#58 AI Expert: Automate or Be Automated | Dara Ladjevardian

It is the only solution to prevent mass revolution.

This guest today is so interesting. He's built one of the leading AI companies when it comes to taking our human mind and recreating it online.

If any video you see online can be AI-generated, how do you know what to trust?

I almost literally hit post on a video that I didn't even create.

Exactly.

What jobs are the very first ones AI is going to take?

Man, I feel like—

interesting.

The most unique thing that you have is your mind. When everyone has the means of this hyper-intelligent AI tool, people are gonna pay attention to the things that stand out.

Hi and welcome back to The Big Deal Podcast. I'm Codie Sanchez, and this guest today is so interesting. He's built one of the leading AI companies when it comes to taking our human mind and recreating it online. If at any point you have worried about all of this technology coming into your life and how to make sure you are a winner, not a loser in the AI race, this episode is for you. So without further ado, CEO of Delphi. So yesterday I go on my Instagram, I click on a video and I'm like, God, I don't remember recording that video. I didn't even catch it. My team had used AI voiceover on me. The only reason I caught it is because my new president was like, your voice inflection sounds weird. I almost literally hit post on a video that I didn't even create. And I think that's just touching the surface with AI. Now you live in this world every single day. I guess one of my main questions is, do you think it's true that no entrepreneur is worried about AI taking over their jobs as entrepreneurs? Or should everybody be thinking about what AI might take over?

I mean, I think there are two parts of it. There is definitely a certain kind of person who's the doomer, who is absolutely terrified of what's going on. But this isn't the first time there's been something that has caused doomerism. You know, you think about Neil deGrasse Tyson's Cosmos, you know, you watch that show and you're like, oh my God, the world is huge. Life is insignificant. Why should I do anything? And then there's another kind of person who's like, wow, it's actually amazing that I'm alive and I should like, I'm empowered to do all these things and like work. And so I think AI is just another thing that people like to create a catastrophe out of. Whereas instead it's more like, okay, it's a new thing. It's going to change a lot of things, but. As humans, we find meaning, and there are new ways of working, and we don't have to do the things we don't want to do, and we can do more of what we want to do.

I think I'm— I fall somewhere in between. Like, I'm simultaneously excited about AI and scared about it. What do you do to not get left behind in a world of AI?

I think you have to start with the end in mind. You know, there's a lot going on today. It can be very easy to, you know, there's a new paper, there's a new prompting method. I think that's actually just fully a waste of time because these are all incremental steps towards what is inevitably a free, pretty much instantaneous, hyper-intelligent AI at your fingertips. Like, that is the end in mind. And so we're on the path to getting there, but you should be planning for that as the end and not getting distracted with, okay, cool, new metrics, new metrics. We're going to get there. It's inevitable.

So like, I'm a super normie. So this morning I'm like, AI for idiots, Codie version. What does somebody who doesn't play in your world at all, we're not even on X, you know, we don't know anything about AI except maybe that we've heard ChatGPT and we've been confused by Sam Altman on TV before. How do we start actually playing with this stuff live? Because you created a really fancy company that does this now, but you probably started with tinkering too. What, how does an average person tinker with AI today?

I think it's good to know the historical context of AI and how it's grown. For example, when I got into AI in 2014, the coolest thing AI could do was tell you whether a picture was a cat or a dog. And then I remember in 2015, I did a project that predicted whether Trump was happy or sad based on his tweets. Wow. And so—

Were you right?

Yeah. Yeah. Fully right. And it's all pattern recognition. Like, AI is just math.

Mm-hmm.

You pass in a bunch of pictures, of a cat or a dog, think about you have a child and you show that child 100 pictures of a dog and 100 pictures of a cat, and then you show him another picture that's either a cat or a dog. He's probably gonna know which one it is. That is pretty much AI.

Mm-hmm.

It's pattern recognition. And when you start it from that set, you're not as scared. Cause I've seen people who try to start at like level 5.

Yeah.

Without starting at level 1. And they're like, oh, this is conscious. This is a new kind of being. And you're like, no, it's just math. It's pattern recognition. Just as babies are, just as we are, we learn things and we predict things based on our experiences.

Interesting. Okay. So if that's where it started, what's next? Where do you think we're going next?

Yeah, I mean, I, like I said, I think the end in mind is AI that is hyper-intelligent, generating realistic videos, generating infinitely all night, improving itself, and I think if you've seen in the past, whenever there is an era of abundance, mm-hmm, you know, there's so much of this one thing, barrier to entry for creating anything is low. Mediocrity is gonna be the norm because anyone can create anything. The pendulum swings and people really start to value curation and trust.

Mm-hmm.

And so I, I'm actually like more bullish on the creator economy and like the individual economy, more bullish than ever because I, I really do think that. When everyone has the means of this hyper-intelligent AI tool, people are gonna pay attention to the things that stand out.

Hmm. Will it be harder to stand out?

100%. Easier than ever to get distribution, easier than ever to be really good at something, harder than ever to be the best. And so I think 80/20 likely becomes 95/5, where the 5% really reap the benefits of the 95.

Yeah. I mean, you kind of think about it. I remember Naval had a quote that was like, the robot army is already here. It's just inside your computers. And so in Armageddon or some of these, you know, Arnold Schwarzenegger movies, you think, okay, there's going to be this robot army and we invested in Figure, so it could be, but, and they're going to be out there, but it's not so much that it's in so many words, they're inside of your computer today. And the war is sort of who can utilize them to have the highest impact.

Exactly. And I think a lot of people think, oh, AI can do all of this stuff. It means I get to work less. I actually think it's the opposite. I actually think. You need to work more in order to stand out because AI, you can only reap the benefits of AI if you have something worth scaling. AI can only build off of a foundation that you have spent time building. If you haven't built that foundation, you're just gonna get a plateau that everyone else in the world is getting.

And what would be like an example of that today?

An example of that is, you know, I could go to ChatGPT and say, generate this SEO-optimized blog post on AI.

Yep.

Um, and anyone in the world can do that. Like, and everyone in the world will do that. You can imagine SEO's not gonna mean anything anymore because everyone's gonna be doing this. But since I've spent time learning AI and learning the nuances, I can still leverage ChatGPT to help me write a blog post, but I can also add my own nuance so it stands out. You know, it's, it's building off of my own experience, versus what ChatGPT generates, which is the average of the, of the internet.

Mm-hmm. I guess it's not that dissimilar to, if you think about it today, yeah, you get a lot of emails in your inbox and it's become easier than ever to send emails in your inbox. And yet subject line matters more than ever. So like in order to stand out, you're just going to have to be more creative because that first act of creation is no longer the barrier to entry. Like I saw this graphic today that was like, 3 levels, and it was a bow and arrow, a guy shooting a bow and arrow. And the top level was what most people do, and it's like 93%, and it's just a guy holding a bow and arrow. Like, he never actually shoots the target. And then the next level was like, I don't know, 50%, and it's like shoot once, miss. And the 1% are the people that shoot 452 times. And so now it'll be a lot easier to actually shoot 452 times, but harder to get to the 1% that actually matters.

Yeah, and you need to build the character to deal with those 450 failures and, you know, push through. Uh, and I'm glad you mentioned email because we talk about thinking with the end in mind. If you think about what happens with email when every single person is using an AI agent to send hyper-personalized emails to your inbox every single day, people are not going to check their emails anymore. I actually think email is going to be a thing of the past. Uh, be, because if you just think about it, you're just, I'm already getting spammed on my email. I'm already getting called by these random AI agents and people are hopping on this opportunity. But what, what likely is the future of email is something that you don't even check. You have this AI thing that knows you and how you speak and how you think about things, and it only surfaces to you what the opportunities you care about are. And that sounds great to me. I don't know about you. I, I hate looking at my email.

That is true. I use it increasingly less and less.

Yeah.

But then how do we get to people? Will real life become so much more important because because email has become sort of bastardized?

I think it goes back to doing the work to make yourself a unique and interesting and valuable person. Because I imagine in the world where there are AI agents communicating with each other, and, you know, an AI agent is something, an LLM or an AI that can take action. It knows you and it knows the value that you can bring to others. And it is going to convince other AI agents that, hey, this person has great stuff. But that's why the foundation is important because you need to have that great stuff worth offering other people.

