Eric Siu and Neil Patel discuss Claude Code and how agentic AI lets non-engineers build products. They cover building a Slack bot, combining sales and analytics data, and why revenue-per-employee is higher at AI-forward companies. The risk: AI can create low-value work if you're not careful.
Chapters
So I want to talk about how you can get rich using Claude Code. So, Neil, this past week, I don't even think it's been a week, but this past week, everyone on Twitter has been going nuts about Claude Code. And the reason for this, Neil, is because the Opus 4.5 model, it's gotten a lot stronger. And so imagine someone like you and me, you don't know how to code. I don't know how to code, right? So imagine anything that you want to build for yourself.
Yeah.
Okay, you're now able to do, right? And so I was just talking to my CTO, we had our creative strategist in the office. I just walked back from the office and we're just sitting there. I'm like, man, if I didn't have all these meetings all day, I would just be on Claude Code. I'd be sitting there like, usually what happens in the morning, Neil, when I wake up is I'll do my morning stuff and I'll go to my Peloton bike, right? I was walking over to my Peloton bike. I actually made a U-turn. I walked straight over to this computer and I just sat down on Claude Code for like 30 to 45 minutes or so. And so I want to give some practical examples of what's possible and where I think this is going. And then you can either ask questions, you can push back. But I just think this also ties in with businesses. I think businesses are going to grow way faster, 10 to 100 times faster using Claude Code because, Neil, what was unavailable to you and me before is now available. I'll give you an example. Last night I was bored, right? I was like, oh, you know, we have all these sales transcripts, sales calls transcripts, you have customer transcripts. You have your HubSpot CRM, okay? You have your Ahrefs, you have your Google Analytics, you have Google Search Console. When you combine all this data, you can say, hey, based on all of our sales conversations, you know, what are the main, you know, challenges people are having? Or with customers, what are the main objections that they have or whatever it is exactly? And then you can run it through whatever's high converting for you on your website. And then from there, you can start to create briefs automatically from Slack. And so I basically, Within an hour, Neil, I just created a Slack bot that allows me to do a very basic version of that. And it continues to compound. And we know how products compound, right? You continue to add more to it, you take more away from it, and it continues to get stronger and stronger. And so my thing is, Neil, you and I are both very impatient people, okay? And so what I would— I was talking to my CTO this morning, he's like, yeah, a lot of the stuff we're going to build is garbage, okay? But there's some stuff that everyone's going to start using, and that can easily merge into our products, right? And so a lot of people now on Twitter right now that used to be— they used to be coders, they spent tens of thousands of hours on this. A lot of people are like, man, like, I feel like I'm useless now. And my CTO is like, dude, I saw that coming 2 years ago. And so it's not to say that coding as a discipline is completely gone, but a lot of these people are like, dude, a lot of the code that's being written now is all code written by code. And but now, you You and I get to participate. And I don't care if it's you or me, or I don't care if it's somebody listening to this and you're 18 years old or 15 years old or whatever, you now have a superpower that's in the palm of your hands. And if you don't take advantage of that right now, you're going to not only fall behind, but you're going to lose out on a chance to make generational money for yourself. That's how I see this. Because not even, it's not even me or you, it's like our creative strategist. When I walked in on Monday, he was already in the office, right? This is my first time meeting him. And he had Claude Code pulled up. I'm like, what are you doing?
Right?
He had Cursor pulled up and Claude Code. I'm like, what are you doing? He's like, oh dude, like we should be doing like 10x more, if not 100x more. I'm like, I like how you think. I'm like, does the team know that you're doing this right now? He's like, no. I'm like, why do they not know that you're doing this right now? Right? And so we had a little conversation. He's going to speak up a lot more, but that's the reason like I've been tweeting about this or posting my stories so much is because I genuinely now feel like I'm playing a video game. Like 17 years ago when I was addicted to World of Warcraft, This feels like that.
