Eric Siu built an AI tool that watches his screen to identify skills and inefficiencies in his work. He's rolling out AI apprenticeships and training pods, and discusses how AI is changing hiring and marketing workflows. The conversation also covers AI overviews citing brands and the broader impact of AI on how people work.
Chapters
All right, everyone, we are live. Neil's in a new home finally, and we're just joking that the last time he had a nice, really nice background, it lasted for like a month. And this time he has a nice background again. So, you know, Neil has this issue with business. He's very focused, but when it comes to his homes, he's very unfocused. And when it comes to his car, he's very, very unfocused too.
You know, I bought a new car, right?
You got the Model Y, right?
Yeah, it's the best thing ever. From a self-driving perspective, shittiest seats I've sat in a long time for a car.
That's all you complain about is the seats, but you miss out on like all the good things too. The fact that you don't have to drive is great.
But I have back pain, that's why. Oh, so like I notice it within like 15, 20 minutes of sitting. Um, it's so much so that I've been using the self-driving and I started researching every single car company that has self-driving technology out there. I couldn't find good results. So then I plugged into LLM and it's breaking down how Mercedes is going to do a good job. And I like Mercedes, their cars are really comfortable. And what I started doing was I asked the LLMs to find me, or Google to find me every YouTube video of someone test driving the Mercedes self-driving. And then it ended up giving me results and I started watching the video and it was making so many mistakes. I was like, oh, I can't use the Mercedes self-driving technology.
I think that the lesson to learn from this one from a business standpoint is that even just a philosophical standpoint, so this comes from Thomas Sowell, who's like an economist, right? And I think he's still alive today, but one of the quotes that really sticks out to me is, There are no solutions, only trade-offs. And so you might say, okay, the Mercedes seats are a lot nicer and it looks— it's just more luxury in general. Okay, you get that. But then you miss out on— the trade-off that you miss out on is the cream of the crop when it comes to self-driving, right? And then the trade-off is when you drive the Tesla, you get crappy seats, right? And so if you look at everything across business, almost everything that you see, there's a trade-off at the end of the day. There's no perfect solution. There's always a trade-off.
But you have to figure out if the trade-off is worth it. And I think that's where a lot of people mess up. So like for me, I have back pain. I can't do the trade-off of a Mercedes and use their self-driving for no back pain because the trade-off in this situation is self-driving technology that may get me in a crash and may kill me because based on the videos I saw, it's not that great. And the same problem goes in business. When people pick trade-offs, you may prefer one over the other, but you got to really analyze the consequences and figure out, can you even— is this consequence something that you can even afford to have?
Yeah. So by the way, Neil, I just came back from an AI mastermind last week in Costa Rica. I thought I could maybe start off by sharing experiences and you can talk about cool stuff that's happened in the last 7 days.
Let's do it.
I was in Costa Rica. By the way, I've never been to Costa Rica before. Have you?
No, I've never been to Costa Rica.
So Costa Rica is really nice. So one of the guys in the group, his name is Alan, he recommended Costa Rica and he did a great job of planning the event. And so I'll reveal what I can. I can't go into too many details, obviously, but I'll just talk about maybe the setup for this thing and how I think everyone should think about at least forming an AI text group or even like an AI founders chat or like a whatever role that you're in, I think it's really helpful to have peers. And maybe that can be like a follow-up to this piece. But I have always talked about why it's helpful to have a forum. Like, these are like-minded people.
Okay?
So Neil might have like a group of people that he texts all the time. And I, you know, Neil and I, we text all the time as well. You want to have people that you can bounce ideas off of, right? And that's always been very helpful. And so that's why I've always been a fan of like groups like EO, Entrepreneurs Organization, or groups like YPO, Young Presidents Organization, or even Hampton, for example, Because they help facilitate that. Now, what I would say is I was lacking a forum that would just primarily talk about AI. All we wanted to nerd out about was business and AI. We don't want to talk about partying. We don't want to talk about anything else. We don't want to talk about our relationships. All we want to talk about is AI and business, right? And so we went to Costa Rica basically last Thursday to Sunday or so, and it was beautiful. So the whole idea here is that, you know, you do it in a nice spot. We had like a big house and it was like overlooking the water with a nice pool and everything. And just every day we'd wake up, we'd have someone cook breakfast for us. They were amazing. And then from 8 AM to 12, we'd just be talking about business and like AI shares and things like that. And then afternoon we play pickleball and then we probably—
Wait, let's go back. AI shares. So you guys were talking about like stocks and all that kind of stuff and which shares you guys own?