Interesting. So by an AI agent, you basically mean when I get an email in my inbox and I can tell it's like made by a computer, you know, it's like, hi Cody S., uh, you know, I'd like to sell you XYZ service, you know, please reach out at this time. I loved your most recent Instagram post, but you can tell it's somehow copy and pasted. This is an AI agent, basically.

That would actually probably be probably just be someone typing into ChatGPT and copying and pasting. An AI agent would be like a step further where it actually scheduled something on your calendar. So it's, you know, multiple steps of reasoning, not just one thing, but one thing, you respond, and then it schedules time or it takes an action.

Interesting. So do you think in like, how far away are we from outsourced virtual assistants being irrelevant by AI, let's say.

I think it's already happening. I think it's just a matter of adoption. And, uh, you talked about the Terminator. People just have this fear of AI, like they're just default, oh my God, I've seen so many movies, this must mean it's bad. And I personally hope to see more positive outlooks on AI and what it can do and what this like post-AI world could look like for all of us.

So basically like right now today, I have Virtual assistants for many of our companies, we might pay them $800 a month to $3,000 a month to do varying routinized tasks, such as scheduling, travel, email management, etc. Do you think right now today there are AI agents that are better and how much cheaper than $800 to $3,000 a month?

Well, it's kind of like power law. Maybe you have like a really, really good assistant who knows you super well and can be proactive. I don't think the agents we have right now are like that.

Yeah.

But at a bare-bones assistant that can manage your inbox and even like schedule traveling for you, and it doesn't require that much personalization to your life and your nuances, yes, it's there. People are already doing it.

And in a year, will it be so good, it will know me so well, that it will make even somebody I've worked with for 2 or 3 years seem less relevant?

Yes. And I think personalization and proactivity are the 2 things that I think over the next year are really going to be the things that take AI from a nice-to-have to a Wow, this is ingrained in my life. I don't have to think about it. It's reaching out to me. It knows me and knows the way I respond to words. And I think that's when it becomes widely adopted.

Interesting, because it is actually hard to teach people to care, I think.

Yeah.

You know, my dad always said you can lead a horse to water, but you can't make them drink. And so when you have employees, as you, you know, do too, at some point there's just this like give a fuckness that is often missing. And if AI has found a way where give-a-fuckness doesn't actually matter, but it's just math and processing, that is going to be so valuable. Totally.

And I think, you know, hiring is still a play, but you just want to hire missionaries instead of mercenaries, people who do give a fuck.

Interesting. And so do you think in the future you'll just have fewer smarter humans who work for you?

Yeah, I mean, I tell everyone we hire, I want a sub-50-person public company. And I think we can, you know, you can do the job of 10 people. And specifically, AI really benefits generalists, um, because it allows anyone to be a polymath. Before, to be a polymath, you really had to be like a special kind of person. Uh, but now AI democratizes access to specialization. You can become a specialist in a couple days. I, I don't know anything about pricing or branding, but I've become a branding and pricing specialist just by using OpenAI's latest, uh, $200 per month model, which sounds ridiculous, $200 per month for a model, but it's like having a consultant at your fingertips that has a PhD in, in, in, you know, branding or pricing.

Yeah. And explain what a polymath is to anybody that doesn't know.

Polymath is someone who is an expert in multiple things. You know, they can context switch, they can be a great front-end engineer, and then play the violin, and then they can also be an amazing football player.

Yeah, so this is kind of just like— what was that movie where you took the pill and all of a sudden you became superhuman?

Limitless. Yeah.

This is like an enablement of that.

100%.

Interesting. So we won't— well, theoretically, it is the same thing because the pill didn't make you buffer or stronger or whatever, but it did make you have this mental capacity that seemed absolutely unrivaled. Totally.

And I think, uh, learning and education, I personally think, is one of the most optimistic outcomes of AI because you see something— Bloom's Two Sigma, it's a study that shows that those who have a private tutor are more likely to succeed in life by like 2 standard deviations. And obviously those who have private tutors, you need money, and it's just not something that most people have access to. But AI allows that one-on-one learning experience for any kind of subject. And I think when, when the world is more educated, ultimately it results in new innovation, higher quality of living. So I think it's a very, very, very positive thing.

Yeah, that's interesting. Is that also the study I saw? I saw something come out yesterday on X about Alpha School in Austin, which is Elon's new school. Did you see that?

Yeah. Yeah.

And basically Alpha School came out with like 2 or 3 standard deviations, um, of scoring by their, their students than a traditional school. Which is wild because that thing's only been in existence for like, I don't know, less than 5 years, I would say. And it is essentially my understanding, I don't know if it's different, is that it is individualized education per student that mandates about 2 hours of active learning assisted by AI per day. And then the, as soon as you get done your programming, whatever you're supposed to get done for that day, that would be like, oh, you get through your history test, you get through your mathematic tests, for the rest of the day you get to use your curiosity and dive deeper. And so these students are actually, they were a they were a little petri dish. We weren't sure if that sounded like a good idea. I mean, back in the day, somebody had told me, yeah, Cody, you can work for 2 hours a day at school and then you can kind of go fuck off and do whatever you want for the rest of the day. I'd have been like, this seems like a bad idea. And yet they kind of tested it and it's working.

Yeah.

Which is fascinating.

It is. And it's, it's both active learning and accountability and personalization. Like, think about online courses. Online courses aren't doing so well. Course completion rate is in the low. And it's because it has, they have these assu— this assumption that the 20 modules, I need to experience them in the same order that you need to experience them. But we're very different. We learn differently. And so the ordering of how you learn is different. And I mean, that's kind of like what we do. We have authors who have converted their books into these interactive learning experiences. And instead of just someone reading a 400-page book, it's almost like a choose your own adventure. You, you choose what you wanna learn and then you get that proactivity on how it applies to your life. I mean, how many times have you read a book and you're like, wow, I learned so many things from this book, like this book right here.

Yeah.

And you don't actually apply it for sure. And so I think active learning is, is really what turns learning into something that is applicable and people can actually measurably get better in their lives.

Explain that a little more, cuz I think you guys are one of the best in the world at it at Delphi and increasingly your roster of, you know, big time thinkers and, uh, famous people who are using it is incredible. But traditional education would be like, okay, we have a curriculum. Here's what we go through every single day. You can either self-do it yourself at your own pace, or you can do it with the class, let's say. But you go from step 1 to 2 to 3 to 4.

Yep.

What does, what does education look like with AI now? How exactly will that change?

Totally. I mean, people are already using ChatGPT to learn, like, a lot of things. I am. Uh, If you look at Wikipedia, which has documented all of human knowledge, mm-hmm. Very easy way to learn something. Wikipedia exists, but does anyone actually learn through Wikipedia? We read books, we watch YouTube videos, and I think there's an insight there about how we inherently trust things that come from human. It's just like a human need. So the next version of learning is an outcome, like becoming a millionaire. And learning a little bit about yourself, like almost like an initial survey, and then having this guided learning path over the next 30 days to make sure you understand the concepts that are relevant to you, to make sure you're applying them, and then evaluating whether you truly understand those concepts, and then continuous management until it becomes a habit. Could probably leverage concepts from, uh, you know, James Clear's Atomic Habits to see, you know, how can you take a book and actually embed it in someone's mind for practice.

Interesting. So like currently, let's say right now we teach people to buy businesses, so they go through a course and curriculum that's, you know, step 1 through 10 and how to buy a business, but it's linear. And in this future world, and we're using Delphi for a bunch of stuff in there, so I'm really excited about it. It would be like, okay, Cody, you tell me a little bit about yourself and you rate, I don't know, a beginner level for business buying. So here's like all the opportunity areas that you're not very good at, as evidenced by the questions that you took. And so because of that, you're really, you know, a lot about accounting. Yeah, but you actually know nothing about how to communicate with sellers. You know nothing about how to actually find deals. So instead of wasting any of your time on accounting, you already passed that by. You're going to go straight to these other things and you're going to score them as you progress. And then finally, when you've reached like the completion level that we think is appropriate, you'll get your little certificate. But also there'll be some sort of continuous learning option like a Duolingo, but tailored specifically to you.

Exactly. Exactly. We, we take a lot of inspiration from Duolingo in terms of, you know, how can we make something learn— learning that's engaging and fun and incentivized.

Interesting. And so in the future, will we sit in classrooms with chalkboards, with teachers, everybody teaching at the same level, or is this going extinct?