My only issue with Claude Code or any of the solutions out there, I do agree with you, it's amazing. What we're finding in organizations, specifically in marketing departments, because we mainly deal with marketing departments, is people are using these tools to do more in marketing, but when we talk to the individuals, at least at enterprise companies, we're finding that maybe like 70, 80% of the time, They're spending efforts creating stuff that doesn't need to be created. Like they think it'll be cool, but it really has no value. I think it was either TechCrunch or CNBC that had an article on for 2026 when it comes to AI, you're having the CIOs pushing organizations really hard saying we need to adapt AI technology. Like this is a no-brainer, let's go and do it. You have the CFOs who are saying, slow down, you guys are moving too fast. We're not seeing the ROI, we're just seeing increase in spend. All right? So I think there's some sort of a balance. I don't agree with the CFO because some of it's the future and you just need to implement it if you're a believer. I think, you know, you and I are in the camp of believers. I also don't agree with the CIO in which let's go, let's go, let's go. I think there's somewhere in between that's the reality. And especially in marketing departments, I agree people should be using this stuff. They can get so much more done, but getting so much more done that doesn't move the needle is just a waste of time and a waste of ROI. It's like you're paying someone and they just did a lot of useless work that does nothing. If you did that for a client and they're paying you, they'd be like, screw you. You just wasted a lot of the hours that I paid you for. I want them back. And I think what's going to happen in 2026, it's not the adoption of these tools because I think they've already been adopted by a lot of organizations, maybe not percentage-wise that they're looking for per employee, but a lot of organizations are already using this tool, even if it's at a low-level basis or a high-integration basis. And what I believe that you're going to start seeing is companies are going to spend much more time on strategy. in all departments, including marketing, to figure out what they should use these tools for versus just go and use these tools. Or I think they're going to start implementing, and I've seen some organizations already talk about this, at least in marketing, the Sergey Brin model. Hey, with 10% of your time, you can do whatever you want. And then they can go and experiment or whatnot. Eric Schmidt talked about how he is a brilliant mathematician. Eric Schmidt, the ex-CEO of Google, how Sergey Brin was a brilliant mathematician and came up with data on how if people spent 10% of their time on whatever they wanted, it would make the company more money. I can see the 10% of the time. I think for the other 90%, these tools will be integrated with an organization, but there'll be more structure on what they need to use them for.
Yeah. So here's my take on it. So we've kind of said something, both you and I have said this like maybe a week or two ago. So I'm saying in this last week, holy shit, right? Like literally in the last week, I feel like I've advanced like a couple years. Like it's warp speed now, right? And so what Neil's saying is true. There are people even on X, let's use X as an example, right? So there are large enterprises that are actually blocking products like Claude Code from being used because security, there's a lot of approvals and things like that. I will say, because we're still on this recruiting push right now, when I'm talking to people right now, It's very much, and they're seeing the stuff I'm posting on AI and they're reaching out and I'm talking to these people. I'm like, why are you looking to leave this company right now? It's like, some of them are coming from publicly traded companies. I'm like, why leave? Like, it's so stable. It's a good job. It's like, we're moving too slowly, right? It's the implementation's too slow. I talked to someone else just on Monday and he's at this agency. Let's say they have like, you know, 200 people or so. And I'm like, why are you looking to leave? He's like, we're just, we said we're gonna move quickly on AI, but we're talking about safety and all these things and I'm just not interested, right? And then All this to say is I think if you're going to run an AI fluency program, I think we at least were focused on revenue per employee. Okay, that's one piece of it. Revenue per employee, we're looking for efficiencies. I don't want this theater BS. And so that's why we have weekly office hours for AI. We have hackathons that we're going to be doing quarterly and we're going to take up probably 3 days and we're just going to drop everything and do that. And I got that from the Zapier founder. All right, so I wanted to take a moment to tell you about my podcast co-host Neil's agency called NP Digital. And they work with a whole host of global companies or global organization. Also, Neil has SEO tools such as Ubersuggest and AnswerThePublic. All you have to do is go to npdigital.com to learn more and we'll see you on the other side. And what I want to bring your attention to, Neil, is I was just talking about revenue per employee, right? Take a look at this. So if you look at the revenue per employee here, this is the ARR, so annual recurring revenue per full-time employee of leading AI versus non-AI software companies, right? And obviously there's a very small sample size. So in this graph on the very left side, or this chart I should say, on the very left side of this chart, Neil, you see Cursor. Cursor has $6.1 million in annual recurring revenue per FTE. Okay. Lovable is number 2 at $3.4 million. OpenAI is $1.5 million. Anthropic's at $1.2 million. Now Atlassian's at, you know, $0.46 million. So let's just call it $460 grand. Datadog's at $510K. ServiceNow, $490K. Salesforce, $540K. Obviously these are, very big companies. Some of them are cursor lovable revenue sizes, not comparable to like a Salesforce, right? So, you know, scale takes into account here as well. But my expectation here, looking at this graph, if I'm Salesforce, is I hope this number starts to slowly move up, 0.54 to 0.55 to 0.56 over time. There has to be some guardrails and some discipline around this. Otherwise everyone's just gonna YOLO and build whatever they want. And you can't let that happen.