Oh, no, no, no. AI shares as in like we're sharing things around AI and business, not the stocks.
But we did talk about stocks.
Yeah, it's like the stuff that I showed you already. You saw the stocks that jumped like 8,000% right in a year. It's like that. Yeah, yeah, yeah. We did talk about investments. We did talk about stocks. I think it's just really anything around business, right? And I would just say that going into some of the AI learnings, at least from my end, one thing is after I came back, I started off this apprenticeship at my company. And this was a result of one of my explorations that I did where I was talking about some of the challenges I was having, right? And so, I noticed something with some of the people that I trained up at the company. I would spend like, I would do these 30-minute meetings and I would do these meetings like 10 to 15 times or so, okay? But after I did these meetings with these people, these people were flying and I didn't have to tell them much. They were just very high agency people and they were just figuring everything out and they were doing the jobs of like 4 or 5 different jobs at once. I'm like, dude, Why don't I just figure out how I can duplicate that across like more people? Why don't I like Jedi train some more people in the organization so then they can Jedi train more people in the organization and spend time like I was doing too, right? In terms of scaling it out. So, um, we created this pod of one apprenticeship and I already started rolling it out this week. So that's one thing that I immediately did. I'll pause for a second here.
So some of them have already been doing this, right? Where they rolled out this pod and doing this apprenticeship. Like, what are people getting from it? Like, what's What are the results that people have been experiencing that have been doing it for longer? I'm making assumptions here, but I'm assuming some of these guys have been doing it for longer because you guys are all sharing new ideas with each other.
Yeah, we're always sharing ideas and things like that, but I think when I'm actively spending one-on-one time with some of these people that are very talented and AI-forward, and I'm kind of guiding them along the way in terms of kind of my speed, my expectations, and teaching them along the way, that's why I say it's like Jedi training. They tend to move a lot faster because I asked one of the people that I trained, I was like, hey, how impactful was it? Because I don't think anything of the time I spend with someone, right? And this guy was like, oh, it was honestly like a 9 out of 10. It was massive for me that you spent the time doing that. And I was like, okay, I'm going to pick 6 other people in my organization to do this with, and then they're going to train the rest of the people, right? In terms of just spending a lot of time with them. And so, I've been doing that this week and it's already been a high-leverage thing already where I'm like, whoa, this is really good. And I haven't thought of doing like an apprenticeship before, but this is what I'm thinking about. So when we talk about this AI apprenticeship, if you want to join my company, we're hiring more. I'll actually spend time with you if you're that damn good. If you're not that damn good, I will spend no time with you.
Well, if you're not that good, forget the time. Are you going to even actually hire them?
Well, we make hiring mistakes sometimes, you know, like how often do you get hired right? That's the question.
Yeah. I looked at my track record. Last week we were actually going through every single company I've been through and looking at how many times I hire a good manager that works out. I'm talking about a manager who typically leads a division or a project, whatnot. And this is literally 20+ years of doing business. I think I'm what, 25, 26 years of doing business at this point. I'm a little bit better than 50%. I was just like, oh, maybe I've gone better. Maybe I'm at 70, 80, 90%. And I'm talking about we have really strict hiring processes. I've done much better later on, but I'm still not batting like 70, 80, 90%. It gets really hard when you're hiring a CEO. I'm not talking about lower level or middle management. I'm talking about senior management looking to lead a country or a division or do something new. Roughly 50%, sadly.