Well, I do think the in-person experience is definitely always important, and I don't wanna live in a world where we devalue that. I think. AI is about providing access where previously there may not have been access. So maybe there is a classroom, uh, you can communicate with other students, but then in the off time, you know, questions that you may be embarrassed to ask in class or a private tutoring session. So you show up in class the next day more knowledgeable, you can participate more. That, that is a future that I'm more excited about. There may be a world where humans don't value in-person communication, but I just think that, you know, that's a kind of habit change that is going to take decades and even centuries to really manifest, that we don't value in-person times with, with humans.

Right. Yeah, because at least for now, probably the thing that's super repetitive and easy to replicate is the knowledge. It's like 1 1 2, no matter what math teacher you have. But what doesn't, uh, standardize is like, oh man, every time that teacher sits down with me and says like, hey, I can tell you're struggling.

Yeah.

It's okay. You're doing a great job. Go take a walk outside. Go get a little vitamin D. You need a little coffee here. Like that almost mothering or human EQ component will probably be one of the things that computers are sort of last to integrate. Although they could at some point too say like, good job, buddy.

But they could. But same, same as the Wikipedia thing. I think inherently there is just like this unknown. Psychological thing where I am talking to a human.

Yeah.

I'm not learning from Wikipedia, which is a summary of that human, or their AI, which is, you know, good access while I can't have them, or maybe I'm too embarrassed to tell them the thing that I want to tell them.

Mm-hmm.

But in the end, it, it is the human. And, and we've seen that behavior with Delphi where someone is, you know, learning from someone's digital mind and then ends up wanting to meet them even more.

Yeah.

It's like, oh, like I, I want to get to know this person. I really, really resonate with the things that they say.

It's true, because if you think about it, you know, right now when I meet somebody, let's say, who's read the book or followed some of our content, we already have this like parasocial relationship, right? Yeah.

Yeah.

And then if they were able to actually have me communicate direct with them and answer questions, they would really feel like they know me. And then if I'm able to like get a bunch of information, let's say, from them so that every engagement we had, we got to know them too, that would be kind of interesting. And doesn't exist today.

Totally. And you know, you're, you're a big creator, uh, you're pretty well known. So parasocial relationships are pretty common with you. I think parasocial relationships going into the business world is pretty interesting. And what I mean by that is, you know, there's that, that leaked email from Mark Zuckerberg to Peter Thiel where he says, like, I think my reputation with millennials and Gen Z are particularly important. And then a month later he has a perm and a chain and like really, really caring about his brand. And he's a very calculating guy, like, none of that is by accident.

Yeah.

And so I think, uh, the, the insight that he has that I have seen with having my own clone is people connect with people and less so with, uh, brands, like nameless brands. And so from the business owner perspective, I think, you know, the founder-led influencer, the founder, founder-led marketing is gonna continue to become more as this idea of parasocial relationships with business owners starts to be more commonplace.

So true. Like, if you have a business today and you don't have an ability for your customers to relate individually with you or whoever you have as the face of your business, you're probably missing out on a big monetary opportunity. Totally. Like, why do you think they put Jared for Subway, you know? And that one could go a little sideways, but it's because they needed a face to go with the brand. We don't associate with corporate logos the same way we do with humans. It's That's a really good point. Explain what a clone is at Delphi, because I imagine people immediately think of, like, I think that movie with Joaquin Phoenix where he has Scarlett Johansson in his ear and they become best buds, and then she's best buds with everybody and his heart's broken. What is this clone that you guys do?

Yeah, it's definitely not her. And then people love to reference that movie. I think it's like another dystopian movie, so I don't love it. Uh, and we're also moving away from the word clone because it gives that dystopian taste.

What do you call 'em?

Digital mind. Like, what we're doing really is the most unique thing that you have is your mind. You know, your expertise, your experiences. And up until now, the way that you scale that mind has a cap. You can only be in one place at one time. You write content to scale your mind, but it's not you. It lacks the personalization. There's no way to scale personalization.

Yeah.

So a digital mind is something that understands your tone, your style, your thinking. Allows you to scale your thoughts like a book, but is bidirectional and personalized, like talking with you would be.

Yeah, I mean, you could even think about it, you know, in a non-dystopian way. Like, let's say you have an accounting practice right now, and you have a process and systems that you use for every single person you input. And then at one point you're like, I'm so overwhelmed, I can't do anybody else's accounting, I need to hire somebody else. And then I have to train them on everything that I know about accounting. And so I hire another person, and then I hire another person, and that's how we get, you Ernst Young and these big huge companies. Well, in the future, part of that can be, hey, I have all these systems and processes. Every time I onboard a client, I don't actually hire a new person. I just have them interface with my AI. And so I've essentially become not only superhuman mind, but superhuman hands, because as opposed to having to hire thousands, I have now individually created a business that is based on one sort of dataset, me.

Yeah, and think about when you scale to thousands, at least from what I hear, the biggest problem when you scale as a company is that people stop making decisions in line with your culture and how you do things. Like, that's where things really break. And so, uh, Sam Altman talks about the, the one-person billion-dollar tech company, but maybe there is the one-person billion-dollar company in general. It doesn't necessarily need to be a software company.

Oh yeah, absolutely. I mean, and also I think if I remember when I first started a business, one of my favorite mentors said to me, he's like, before you even start, I'm going to tell you what your biggest problem is going to be. I already know. And I'm like, perfect, tell me. He's like, no, no, I'm going to write it down. He put it on a little piece of paper. He's like, I'm going to put it in an envelope. I'm going to date it over the part that I zipped it so you can see that I haven't changed it. Then when you have your first problem, come sit down with me. We'll talk about it. And so I remember I started the business. I'm like, all right, whatever, weirdo. So I go, I start the business, sit back down with him later. I'm like, I have this huge issue. It's a total nightmare. He's like, wait. Pulls out the envelope, hands me the envelope, and he goes, before you tell me, I want you to remember this. You know, I still have it. And I said, you know, he said, what's your issue? And I was like, well, you know, this person's doing XYZ thing. And he opens up and he just says, the best thing about your business, the people. The worst thing about the business, the people. And so that is very true. Every time you get in business, scaling humans is really, really hard. And so I think this will be a huge unlock for business leaders. If they think about it. That's why I did like that idea of like entrepreneurs aren't as scared about AI because they're thinking about how to use it, not how to get used by it.

And even if they want to hire people, it makes their team more effective. And we have CEOs whose teams like love the clone because they don't want to— they don't want to bother the CEO and they're using it to— okay, is this how he would make or she would make decisions?

What do you think is the most interesting thing happening in AI right now that most people are not aware of? Like, today, right now.

I really pay attention to the psychological impacts of AI.

Hmm.

And so I, I find these video generation models really interesting because, again, I always think with the end in mind. Like, what does this mean for our digital behavior. And if, if any video you see online can be AI generated, or any picture, and it's near perfect, how do you know what to trust? Are pe— people are either gonna trust everything, or default, it's gonna be, I'm not trusting that. I'm not trusting that. Uh, so I really think it's interesting as these things get better, like, will social media still hold the role? Like what, what technology needs to be available to make social media something that we can still use? Probably some sort of like trust authent— authentication layer where it is like, this is AI generated, this is true, this is false. And you have these companies like Polymarket and Kalshi, they are these companies where you can bet on events. And I don't know how they validate whether an event actually happened or not, but they, they have found a way. Uh, so I think it's really interesting what the future looks like of trust. And are we going to be okay with not knowing that something is true or false? Or are we going to, to want to know like, this is a deepfake, or, uh, my voice was stolen and now it's being used to call my grandma to send her money. Uh, so I, I think that's a little scary and I think there needs to be solutions around that.

It's a great point. I was off Twitter for like 4 days this week and all of a sudden I came back and it's like the CI, 'CIA confirms there are astral-projected huge beings on Mars that came to the US that built the pyramids. Did you see this?' And then there's, like, some pillars underneath the pyramids.

[Speaker:LATIF] Holy moly. I told my sister, I was like, 'This is so cool. This changes everything.' And she's like, 'That's not real.' [Speaker:LULU] Yeah.

So interesting. So the—and—but my immediate take was like, 'Okay, I guess maybe. I don't know.' Like, but what are you gonna do? Because there's so many wild things coming out nonstop. You're like, 'For sure.' JFK assassination happened by the CIA. I have no idea how to prove any of that. So what does happen in the future in order to know if things are truth or not? Do you think that the next biggest AI company is actually just a— is it just authentication? Will there be some stamp or is AI smarter than a stamp?