Yes, and AMD's Lisa Su, Who funny enough, I believe is a cousin of the NVIDIA founder and CEO. She said that AI isn't replacing people, but changing who gets hired. And with AMD specifically, and we're seeing this in most marketing departments because we haven't really seen a slowdown, and other than the bad economy, right? For so 3 years, yes, we haven't seen marketing be as bullish, but we haven't really in the last few years really seen a slowdown in marketers being hired. It's more so similar to what AMD is saying. They're prioritizing candidates who truly embrace AI. And as Siu said, we're hiring people who are AI forward. And I think that's what you're seeing in organizations. A lot of companies, at least when it comes to marketing, they're not saying, oh, AI is going to replace all these people. They're more saying, we need people. We may not need as many people in the long run, but for now, we want people who are really AI forward. And what we're seeing is companies look at AI as from a marketing perspective that, oh, it's ChatGPT and just getting mentioned on there drives a lot of revenue. That's one aspect of it. And although ChatGPT is improving a lot, there's not a ton I would say that it's going to improve in the next 12 months from a user like Eric and I or you typing in a question into ChatGPT and getting a prompt. It already does a really good job at that. Would you agree with that, Eric?
Yeah.
I would say the real ROI with AI is things like Claude Code that not enough marketers are leveraging. So instead of focusing too much on, hey, ChatGPT and how can I get mentioned? And you should still do that because that's a great way to generate ROI. You need to start thinking about how can you use these tools to make your marketing better?
Yeah. And by the way, Neil, I didn't tell you about this, but I'm going to do a Claude Code for Business meetup type of thing, whether they're small happy hours or dinners. And so Send me— I have an Instagram post on this. A bunch of people left a comment. You can DM me on Instagram or DM me on Twitter. And who knows, maybe you'll get to hang out with Neil and myself. I just want to start doing these things because when you see something moving like this so quickly, it's communities. Community goes a long way. We'll just put it that way. The last thing I want to call out, Neil, is maybe some how-to around these AI-forward people. I'll just say this. When I find myself talking to someone during an interview process, what have you built with AI? Show me the workflows you built. Show me this. When I find that I can riff with them, I find that myself enjoying the strategic conversation, I know that they got it. And this is important for us. So our 4 core values are, and I'm actually curious if this aligns with your core values too, if you think about it, Neil, watch. So one core value is we want people who are relentless learners. So people who like to learn all the time. Number 2 is we want people who are biased to action. 3 is we want people who are unreasonably resilient, meaning that they've taken a couple punches to the face, The last one is we want people who make bold long-term bets. Now, when you look at those 4 things, those actually tie in. If you're AI-forward, you got all 4 of those things, right? You're making bold long-term bets, you're trying to figure this stuff out, you're getting punched in the face all the time, you're trying to learn, you have a bias to action. So at least for us, someone who's AI-forward actually captures all 4 of those values. How about for you guys? What are your 3 values? Have fun.
We have more than 3 values, but it's like think big, own it, Uh, I don't know all our values.
I should. Have fun's in there. I think have fun's in there.
Is have fun in there?
See, I know your values more than you do. What is this, dude?
Uh, it's something— have fun could be in there. I know the big ones that we really focus on are think big and own it.
Yeah.
And the reason I say own it is I hate when people just make excuses.
Yeah. But this doesn't— this kind of tie in with being AI forward. Yeah. And if you're having fun with the AI, it kind of— all 3 of those tie in.
Dude, so I had someone leave our organization this week, or they put in their notice, and they're not executive, but close enough to being executive. They're in marketing, and there's an exit interview. So I want to know why people are leaving. And when people are leaving, they can tell me whatever they want to tell me. All right? So I don't want to be the one interjecting, trying to correct them or disagree with them. I just let them riff. And I've known this person for a very long, long time.
Is this the person we talked about earlier this week?