Yeah, that's the thing. I think I'm probably around 50%. I need to run the numbers on this. I might be around 50% or maybe even slightly below 50%, but I think we all like to think that we're really good at hiring when we're not because we're all different people and people are very complex and we all have different personalities. Like, Neil's a very different person than what I am. I'll share one more thing, Neil, because I don't want to take up too much airtime here. One other thing that we rolled out ASAP is, so I built this using Codex. And so, you know how Meta right now, they're basically, they're kind of seeing how people are, their employees are doing things and they're making processes with it. Or like if I, some factory companies are making you wear a suit so they can see how you run certain processes when you're building things. Are you familiar with this?
Yes.
Okay. So I made something. So we're talking about this, but I basically made something using Codex that It's been watching me for this last week. So, it'll take screenshots every now and then. I don't even know what screenshots it's taking, but it's watching my screen and it's telling me, hey, this is how you work right now. These are the skills that I think you should be making, meaning that these are processes that I can then— these skills I can hand off to other people and make them very portable. But it's also telling me in which ways I work very inefficiently and how I can work better, right? And so, I've been running it on myself and it didn't take me that long to build. I was actually pretty shocked how long it took me to build it. But I'm like, oh man, like if we run this with the kind of the understanding that, hey, this is going to help make everyone better and we're going to basically, it's going to make you work more efficiently and it's going to make a lot of processes like skills that you can then hand off to a robot to take care of so you don't have to do it. I'm like, that is super high leverage. I have a lot of these other things I can share, but that was another one that I got working. Like I built it on the plane, I think. And then like now it's just working. I'm just staring at it right now. So just a quick break. If you want to run LinkedIn personalized ads and landing pages, You can't do it on LinkedIn ads right now, but you can do it with Karrot. It's www.karrot.ai. Again, karrot.ai, and we'll see you on the other side.
I was in Colombia yesterday. Funny enough, it is Friday while we're recording.
Country?
Yeah, Bogotá. I left Wednesday afternoon-ish or early afternoon-ish, got there Thursday early in the morning, like 2:00 AM. Right. So I was at the hotel by 3:00 AM. Driver took a little bit to pick me up. I was waiting at the airport. I'm like, come on, man, there's no traffic here. I was speaking at a marketing event, IAB, Interactive Advertising Bureau. And then I left Thursday right after my speech, right? And then I got back home, call it around midnight.
That's not an easy flight. That's a 7-hour flight, isn't it?
More because there's no direct flights. So connecting and I'm like sitting and crammed, my head, my neck and back just hurt because You know, I was up sleeping in an airplane seat for a while. But nonetheless, when I was there, I was talking to some of my team members and we were having this debate on do you fire marketing employees because of AI? And what I mean by this specifically is there's 2 camps right now when it comes to AI and employees. Some are saying, hey, we're letting go of people because we're making things much more efficient. Okay, I'm not talking about the companies that are saying we're letting people go because of AI and they're not growing. A lot of those are actually just letting people go because they overhired and they're just cutting. But there are actually businesses that have made efficiencies that are cutting due to AI. But there's another camp, and my team and I were talking about this, and some of them are like, you can get— eventually you can just get rid of a lot of people and you'll have way less. Like our old VP of marketing, Salo, used to make the pitch to me, he was like, Dude, if you want your company to succeed, and he's a friend, so when he left, I had an exit interview. So I let people say whatever they want and you get feedback because it doesn't matter what you believe. It's good to hear what other people have to say. And I remember Saulo telling me, you should get rid of half to 3/4 of your company and just automate everything through AI. Now there's differences of opinions, but I believe in marketing, what we're seeing on the ground working with companies, Because of AI, people aren't really putting too much pressure to pay us less. They're putting more pressure to get more. And what we're finding is if you go the route of just cutting a ton of headcount and you're using AI to automate a lot of stuff, the quality isn't there yet where clients will be happy or the marketing outputs will produce the results that we're looking for. So what we're finding in, in Today's world, at least in marketing, and we're finding the winning combination is not necessarily firing because of AI, but more so maintaining because of AI. I'm not saying hire a ton more, I'm saying maintaining. And then you're using technology to make every single person better, more efficient, so that way you can create more outputs for your customers or for your own company that you're working for. And I don't think there's really another way because if you end up cutting staff, It is still hard to get a lot of high-quality output done in marketing and people right now, we're not hearing the pressure of we need to pay less. We're hearing the pressure of marketing needs to do more. You need, we need better results. We need a, you know, and this is just globally. I'm not talking about just our customers. I'm just, when I go to conferences, I'm hearing everyone say we need more done faster and better. Right? I don't really hear we need to cut costs.