I think it's such a difficult problem to solve because you get into like political bias and like what's actually right, what's like, what's true and what's false. And then you have like the Orwellian 1984 view where if someone has control over what's true, then like they can control the narrative of history. So it would require creating like an unbiased AI that has access to just like the stream of the internet in real time. And maybe it's something like what I mentioned is the future of email, where the future browser is something that can filter things that it knows are true or has some like importance in your life. Because like you said, we see all these crazy things these days on on television and on social media, and it's just like, "Oh, cool." Could be. Yeah. Aliens. People just, like, don't care anymore.

Like, it's kind of crazy. Yeah, I mean, think about what they would have to do to break through and make me really concerned. I mean, they've literally released multiple times that aliens exist from the government, and now there's this thing about CIA and the pyramids, and you're like, "Yeah, I don't know. Anyway, I'm hungry." And so what does that look like? They're aliens?

Like, okay, cool. Cool.

Yeah. Yeah. Sure. We need breakfast. Yeah. Um, I guess the other question for you there is like, what books would you read for you to understand the negative consequences of AI? And what books would you read to understand what could be possible positively? Because I think it's useful to see both sides of the coin.

Totally. Uh, there's obviously AI Superpowers, I think by Kai-Fu Lee. Okay. That's, that's more on the negative consequences. Um, and then all of Kurzweil's books are great because he was so ahead of his time. I actually got into AI reading his book, How to Create a Mind, in 2013, where he talks about how you can leverage computer science to recreate someone's mind. And, uh, yeah, it was, it was reading that book and then using LLMs in 2021 that made me realize, wait, the technology he was describing in that book is pretty much LLMs. So this, this guy was really ahead of his time.

I think, uh, explain what an LLM is for anybody listening.

LLM is a large language model. We talk about AI being just a pattern recognizer. So an LLM is just like a huge pattern recognizer trained on all of the words on the internet and can see a pattern in a sentence and predict what the most likely next word is. So when I told you Old MacDonald Had, what's the next word? A farm. Yes. So you just, you're an LLM. You just, you just did that. We might be LLMs. I mean, that's a whole nother conversation. But, uh, then The Fourth Industrial Revolution, I think, is a good book on all the potential good things of AI. I think Peter Diamandis is very optimistic about AI, and his books really go into, like, AI with biotech and flying cars. If you really want to, like, walk away feeling hyped about the future, definitely check out his books.

I like it. Would you also throw in there some of the dystopian fiction and, like, the utopian nonfiction? Would you throw in there, like, like a 1984? Would you throw in like a The Diamond Age? You know, um, what, what else would, should people be reading even to like break your frame of how you think about AI today? I think it's a great point.

I think with AI, you almost have to like forget everything you've ever learned and like try to rethink things from first principles from the ground up. Diamond Age is great because it talks about, uh, you know, it has that, that personalized tutor, the illustrated primer that can grow with you and teach you things. And I think that's exciting. And then 1984. Is great because it's like the consequences of what could be. Yeah. And I think knowing the consequences of what could be allows you to be intentional about the decisions we make today to make sure that is not what happens. Yeah.

Actually, Diamond Age is one of my favorite books. Did you know that is the most often cited book by tech CEOs for their daughters? Really? Which kind of tracks because it is about this, you know, kind of magical education book that you give to your that's given to a young woman who kind of has no prospects and awful things happen to her, and then her life changes.

It's the promise of having a mentor or having someone who cares about you and can guide you.

Interesting. When you— you mentioned something weird that I want to go on a tangent on, which is, are we LLMs?

What does that mean? So in Kurzweil's book, How to Create a Mind, he describes the mind as a collection of building blocks. And how complex intelligence results, uh, from that collection, that organization. So it's a hierarchy of pattern recognizers. On the low level, you have pattern recognizers that recognize the shapes and letters. So M, I, L, that gets fed into higher level pattern recognizers where it's sentences, and then there's paragraphs, and then there's meaning. And at the highest level, there's consciousness, and you're not aware of all of these lower-level pattern recognizers. And so what is also a pattern recognizer? An LLM. It takes inputs and it has outputs. It recognizes things. And so something that I think about a lot is like, are we just a hierarchy of GPT-10s? Um, and what we're experiencing is consciousness, similar to how he described the mind working. And I also think about that a lot in terms of dreams. Um, because if you see some of these AI-generated videos, I don't, I don't know how vivid your dreams are. I have very vivid dreams, but these dreams mimic exactly how these AI videos look. Interesting. And so it's like when we're asleep, maybe the LLMs don't have as much power, and so they are lower-level LLMs. Uh, so anyways, it sounds a little crazy. I think even if it's true, it's similar to the cosmos idea of like, oh, we're so insignificant. Or the simulation hypothesis where what if we're all in a simulation? It's like, okay, we're LLMs. What are you going to do about it? You still got to make meaning out of your life. You still got to work hard. You still get to do good things.

Like, so what? Yeah. Yeah. So what, what keeps you out of what I think happens with some people when they start talking about this, which is like becoming an existentialist and you're like, why does this matter? You know, what is life? What is meaning? How do you drive meaning from something super technical like AI and the belief that at some point like, will AI just do all our work for us? Will we not need to work anymore? And will we live in a utopian world in which AI does everything?

Yeah, I mean, I think you have two kinds of people in this world. You have people who— there will likely be universal basic income. I like, that's the only solution I see to what's about to happen. Really? So many people are going to lose their jobs. But a lot of people, I imagine the listeners of this podcast are like ambitious people who like want to win. Uh, That's not the people I'm talking about. I'm talking people who go to work and they don't necessarily love their job and they would much rather like spend time with friends and family, be creative, um, maybe just find joy in, in things like that. And then the more interesting thing for me, and I think listeners of this audience is for people who still wanna like win and find meaning, like what does that look like? And I think that's where. Really starting with the end in mind and thinking about the psychology of humanity in a world where mediocrity is abundant and the barrier to entry is super low. What does that mean? It means we have to work harder to stand out. We have to, we have to think about our relationships even more. Customer service matters more. Treating people with respect matters more. Brand and creating something unique matters more. Uh, so for how do I stay motivated? I. Very much believe in what I'm doing. And I think I have a lot of good motivational sources that keep me going. And I just go back to the Wikipedia example where I truly believe humans value humans. And even when AI is very good, uh, we, we care for one another in some way or form. Yeah. You know, it's interesting.

They're building a Skims store on the street. I don't know if you saw it. And I saw this Instagram the other day. About taking every single item from Skims collection, sorry Kim, and comparing it to the Amazon Basics, which are like somewhere between 200 to 400% cheaper. And yet they were explaining what was actually the material of the clothing. So in order to decide if one was better or not. And the moral of the story is like almost everything that you can get from Skims, let's say 80% of it, Pareto's principle, 80% of it you can buy online at Amazon that appears to look exactly the same and be the same material, but it's about 200 to 400% less. And yet Skims sells out of almost every single collection. And so why?

Because brand matters more than ever in a world. Status, associating what, like, uh, communities, what you want to be, what you want to associate yourself with.

Right. So in the future, you might not need to differentiate as much your actual product, but you certainly will the packaging. It goes back to the Tiffany's model. And who you are. And who you are. Explain that more.

This goes back to the Mark Zuckerberg point, like where I think Kim Kardashian, a lot of people look up to her, like she, she is a status symbol. And I think as a business owner, uh, especially as you have a lot of get-rich-quick people who are building these like shitty AI tools and people are going to try them like, damn, this sucks. Oh, it's because the founder had poor intentions. And I think people will start to evaluate the people behind the products. More so than they do right now.

So in a world in which AI makes everything easy, the thing that's hard to replicate is you and your own ethics.

Yes. And I think it's a hard thing to work on. Like, how do you work on yourself and your own ethics? And I personally remember when I was working on my first startup, and my first startup was really hard. I was in, like, in a dark place, super lonely, and literally spending 3 weeks just, like, writing my stream of consciousness thoughts, like, really understanding who I am and what I cared about. And coming out the other side and having a very clear understanding of who I wanted to be and how to catch myself when I'm slipping.

Interesting. Your first— you've had a couple shots on goal to start up Land. This one is appearing to be very successful. There'll be some fun announcements in that in the coming months. But what do you mean when you say that your first one was really hard and you were in a deep place?

