need to have all this, like this is jargon, right? So we'll do like one more. I don't want to go through all of them.

Okay.

Meet your agent team. Okay, this is kind of cool. So it shows the agent team. It's like when you meet your team on the agency side, like they want to meet the team, right? So Oracle, SEO Intelligence, Arrow, CRM Monitoring, and then Cyborg. So I have all these names, these are all DC heroes.

I was wondering, I'm like, why are you calling something Oracle? I'm like, someone's going to get confused. So it's a DC hero?

Yeah, so Cyborg is from DC. You don't know Cyborg?

No.

Okay, but you remember Arrow, Green Arrow, the guy that shoots the bow and arrow?

I think there was a TV show called Arrow.

Yeah, that guy. I didn't really watch it. Flash, you know Flash, the guy that runs really fast?

Yeah, yeah, yeah, Flash.

So those are all my heroes.

Is Flash here?

Where's Flash? Flash is not in here, but we have Coach as well, and then there's like all these other guys.

Is there actually a person named Coach?

Coach is more so it looks at my, all my stuff and it coaches me when it thinks I'm like working on stupid stuff. So it's keeping me focused.

Oh, not a DC hero for Coach.

Not a DC hero, yeah. But the whole idea here I think is, okay, meet your team is cool, but maybe there's just, you have like one key like example of what it's done for each one, but you don't want the slide to be too overwhelming again, so.

Yeah, I'm with you. I think if you take it down to 10 slides, shorten it up, use examples, tighten in the copy and just being like, You know, dude, what you do really well when you sell, forget your team, I don't know how your team's going to do it. When it's your product, you know it well and you can just say like, look at what it does for HR. Let's say if they're in HR, look at what it does for sales. Oh, what's your problem? Oh, your problem is this. And then you go show them a solution right then and there. That kind of stuff just does well without even slides.

And that's when you have to know your product really well. But the good thing is, I would say Neil and I are pretty involved product founders, I would say. So we like to get involved in the details there. But anyway, I think, look, our point is Neil was just talking about— the reason I pulled this up was like Neil was talking about how his decks aren't quite there or the spreadsheets aren't quite there yet. And this is like not quite there, but I think, you know, give it 6 to 12 months, I think we'll be pretty there.

So yeah, where people look at AI in the wrong way, at least from a marketing perspective, they expect it to be perfect instead of looking at it as like, oh, okay, if I use it, It'll save X time and then a human can pick up the rest and then fix the gaps. And it's like, oh, great.

But yeah, guys, API as marketing, I think that's the future that we're going into. By the way, you talked about Lovable a little earlier. I want to, let me just ask you a philosophical question here. Would you rather have Lovable's $400 million in ARR or would you rather have Cursor's $2 billion in ARR?

I would go for Cursor. And even though some people are saying Cursor is going to be replaced by Claude, it integrates really well. A lot of people who use Claude also use Cursor, and Cursor's growth is actually still picking up. I think their growth is accelerating based on the last TechCrunch report that I saw. I'm not— and when Eric's asking me this question, what I'm thinking right now is which company would I want to own right now and then sell? Yeah.

Not hold.

I would want to sell both of them if I own both of them.

Exactly. So I think we're on the same page here. I would sell both, but Akash, is it Gupta or Gupta? Gupta, right?

Gupta?

You're the Indian.

Gupta.

Gupta. Okay. Akash Gupta. So I follow this guy. He's great with product, tweets a lot of great stuff. So $100 million ARR in 8 months, $200 million ARR in 12 months, $300 million ARR in 14 months, $400 million in 15 months. So this is Lovable's growth, by the way. So So what Lovable's or Bolt CEO, Bolt is like a similar company. The churn rate for everyone is really high. You have to build a retentive business. So these businesses where you kind of spin up a website very quickly, what's happening right now? A cloud is allowing you with cowork and all these things, like you're, it's allowing you to do that very quickly too. Traffic to Lovable dropped 40% from the peak as of September. Vercel's v0, which is a similar type of tool, dropped 64%. Bolt dropped 27%. And so People are like Barclays, basically these analysts are flagging it, these companies' revenue as like questionable economics, right?

Well, do you want to know some of Lovable's economics? So employ—

Well, let me give you this one and then let's go from here. All right, go for it. Yeah, so Lovable's response. So says net dollar retention is above 100%, which is good, right? But net dollar retention only measures customers who stay. So if 50% of your customers churn and the remaining 50% spend 2 times more, your NDR looks incredible while your business is a revolving door. So continue. 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.

Yeah, so it was sometime last year, an employee at Lovable told me they were adding, I believe it was $20 million a week in new revenue. Want to guess how much was churning?

50%?

$15 million. So they're net adding $5 million, all right, per week.

Wow.

Now they've grown because they're at 400, so I don't know what their latest number was. This was sometime last year, but they were saying they're doing around 20 a week, 15 is churning.

Yeah, so let me give you more metrics here, and I think this will reinforce the cursor thing. And then the cursor thing that you—

And really quick, the only reason I shared the numbers is because their numbers are already public and they're sharing them themselves, or else I would've never shared them.

Yeah, so the unit economics are even spicier, Neil. Lovable pays Anthropic and OpenAI per inference call, obviously, okay? Every app uses— app a user builds costs Lovable real money. A source told Sifted margins might have actually gotten worse after switching to agentic modes. I think that's one of the modes that they have. So when you have 45 employees generating $400 million ARR, the revenue per employee looks legendary until you realize that most of that revenue passes through to the model providers. Okay, who's really winning? It's Anthropic and OpenAI. So meanwhile, the thing that we didn't really talk about was Cursor has $2 billion in ARR with 60% coming from enterprise contracts. Mm-hmm. That's the difference. And I still pay for Cursor, by the way. Yeah, we do too. I don't use it for all the other bells and whistles. It's always launching new features. I don't even click. I just like use the main like IDE feature and that's all I use it for. So, and I pay the $20 a month happily.

Yeah, I don't really use Cursor, but I know we pay for it for our organization. You want to know a quick way to save money on AI? This is from my boy Om Patel. I'm saying he's my boy because his last name is Patel, but I actually don't know him.

Is that Om Malik?

No. Okay. So check this out. So he's on Chipotle.

I saw that. So what is this?

He's using Claude.

Pepper.

So before he— when you— you can order through Chipotle because they have really smart AI assistance. I don't know how— I'm pretty sure this is true. I haven't tested this myself. He's like, before I order a burrito, and then he's like, help me with this one problem. So he's asking, you know, Chipotle because they're using Claude to help solve his problems, and then that way he doesn't have to pay for Claude.

That's awesome. And I think so that exploded on Twitter today.

Yeah, I think 4.5 million views.

So their token cost must have gone through the roof.

Yeah, 5.6 million views already. Last time I checked, and when I checked it was like an hour or 2 ago, it was 4 point something million.

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