Yeah. So here's what I think about AI as it relates to staffing in general. And as you know, we've talked about this on the podcast where I believe that my stance is that AI is going to create— it's going to lead to more job creation in the long term. And sure, there might be some short-term job displacement, but what I haven't changed my stance on is that someone with AI is going to replace people that don't use AI. So as an example, if Neil and I are fighting right now, let's say it's a physical fight, okay? If I'm holding a knife and Neil's holding a gun, over 1,000 times, who's most likely going to win? It's going to be Neil with a gun, right?
Assuming I can aim right though. Yeah.
Well, assuming if we fight 1,000 times, you're probably going to win most of them, right?
Yes.
Do you agree?
Yeah, I agree with you.
Neil, you keep the gun.
Okay?
You keep the handgun. Now I have upgraded to a rifle. Out of 1,000 times, who's more likely to win now?
You, because a rifle, if I'm not mistaken, I've never shot one, or I've actually never even shot a gun, but a rifle, I believe, spreads out. Right? So, it's easier to hit the target.
So, you might be able to— you run out of bullets in like 12 shots. I can shoot rapidly in 30 shots or so, and it's much more rapid fire, right? And then, okay, Neil, I have the rifle now. Now, you get the machine gun now. You are now more likely to win. So, the way I see it now is just like, it's no different than Neil riding a horse and I'm driving a car, and then he's driving a Formula 1 car. That's how I see this stuff. And the crazy thing to me, Neil, I was at an event yesterday, And, you know, I kind of cited the stat. I was like, look, you have 46% of Americans are pro-AI, whereas 87% of Chinese are pro-AI. How do you reconcile that? Right? And like some people, I'm just like, it just makes you faster. And so the way I see it is like, look, we want to hire people that are very pro-AI and are using the tools. Because what I'm seeing, same thing as you, Neil, is that on our end, what we're seeing is people want more work. They want more help. I don't think it's ever going to change that people want help with this stuff. And when we're showing them stuff on the single brain side, when they're looking at how these workflows are working, they're like, can you do this? Can you do this? Can you do this? How about this? How about this? How about this? It's much more of a more, more, more, and I'm all here for it because we're all going to do better if we're using these tools.
Yes, and I learned today that a rifle doesn't shoot spread out. I think that's a shotgun.
That's a shotgun, yes. Yeah, you like shotgun versus gun, you'd have a better shot if you had a shotgun. So anyway, All that to say is that's why I think it's really important. Like Neil talks about his team, well, they have people that will insert into different squads and maybe not necessarily train, but maybe they'll kind of like set up the work. And on my end, I'm like, okay, maybe we can set up a couple of these apprenticeships with, you know, the people that are kind of ahead of the game and figure out how we can spread it throughout the organization. There's no right or wrong way to do it. You can even combine these things. But I just think that the things that I'm talking about over here, like the thing watching me to build skills and all that, I think we're going to need more and more of this because people want more work done. Yeah.
Yeah, and on a side note, when we're talking about AI, constantly when I go to events, and I don't know about you, Eric, but most people want to talk about how to get included into ChatGPT, and I don't see as many people care or want to talk as much about AI overviews, but there was a BrightEdge study that came out. Did you know that AI overviews cite brands, like a company brand, 2.5x higher, not percent, 2.5x. So 250% more likely to get a brand to cite you on AI Overviews or AI Overviews to cite your brand than ChatGPT because they're more likely to do it. In essence, AI Overviews prefer citing brands over ChatGPT.
Why is that?
I don't know. It's just the way— and AI mode is— so when we look at the data, ChatGPT is 2.37 brands per query. AI Overviews is 6.02 brands per query. Google AI Mode though is less than ChatGPT. It's 1.59 brands per query.
Cool. This is awesome. Well, that is, I think that will wrap us for today. Hope you enjoyed this and we will catch you tomorrow.