What happened? So when I quit my first job to start a company, I followed a common piece of advice in Silicon Valley, which I now completely disagree with, which is just like, you can start a company for the sake of it, you know, throw spaghetti at a wall, iterate towards an idea. And I think that's really dumb, uh, because one, startups are super hard. And the only thing that's motivating you through that 7 days a week, if, if if you really are passionate about what you're working on. And two, if everyone is throwing spaghetti at the wall, what's left? Like, what is left to work on? Like, the only— all the low-hanging fruit of ideas I think have been grabbed, and the only ideas worth pursuing take a long time to really, like, manifest and flesh out. And so my first startup, I left, didn't really have a plan. I was a solo founder, didn't really have any mentors in my life. And I also didn't know why I was doing what I was doing. I had no foundation. Mm-hmm. And, uh, an analogy I tell every person who joins Delphi, which they laugh at because it's, it's kind of like dad analogy vibe, but you know, bamboo in its first year of life is only 1 foot tall. And then after that first year, it grows super fast and it doesn't fall. And it's because that it built that foundation. And I felt like in my first startup, I hadn't spent the time building the foundation. And so while, you know, built this startup Friday AI Assistant, got like micro-acquired, not something crazy, minorly successful. I don't view that as a success. I do, I view it as a great learning experience for a period where I found out what my foundation was, uh, found out what I cared about, found out what drove me and guided me towards my next thing, which was, okay, I need to find a co-founder and learn from someone great. And that's why I moved to Open Store to work for Keeter Boy and met my co-founder there. Interesting.

Would you not do another company again without a co-founder?

I don't think I'm going to start another company after this. I think this is the one. But yeah, I definitely think— as an investor, great answer. Yeah, I think this is the one and I can go into all reasons for why I say that. But, uh, I— my co-founder is amazing. It's like a marriage. We have the same values. We care about the same things. We're very different. I'm go, go, go fast, break things. He's like an infrastructure engineer. So very detail-oriented. So that complementary skill set grounded on the same values is— has been huge for us as a company.

How does one go and find a co— like, how would you go and find a co-founder that aligns with you? What's the framework?

I don't know if I have a framework that's repeatable. I can just say that my mentality was, okay, Keith Rabois is a super successful founder and investor. He has this kind of crazy mindset about work where You know, you got to work 7 days a week. The movie Whiplash is like a great movie, uh, which is, I personally love it. We make all of our new hires watch it. And so I was like, okay, the type of person that's going to go move to Miami, which isn't a proven out tech scene, to go work for Keith is someone that is open to this work ideology and someone who's open to risk. Because moving to Miami is just like, why would you do that if you work in tech? So that's why I ended up working at that company. and Sam and I initially were super competitive. Like, we did not like each other because we wanted to be the best at the company. Um, but then Sam Altman has this blog where the best way to determine if someone could be a good co-founder is to go on a high-stakes travel trip with them. So Sam and I went to Santo Domingo with very little research of like the vibe of Santo Domingo. It's not really a touristy place and we definitely had some high-stakes situations that resulted in us being like, you know what, I think, uh, I think this could work out.

That's a great idea, actually. It's probably a great idea for dating too. Totally. Like, if you're gonna date somebody, skip the 2-year process to figure out if you like each other and go do something really, really hard, the two of you together, for an extended period with no plans, with no plans, and see what happens. Yeah, yeah, it was great. Interesting. Yeah, that could have gone really bad.

Yeah, it almost did.

Yeah, yeah, Santo Domingo is a little sketchy. I've been there before. Um, you said something that's wild to me. As an insider in AI, on the forefront of what's happening, you think the future of UBI, universal basic income, is almost guaranteed.

I almost feel like it is the only solution to prevent mass revolution. Wow. Because people are gonna lose their jobs. Not everyone is like a type A person. I don't know if you can teach type A. Maybe, maybe it is an educational thing and maybe the illustrated primer and, you know, having a personalized mentor, more people can become type A, but I think people are just not like that. And so what is the alternative? Uh, they need enough to sustain their life and maybe we have nuclear energy and maybe it doesn't even cost money to have energy. Maybe we have vertical farms and, and food is free. So I think that's the cool thing. And Peter Diamandis talks about this a lot. Where, uh, all of these technological inflection points are happening right now. And I think it's a great time for biotech to be working well and for energy to be working well because for AI to really work in our society, we need some of these other things figured out as well.

Yeah, you know, I'm such a traditionalist in some ways, but it seems like there's a lot of studies that say humans without purpose are not happy. That actually you have a higher likelihood of dying post-retirement for no physical difference in that you don't have a reason to wake up. Any longer. Yeah. Uh, that also people who work versus those who do not work, uh, are happier than those who do not work. That actually there is this, you know, sort of, um, 7 million young men in the country who are choosing to opt out of the workforce today and, and thus have deaths of despair. So what happens to society if we're given income but we're not needed? Don't you worry about that?

I do, but If you think about human history when we didn't have like all of these jobs, engineer, uh, SEO specialist, what was the meaning that people had? It was really their communities. You know, what you can do for your family, what you can do for the community around you, um, art and creativity. And so I think one, AI democratizes creativity. You know, anyone can be a movie director. I personally think that's awesome because I've always had dreams to like create a Christopher Nolan-like movie. I don't have the skill set to do it, but maybe with AI I can try something. And I think, I think that's personally very exciting. Yeah. Uh, and then the community, I think maybe we'll go back to like a community-based society, which I think people are usually happier when they have a community and they're not just adding value to the shareholders, uh, but they're adding value to the people they love. Interesting.

Um, Do you— let, let's go tactical here for a second. Yeah. If AI is gonna take jobs, but people still need them pre-UBI, let's create some bridge between the two. What jobs are the very first ones AI is going to take?

Man, I feel like my answer to this has changed so much. It changes, I think, almost every month. Mm-hmm. For a while, I thought it was software engineering. Mm-hmm. Because, uh, Well, I think actually software engineering in the way it is today, I think is, is not going to be a thing because you already have things like Lovable and these other companies where you can generate a website or app just in like a few seconds. And I think the important thing will be nuance and taste, like making the experience good, making the design good. If you're a system architect and you want to build a scalable database solution, you have to have that kind of knowledge to make it beyond just the AI slop of build a website that can tell people their finances, right? So I think software engineering in the way that it is today, for sure. Consulting, 100%. Like McKinsey, uh, all of those big consulting firms. I don't know how they are thinking about things internally, but, you know, doing research and creating a report and creating a PowerPoint deck is almost like the easiest thing you can do with AI, especially with OpenAI's newest model, O-1, and then O-3, where it is so good and you can just like pass in everything about your business and the problem you're trying to solve, and it will give you like a pretty good one-pager on like what you need to do to solve that.

So if there's a business owner listening right now that has problems in their business, one of the things they could do today to implement AI is go to OpenAI, use these two more premium models, and try to run every single problem they have through their new co-founder, which is AI.

Yeah, I'm trying to get my dad to use it.

Interesting. Um, God, I cannot imagine my mother or father using AI right now.

You really gotta teach them. You really gotta sit down with them and teach them.

What's your, what are your step function process to teach somebody who's never used ChatGPT how to do it?

It's, it's hard. I think, uh, One, starting with the foundations, which we talked about, like, you know, this is just math and this is the evolution of AI. And then just like when your parents ask you a question, like they usually do, I just send them ChatGPT, like, ask, ask it. And they're like, oh, wow. And then they start to use it more and more and they start to get creative for how they're using it. And I think the key thing for learning how to use these AI tools right is knowing that it is fully based on your, your own creativity. Inputs lead to outputs. And so there aren't some strict rules. We're in the very early days of prompt engineering. And so that's why I think unlearning most things that you know and coming in with the mind of a child where you're like really curious and you're trying different things, like what would happen if I just paste my entire Twitter history into this model and ask, uh, if it thinks that my Twitter content is good. Um, or just like getting creative in the types of data that you can put in there and seeing if it works and, and seeing what you can do to improve it. Yeah.

And prompt engineering basically meaning the questions that you ask actually become their own version of sort of in some way code or writing code because instead of writing code in a traditional sense, the way to use this is by the questions that you ask.

Yeah. I, I almost think that there will be, you know, Robert Cialdini, he wrote Persuasion and Influence, which is how do you say words that get other people to do what you want? Uh, I imagine there'll be like a Robert Cialdini for AI. It's the same thing, like you're prompting me, I'm prompting you, and similarly, like we're prompting AI to get what we want, influence it, persuade it to give us really, really good results.