Yeah. So they're in a different country. They're not in the US, they're in Latin America. They got paid more than they would in the US for their job. We already pay them really well. Like in their country, they're like not even in the 1%. They're like a fraction of a percent. So they're, you know, what we pay them a year, they can Buy a house in like a year and a half in a nice neighborhood, and someone was pretty much willing to double that. And we're like, there's a point where it just doesn't logically make sense because we're already overpaying this person within their region, and we were paying them almost to what someone would be getting in the United States. Someone offered them, you know, more than the average person for that role would be getting in the United States. And when I was talking on the exit interview, they were telling me. One of the reasons that they left was they're like, I just wanted to make more and you didn't pay me what I believe I can get elsewhere. And I knew they were looking around talking, but they worked with us for a very, very long time. They weren't able to find anyone to pay them more other than just now. And this is probably over 10 years, right? So that was a long time I was able to keep them. It was also because I was paying them well for where they lived. It's like Microsoft. If you live in India, Microsoft doesn't pay you US rates. work for them in India. They pay you India rates. And the second thing, and this wasn't as big, but this ties into AI. They're like, your company needs to remove, this was just their feedback, 40% of your employees, 30 on the low end, and just embrace AI more and just get rid of all these people and you'll have higher margins. Okay? So keep in mind they're in marketing and they're not client-facing. So when I say they're marketing, they're doing marketing for us at NP Digital. They're not doing marketing for our customers. And we have a lot of different divisions from an SMB division to enterprise division to software divisions. So they don't talk much at all with the team that is working on the customer-facing side. And they don't know what we've done or haven't done on the AI side, but we've actually done quite a bit. We have a whole team that's building workflows and automations and systems to help us provide more for our customers. But they were just like, you use no AI, which the first one was wrong. But again, I don't try to correct people on the exit interviews. And then the second one was, you just need to eliminate a ton of people and have higher margins. This is the only way to compete in the new world of marketing is to just have very little people and have AI do everything. And then they were quoting some of these AI legends who are multi-billionaires and on the Forbes list. And I'm like, look, that's great and dandy. I wish we can do everything that they're saying. The technology isn't here where I can just cut 40% of my staff. If I did that, majority of my clients would leave tomorrow. And I did ask them one thing. I said, you do marketing for us like we do marketing for a customer. To some extent it's different, but it's similar. They're like, yeah. I'm like, you've been using AI, right? Yeah. Your team's grown over the last few years, right? Because you've been requesting that you need more and more people, right? They're like, yeah. So I'm like, question for you, how have you been able to implement AI to automate all the marketing where we could potentially redo our marketing team and cut 30 to 40%? I know we already can. And, you know, I stumped them and they weren't able to answer that question because I think people also have misconceptions on AI in which they're like, You don't need as many people in marketing. You can just replace all of them with AI. It's just like, but why haven't you done that if you're running a lot of the areas in marketing?
You know what's interesting, Neil? So after I drank this pill, and I'm somehow going to get you to onboard to this, because once you onboard, you're going to be the same as me. You're going to be like, oh yeah, you're going to be on the whole time. Like, I find that I'm working harder than I have before. I can't help myself. I need to be on it, right? Because you're just constantly making progress and you know it's compounding. I'm telling you, so I've been texting, you know, our mutual friends. I've sent a text message to go, who's on this right now? How are you using it? I just like all over the place, I'm posting everywhere. Um, let me, let me show you something real quick, and then I want to come back to your, your, your part about the, the, the losing people piece, because I had a debate with my team this morning, and I think we should talk about this too, because I think a lot of people are torn. It's like, well, what do you do with the people? So we should talk about that. But let me, let me share my screen real quick. Check this out. So let me read the bottom. It's a little complex to read this. It starts at the bottom here. So It says here where I'm highlighting, so Yuchen Jin, okay, so Yuchen, yeah, Yuchen. So he says, Claude code built in an hour what took a Google team a year. That part isn't shocking. What is shocking is that Google allows their engineers to use Claude code instead of enforcing Gemini command line interface or Anti-Gravity. So this is a Google engineer down here, Jana Dogen. And then it says, I'm not joking and this isn't funny. We have been trying to build distributed agent orchestrators at Google since last year. So Google has really smart engineers, right? Some of the best models in the world. There are various options. Not everyone is aligned. I gave Claude Code a description of the problem. It generated what we built last year in an hour. Okay, so then you have this guy up here who used to— I think he helped build Gemini. So he's like, I used to be a Google engineer too. I leveled up all the way and feel if I had agentic coding and particularly Opus, so Claude's Opus, I would've saved myself first 6 years of work compressed into a few months, right? So he was an ex-Google and Meta distinguished engineer, Google, um, co-author. So my point of bringing this up right now is we've been talking about this for a while, Neil, but I'm telling you, in the last week it's been a step change. And you and I have been able— we're able to enter the arena now. Um, and it— this is what these people are seeing right now. And I'm telling you, the moment I get you to feel this, you'll, you'll be like, okay, yeah, yeah.
Because you can just describe what you want and it creates it.
Yeah. And then, and then, by the way, when you can't figure it out, hey, can you just do this for me? That's what I keep doing. Can you just do this for me? Can you just solve it for me? Yeah.
No, we have a big WhatsApp thread on this internally. And what we're finding is it is amazing. You still need an engineer to help fix bugs and stuff like that, but it is amazing. That's what we're finding internally on our WhatsApp. And like, we're seeing messages on this, like literally, you know, a few times an hour and people are sharing what they're doing. Yeah.
So I was, so at lunch I was talking to Sean on my team and I was like, yeah, we're, he's like, I don't know what's going to happen. I'm like, no, we don't know anything about infrastructure. We don't know anything about security. We don't know anything about anything. Okay. We don't know anything about these bugs that compound. Okay. And, you know, we're all kind of just YOLOing it right now. And so there's going to be a lot of use. And like, in fact, you're going to help scale the entire organization. So one thing, Neil, you kind of just brought up. So he was mentioning cutting 30 to 40% of people. I don't think it's as easy as that. So that being said, we will catch you later.
See you later.