Interesting. Um, okay, I want to go back really quickly to we talked about jobs that AI will kill. We think it's software and engineering first. We think consulting next. What are the other top 3 to lead to the top 5?

I think anything that is not relationship-oriented, um, like you can go on Upwork and like get all these one-off jobs. Yeah. I think AI is a much better tool because the relationship doesn't matter there. Now I do think power law comes into play here where the difference between okay, let's use AI to give us this graphic to like the best of the best. Like we're working with, uh, Pentagram right now on our rebrand and they are working with them. I'm like, oh my God, the bar is so high. The difference between like good and excellent is just like enormous where the difference between like okay and good is, is not as, not as wide.

That's such a good way to think about it. So, so it's kind of So it's engineers, it's consultants, then it's, what would I even call that? Like, uh, outsourced service-based people online. Tech, tech-based services who are not the best.

That's why I said 80/20 turns into 95/5 because the best actually can scale what makes them the best. Yeah. Which is why it's actually the bar is higher and you gotta work harder if you wanna, wanna compete. Yeah.

But it's, it's a great point because you're right, actually the best actually think super asymmetrically. Mm-hmm. So they're not linear at all. Totally. You know, they're like, Your new branding is this. I think we should look at cotton candy. Plus his hair reminds me of a wave. And so, and they throw together things that AI just would not be able to connect the dots.

And they ask questions too that make you rethink your assumptions. Hmm.

Interesting. And that at this point is something that AI is not capable of.

I think you can program it to be capable of that, but, and I, I have tried, you know, I've, I've used AI for branding, but it's just like the, The, the difference is large, both in the quality, but I also like the fact that I am building this relationship with branding people who are the best in their craft. Like, it makes me feel good. Like, wow, I know these people and like, they're amazing and I love chatting with them. Yeah.

How do you talk somebody into using more AI that doesn't think they need to today? And the reason that I would frame that is sort of like, I think about a lot of my employees right now. And as a business owner, of course you want people to use more AI because then you get more leverage out of all your people at a lower dollar amount. But there's like a deeper demand I have for them, which is a realization that those who use AI and are able to integrate it will be able to continue in the workforce on a go-forward basis for my company and for others. And those who don't won't. Yep. And so I feel like kind of this visceral need to shake some people and go, you guys, Either you figure this out or you won't have a space here in the future. But I feel this inherent friction between those who are willing to do it and those who just— it's almost emotional— and won't.

Yeah. And then we've seen this with our product. Some of the teams of people we work with are very anxious about our product and almost like want to sabotage. It's, uh, success. And so I think there is two things. One, when you hire and you have a job posting with a job description, update your job descriptions because like we said, AI allows you to be a polymath. You can be a specialist in a lot of different things. So the expectation is not just doing your job, but it's finding new ways to help in the business in any way because you can now, there's no excuse. Like you can learn how to, okay, there's a problem with, uh, our trust score over emails. No one knows how to do that, but you can figure it out. So just do it. So very intentionally saying this is a job, but generalists, like the job is also figuring out what's wrong and what we can do to hit those key KPIs. And I think setting what your KPIs are and what definition of success is in your company allows people to think, okay, in a world where I could be in multiple places at a time. And I didn't have to do the stuff that I really hate doing, what are things I could do to help accelerate us towards those KPIs? And are those things, things that I enjoy doing? Yeah, because then it's— there's an incentive to being like, oh, I'm totally going to use AI to get rid of the stuff that I hate doing. And I have found that there is this huge impact space that only I can do. Interesting.

And so almost the question you have to ask yourself continuously is— I used to say Um, level 1 entrepreneurs always ask how, how do I do this? Level 2 entrepreneurs always ask who can help me do this? But maybe level 3 entrepreneurs ask, how can AI solve this for me? Yep. And it's just not how, it's not who, now it's AI. And so every problem you have, you should be asking, how can I prompt AI to solve this solution for me? When I feel pain, AI. When I don't like something, AI.

Well, I think it's both how can AI help here, but keeping in mind that there are certain areas where the human touch does matter. Communications with people. And, you know, we talked about infinite email inbound and fake deepfakes and bots. And I think people really appreciate authenticity. Yeah. And I think being authentic and upfront now, it may not get you like the overnight growth that people are getting with these AI agents. But in the long term, you'll build a reputation of someone who is trusted. And as the, the bar for trust gets higher, where it's like, I don't know what to believe, but that person has always been authentic and honest. So I, I know when I see that person and what they say, I can trust them. Yeah, it's a good point.

Yeah, we put, um, a bunch of our customer service emails in one company through AI, and the team is really excited about— they're like, look, it can actually prompt to create this personalized email for people. And I was like, curious, show me the email. I'm like, for the love of all that's holy, please never send these to anybody because they felt like computerized sloppiness. Yeah. And so you have to make sure, I think in some ways, that the things you do with AI don't become complacent because it would almost be better to have a slightly badly written human email than to have a mid AI email. -based email. Yeah.

Or what, what I do with, with my digital mind is I just send them my clone. I, customers who have questions, it's in the product. I send them the clone and they know it's my clone. But the alternative is either me not having time to meet over the next few weeks or answer any emails, or me sending them our documentation, which has so many pages they don't wanna read and look for the whole thing. So instead, it's just a much more convenient way. For them to get the information they need.

God, I really need to get my clone going with you, don't I?

Oh yeah, we're going to get it set up in the next week.

Okay. When it comes to, when it comes to AI in particular and like hiring, you had mentioned the resume and that you should put on there that, you know, there needs to be like some AI first. Like, how else do you think about hiring in a world with AI? Is there anything else different that you do today?

Yeah, I don't know if this is different because of AI, but we hire two kinds of people at Delphi. Mm-hmm. Either those that are the top 0.1% in what they do, and how do you know they are the top 0.1%? They are more artists than scientists. Mm-hmm. They're kind of like, like Rick Rubin vibes. Wow. You know, uh, you, you don't wanna hire people who are externally motivated. You wanna hire people who are obsessed with being the best. And so I, I talked to you about this growth guy. We just hired, he's obsessed with the craft and he can just obsess and talk about it for like an hour straight. And I'm like, oh my God, uh, I would work for you. And I think that's a key thing. I would work for you in your best form. Yeah. Um, and then the other kind of person we hire is the usually high school or college dropout or young person, very scrappy, high slope. And you can learn a lot about slope based on one, In the interview process, are they using AI to learn different things? Um, for me, how they talk to my clone, they're curious about the business. They're learning things. They're learning fast. And, you know, our first hire was this 17-year-old kid from Belgium who now has a visa. Now he's in San Francisco and he has designed everything in our app. And some people say the best part of our Delphi is our design. And it's just like some 17-year-old kid who has just hustled and worked super hard to, because he has the potential to be that 0.1%. Yeah.

You know, I was talking with Ankur the other day from Kerry, you know, and one of the lines I loved the best that he said on that podcast, if you guys haven't listened to it, you should. He was the— it's called— he's the founder of Kerry. And so one of our other portfolio companies. But he said a line that you guys live in San Francisco that I forget often. And I think people do in the rest of the country, that in Silicon Valley, the magic of it is that is one of the few places in the world where We unequivocally are fascinated by what young people are obsessed with. That age has no bearing on how intelligent you are or how efficient you are or whether you should be listened to or not. And as like a Latina from a culture where it's like, shh, you know, you're too young, like, nope. That is opposite in Silicon Valley. And it sort of feels like if you're a young person today with tech where you are, man, what a time to be alive because you will be listened to in a way that you never have before, I think, as our ability to go backwards and learn all this stuff is so much less than your ability to natively integrate it. Totally.

And I talked about people needing to unlearn their assumptions in order to like really open their mind in what this AI-first world looks like. And some people are very, uh, rigid in their way of thinking and it's a lot harder for them. But the, the great thing about young people is that full open mind, the sky's the limit. They haven't been told no so much in their life that they've become cynical and it's like default no, it's like default yes. Let's, let's see what's possible. Yeah.

And you know, you know what's fascinating too, in some way in this world today, I always loved that line, which is, you know, um, pessimists sound smart and optimists make money. Mm-hmm. I think increasingly though, with technology, optimists are gonna win even more. Yep. Because pessimists sound incredible on the internet. I hear them all day, every day. It's not possible to do this. Businesses don't make sense. Not anybody could buy them. Not everybody could own a business. You could never make that much money. And it's like, now we're in this change in our content where we're actually we are showing people, like we are doing the thing live with as much transparency as possible to say that it is, this bar is so fucking low, you guys. This is low actually. When you see what you guys are building, your ability to go out and start a vending route and make 5 figures a year and then 6 figures a year. Yeah, that's fucking possible. This is not rocket science. And so I think it's really cool. Do you think in the future then, is it almost, is this the best time to be alive for a young person?

Yes, I think it is the best time to be alive for a young person who is very confident with their intuition. Hmm. Because I think all of the low-hanging fruit of startup ideas, you know, payment system, accounting system, they're all taken. Meaning I, I think the really the only opportunities left are those that are weird or, you know, Keith, our investor says he, he likes to invest in companies where the majority of people are laughing at him. Mm-hmm. And when we started the company a month before ChatGPT, people laughed at us. You know, they, people thought it was stupid. Why would this be a business? And, uh, it didn't work out for like a year. But I think both, if you're doing something that people are laughing at, lean in. And two, uh, it's just gonna take more time for those things to develop and manifest because there's a huge idea maze. It's a weird, it's a weirder idea. It's a new space. You have to travel the idea maze, which is a concept that I think Balaji Srinivasan coined where you have this idea and you're trying to figure out what the best way to execute it is. And there's certain dead ends and you have to go back and okay, that didn't work. But eventually you get to the, to the gold star of the idea maze.

Yeah, it's actually fascinating. I imagine like the inside of a computer or an LLM actually is like a conversation with Balaji. If you've ever, do you know him?

Yeah, he's one of our investors.

Oh, I didn't even know that.

Yeah, yeah, he's got a Delphi and everything.

Oh, he's a friend of mine. And every time I talk to him, I feel like I can see him actively working the Idea Maze because he actively is sort of talking about, and maybe this is the way of the future a little bit, here's an idea, I want to beat it up. So here's my premise 1, 2, 3, 4. Here's why I'm confused on whether premise 4 is real. And you can see him kind of real time speaking it out loud. Totally. Which I think is Interesting because now you can do that back and forth with AI. It's almost like premise, test, premise, test, premise, test.

Exactly. He did it with crypto. He did it with network states. Yeah. And that's, that's why I think it's important to build a muscle of being told no. Like it's the being underestimated is a good thing. People laughing at you is a good thing. Um, Naval has this line where he's like, you want to be a loser. Like you want to be a loser that can't fit in because then you're just going to do things your own way. And doing things your own way is actually where the biggest alpha is.

That's so true. Yeah. Did you have a moment where you were laughed at when you started the company?

So many times. I mean, people were like, one, oh, like you, you're not OpenAI. You don't have any AI researchers. You're not going to succeed. You need to own the model. The model is where the moat is, which actually turns out the model is commoditized. And I think the foundation model companies are at most at risk. Uh, so he was like, why would people want this? I don't get it. And, uh, it just made me realize that people are very much thinking like now, and we need to teach people more to think about the end in mind, like the projection, like where is this going? Why does this matter? And why is this— Delphi is rooted in human psychology. We want to learn from other people. We read books, we watch YouTube videos. Uh, sometimes we pay for coaching. It's, uh, first principles thinking, like, yes, it works. And I actually think the book that helped me stick with Delphi for the first year, which was not easy, not working, there's a book called Loonshots. Mm-hmm. Uh, it's one of our required readings in our company. And in the book, he talks about this concept of the false fail. Mm-hmm. And basically it's where something fails, but not because the idea is bad, but because something else about it is bad, like the execution. And so he talks about Peter Thiel evaluating MySpace as a social network and people are leaving MySpace and some people are like, oh, social networks don't work. Like you shouldn't invest in social networks. But what Peter Thiel found is people are leaving MySpace because it's very buggy. There are, there's a lot of bugs and Facebook does not have bugs and it's working pretty well. So he is like, okay, MySpace is a false fail, so I'm gonna invest in Facebook. And so similar with Delphi, it's like, okay. False fail. The brand isn't good enough for Delphi to work. People are trusting us with their data, so it has to be like a really, really trusting brand, really high quality product.

Mm-hmm. What other books are on your required reading list for your high performers?

A lot. Uh, so we got Loonshots, of course, Zero to One, but that's like a classic tech bro book. Build by Tony Fadell, my, one of my favorite books. He talks about how you wanna follow your intuition. First and then data second. Hmm. And I really resonate with that because in the end, the in— our intuition is all we have. Mm-hmm. I would rather be wrong listening to my own intuition than listening to someone else because at least if you're wrong, you get the data and your intuition improves. Uh, another is, I said Seven Powers, Loonshots, Zero to One, Mastery by Robert Greene. Mm-hmm. How to be the best.

Um, he's on the podcast tomorrow. Really? I know. He's like on my list of like people I'm going to geek out the most about meeting. Amazing. I'm very concerned about what questions to ask. You should send me any if you have some.

I will totally send you.

I will totally send you some. Okay, make me look smart.

Breaking the Habit of Being Yourself, Joe Dispenza, very big believer in manifestation and energy. And then not a required reading, but it's around— is the book about my grandfather, which, you know, successful entrepreneur in Iran. And he was actually like a Rockefeller type of Iran before the revolution happened. And he had a book written about him, about his values. And that book is what led me to creating the first clone of him in 2021 so that I could learn from him and learn from his values. And so, yeah, have I sent you a copy of that book? No, I gotta send you one.

I sent it to all our investors. You know what you need to talk about is this idea of like I hear some of the richest people in the world want to talk to you guys about creating clones or digital minds of their more senior members. So let's say this isn't the real one, but let's say the Valentino family or whatever might want to create a version so that, that, you know, patriarch never dies or matriarch never dies. Is this something that you see the rich and famous doing? And in the future, will all of us? Be able to talk to our AI grandfather version?

Yeah, I definitely don't think it's only rich and famous. I mean, everything we do in our life is almost in the pursuit of legacy, whether it's business for our family. And, you know, we live and we die and then our impact is gone. What does it look like where the decisions we make today have the potential to truly impact the life of someone 200 years from now? Like, we already read books of people from the past, but imagine if something like this existed 500 years ago. Imagine you could learn from all these people and you find out that people 500 years ago had the same problems that we have today. You know, so insecurities, love, business, ambition. And so not only I think is it good for family and connection and passing down values, as that was a huge inspiration for my grandfather. I didn't want the life to, that he lived to be lost. I wanted to make sure that the values and lessons he learned, I could apply. Yeah. So that you could stand on the shoulders of giants who are your ancestors and at least keep their legacy moving forward, but also for the future to learn from the past because, you know, history repeats itself. There are mistakes, but maybe that doesn't need to be the case. Oh yeah.

Well, also, can you imagine? I mean, right now my, my grandparents are going through dementia and, um, and my grandma just had her third stroke and I'm incredibly close with them. And so watching them slowly lose all mental function. Separation is one of the harder things I think I've ever done in my life. Yeah. And I could only imagine what it's like for my father. Those are his parents. And, and I think about often, God, what are the questions that I wish I would have asked them before? Like, what, what does it take to be married for 60 years happily where they like love each other? You know, what, what lessons would you have for having kids who are so close as a family unit that, you know, we get together constantly and are always laughing? Like, the real things that matter in life, not, you know, not what your text was, but, uh, something deeper. Yeah. And if I had gotten some of that from them, that would have been priceless to me.

Yeah, I mean, it's, uh, same thing with my grandparents and my parents. Um, that's why we're working on this interview mode feature where you can get them to answer questions about themselves over a period of time. Luckily, my grandpa had a book, but not everyone has a book. So how do you get that information down so it can be preserved? Yeah, we've—

that's such an interesting idea because for those that are close, you know, with your family, how many times do you— at least for me, I always go to my dad with work stuff. And he's— it's not so much that he's done exactly the same thing, there's just a wisdom. And trust. And trust.

You know, he wants the best for you. Right.

No exterior motives. Yeah. No real— unconditional love. Right. Yeah. And so what would that be like after he was gone to still be able to talk to a version of him? Oof.

And even your kids to be able to experience him the way that you experienced him.

They just get on TikTok, watch me dance. You know, they'll be perfectly fine. Oh man, TikTok. Yeah. Really fascinating. Okay, where I kind of want to end is, um, with all of this going on with AI, there's so many shiny objects, things we can chase. Where should we really be narrowing down today? And how do you decide what is important to you to focus on in a world in which information is totally abundant?

Yeah, I mean, shiny object syndrome is abundant right now. There's like a new thing every day on Twitter. Uh, I think the top things to keep in mind, it is easier than ever to build a sexy demo that goes viral on Twitter. And the fact that it goes viral incentivizes more of that. And so don't get sucked into the hype of a Twitter demo because it is much harder to build something that is scalable. The initial Delphi demo was super easy. Actually making Delphi useful was much harder. Um, number 2 is I actually think pay attention to the motives of the founders of the company. Because not only does that matter for someone who's building like a get-rich-quick scheme and then the AI product that you're using is no longer active. Two, it matters for your data. Uh, you know, you don't want to work with a company that you are talking to or giving your data to, and then they're, they're using your data to train their models, and now you're losing your IP and moat. Um, so I, I actually think getting to know the intents of the founders are really important. And then the last thing is back to what we've been saying over and over again, thinking with the end in mind. So an example of thinking with the end in mind is right now, if you wanted to build an app, uh, you may spend $200,000 hiring some mobile app development firm to build it out. But if you look at Lovable and how fast AI software engineering is moving, potentially in 6 months, you could just pay someone $40,000 to leverage AI tools and just get the same outcomes. So I think timing is an important thing too. Like, should, should you pay to do this now, or should you pay to do this later? And what are the things you should pay to do now? Yeah, interesting.

Yeah, you know, the thing about the founder's intent, you just really hit home for me because you think about 23andMe, right? Bankruptcy proceedings. They have my data.

And it's going to the highest bidder.

And anybody could have it. Yeah. And I have no rights to that. And that has material consequences in a world in which, you know, biotechnology looks different. And, um, and so it's a really good question. How do you figure out a founder's intent? I mean, not, no, no shade. And I don't know what the right side or wrong side of it is, but, you know, Sam Altman and OpenAI is a great question in the beginning. Lots of people could say, well, intentions looked good, and now there's some questions on that. How do you think you actually figure out somebody's real intentions when you're going to give them your dollars and your data?

Yeah, I mean, I am a huge fan of Sam Altman and OpenAI, so I have a lot of respect for what they've done. But as you said, you know, the shift from nonprofit to profit and then training models on people's data. And then there's that one Polygram essay where he says if Sam Altman was on an island, he'd become king of the island. Yeah. and it's like, okay, there's certain data points on people that you can get to see like what their true intentions are. So Delphi upfront, we're like, you own your data. We're not sharing with other people. It can be deleted at any time. Super important for us. And so some founders don't have an online presence, but I almost feel like it's a default for every company these days to have a manifesto, an about page, like why are they doing what they're doing? What inspired them? What the vision is. I think clarity and thought of the vision and the, and the reason I think is important. It's a really good question.

So you should be asking, if you're gonna use AI tools today, who owns the data? Mm-hmm. What happens if I don't want you to have the data anymore, company? Yep. And what's your background? Why are you doing what you're doing? What other questions should you ask somebody before you give them your dollars and data?

What insight inspired the company? Was it like a first principled insight or was it like a get rich quick insight? Yeah. Um, what's the vision long-term? And I, I think another example is, let's say a software says, hey Cody, uh, put in a topic and we'll generate a million SEO optimized blog posts. And you take a step back. Another question you can ask yourself is, okay, what does the world look like when everyone has access to this? Mm-hmm. Is this a tool that I wanna be associated with? Do I wanna be a part of the group that is flooding the internet with SEO-optimized blog posts? Probably not. I think that's gonna get a bad rep.

Yeah, no, I don't. Interesting. Okay, in closing, for somebody who wants to get on this journey, one, they should definitely go to Delphi. They can play with your clone live so they could actually talk to you.

Ask my Delphi.

Yeah, if you like anything that Dara's talked about today, you can literally go there and have a conversation with you, which I think is really cool.

Yeah, ask all your podcast hosts.

Yeah, you literally can basically say, say, um, I like something that happened and I'm watching somebody like Dara put it into action, which I think is also a really good, a good thing to do. A lot of people are talking about things. That's a big problem that I have with AI today. It's like, okay, you guys are all talking about theoretically what AI could be doing, but can you take it all the way to fruition? We just did this on a, on a, a YouTube video. I saw this, um, thing on Twitter where a guy started talking about being able to do this XYZ activity. When we did XYZ activity, what we found is wasn't exactly the way it was portrayed. Yeah, it took way longer. Yeah, it was way harder. There was much more human interaction. And so when we did it, we were kind of like, all right, at the end you could do this. But currently it's like watching a toddler with an iPad. Heavy supervision. Yeah, lots of interaction. And so I think seeing people do it live is super important. But for somebody else who just wants to start on the path of AI, is the answer just go to ChatGPT and start playing? Or what's the playbook for somebody that wants to learn more and doesn't know where to start?

I think first thing is first principles. Like, really try to understand the why of all these things and build up your foundation from the ground up. If you start with floor 3 without floor 1 and 2, building's gonna fall. Mm-hmm. You gotta build that foundation and it takes time, but like, that's, that's how you're gonna grow a strong, tall building. Uh, two, with that first principles, having a long-term view. The long-term view is what it's gonna set people apart. Three is, like you said, getting started with ChatGPT, being creative, not being afraid of things not working. Yeah. And then four is consistency. Uh, because there is shiny object syndrome, because there is TikTok, attention spans are lower than ever. And people are less and less willing to like do the work for long periods of time. And I think that's actually like a huge competitive advantage if you're able to do that. Yeah.

Consistency plus grit. Yeah. What are your favorite, um, GPTs to use? Like I use Perplexity a lot and I use Grok for Twitter, be just like news-based. Um, do you have different ones that you use for different things?

Yeah. O1 is great for those consulting-like projects. Okay. Um, $200 per month, but I think totally worth it. Claude is particularly good for, uh, language and writing. Specifically, you know, if you're emailing someone, everyone speaks different languages, and I don't mean languages as in like Spanish or English. I mean, we perceive things in different ways and we make assumptions and projections on what you actually mean by certain words. So I use Claude for very difficult emails, fundraising, negotiations. Hiring, you know, this is what I— this is what I'm saying. Help me change the wording a little bit so it doesn't come off the wrong way to this person. Wow. Great for that. Interesting.

Do you use Perplexity or Grok for anything, or are those really just the two you use?

I use Perplexity. Sometimes when I meet someone, I'll ask Perplexity, tell me everything I need to know about this someone. Yeah. So like I can be prepared. I use Delphi, obviously, for learning. Um, and I don't think we use Grok or Gemini yet.

No. Okay. For anything. Interesting. Okay. Let's close, uh, with this. Um, if I had to say, are you positive on AI versus negative on AI, you would say?

Long-term positive. Short-term, think there is going to be anxiety and structural change and like meaning and it's going to take some time for people to fully accept. And I imagine when we found out that Earth was a planet in a huge universe and it's actually not the center of the universe, I imagine people at that time were like, oh my God, what am I going to do? I'm not the center of the universe. And then they got over it. And I think there's going to be a similar transition period. I think the government will need to adapt. I think— I do think UBI is important. Yeah, but long-term optimistic.

Okay, Dara, you're the man. Where should people go to find you?

Dara Ladjevardian on Twitter or X, and then delphi.ai/dara-ladjevardian. Amazing. If you wanna talk to my Delphi.

And I do really like your X or Twitter. I just can't call it X still for some reason. I can't. Like I'm Xing today.

It just doesn't work for me. The brand doesn't feel right.

Uh-uh. But I do like what you tweet about because it is from this first-person view of AI, Plus you got a little edge, and I'm into that. So thank you so much for being here today.

Thanks for having me, Cody. Yay!

You and your teens trying to figure life out. What do you wish that you could tell him that you know now that he didn't know then? So we're gonna give you this, you write it out, and then you'll read it to the listeners like you would to young you.

Uh, well, to my younger self, I would say trust the process. The world will pull you in a million directions and make you doubt yourself. And most of them know nothing. They're all just figuring it out. And your best bet is to know who you are, what you care about, follow your intuition, and trust the process.

That's beautiful. Thanks, Dara. Thanks, Abby.