The Exchanges

Q learn your voice. So it can write your, it can help you write your emails or your blog posts in your voice, which is kind of like the next level up from autocomplete, next level up from Grammarly. It doesn't just kind of correct or spell check your stuff, but it actually writes the way you write because it has watched the way you write. Is that the thesis here?

A Yeah, yeah, absolutely. And I think it can be, you know, lots of different types of business communication, but especially like email. So I don't know if this is, this is actually my friend Mike Vernal's idea. I think he suffers from the same thing I do that might be true for you, which is like I'm an incredibly picky writer. And so I will use the models today for generation of basic content, or I'll ask my amazing EA to like draft emails for me, and then I will go rewrite the whole thing. Because I don't like the tone because it doesn't sound like me or because it's not tight enough or because I want to use a certain phrase. And I think the next level of like value and impact is definitely going to be, um, fine tuning to specific voice.

AI assessment note: “Yeah, yeah, absolutely. And I think it can be, you know, lots of different types”

Q is a bunch of work, like creating the store listings, changing prices, writing the copy, all of that. And I think what you're saying is somebody should be able to just like have a product and then the AI should be able to like do the actual e-com management of the, of the sell of, of the, of setting up the shop and running it. Is that what that means?

A Yeah, I think, like, um, it's probably, it, it matches, like, a larger theme that I really think is exciting about AI, which is, like, because all of these skills, and it could be, um, run a basic, like, social marketing campaign, right, or, like, send email to your customers that are likely to be repeat customers, or improve your website for indexing, like, there are a bunch of things that, um, are probably not related to, let's say it's a, Let's say it's a Shopify dropshipping store for, like, a particular type of sock, and you love socks as an entrepreneur. It's not, like, related to the merchandising decision or the design decision of, like, what is the sock I want to give the world.

AI assessment note: “Yeah, I think, like, um, it's probably, it, it matches, like, a larger theme”

Q What are some of your, like, hot takes, uh, or maybe your contrarian takes? Anything that you think that might be counter to what the most people say, most people do, most people are betting on? Do you have anything that is against the grain?

A You know, I'm gonna get, I'm gonna give a, like, a somewhat arrogant answer, which is, I don't spend a lot of time trying to figure out what the entire market thinks, actually. So I'm like, I don't know which of these things are contrarian. I can tell you where, like, my opinion Has changed dramatically. Like, let me give you one example. For many years in including, you know, the tenure of my investing at Greylock, I was one of several people who were like, okay, we're gonna go understand healthcare and digital health. And I was like, oh, healthcare sucks, right? It's a quarter of the economy. It's really important. How could you not want to work on this mission? But it is so slow and the incentives are so screwed up that like trying to enter that market with technology or the speed of entrepreneurship that, you know, Silicon Valley entrepreneurs are seeking is like not a good idea. And we just did a healthcare administration automation company. So I'm like, oops, like changed my mind, real hypocrite here. And like one of the reasons being, I'm like, well, if you think about the mind numbing work that happens in healthcare administration, like billing, authorization, coding, claims processing, like all, like even, not even mind numbing, but just like expensive and manual, like patient support. It's actually really fertile for an AI company. And you know, we back something that…

AI assessment note: “I don't know which of these things are contrarian. I can tell you where”

Q And you said that you started the podcast by saying there's like a lot of, you had a real, the line you said was awesome, which is like, there's a bunch of ways to make a million dollars that could eventually become a billion. Is this one of those companies where it started that way?

A Uh, I think, I mean, I think the market for this company is very deep because people, like, they want a lot of video, and I think more, like, if you just think about the domain of making video, you guys know much more about this than me, but, like, people want a lot of control, right? They want quality, they want specific expression and brand and motions, and they want, like, one person, two people, three people, like, person walking around, product, whatever it is. And so I think there's actually a lot, like, there's a lot we still cannot do with research, and so the company wants to continually pull, like, push the bounds of what you can do. And so I think this is a good example of, like, I think there's a billion dollars of video generation revenue for them or for others, but, like, you know, you actually have to invest in the product pretty deeply, but it doesn't mean that your wedge can't be really powerful across a single use case.

AI assessment note: “I think this is a good example of, like, I think there's a billion”

Q Well, have you heard any, what are some, what's a wild example of how people use this for hacking or for fraud? Oh, I mean, like, for my company, we get emails from me, it's not really me, and sometimes it will have, or, um, like, it'll have a link to something that sounds like it's in my voice.

A Yeah, I think that's the simplest example, which is, well, like, what happens if you can create really Authentic sounding media. Like, you know, are, are your parents, like, you know, going to not pick up the phone if it's a spoofed phone number, and it sounds like you, and you say you need something? Like, that's a bad scenario, and so I think we need more tools to protect against that, and general education about it. So I, I worry more about that, and then I'd say, like, I think of the probability of a bad scenario, I said was like, It is possible. I can't see exactly how we get there, and if you ask me, like, what are the reasons in which, ah, broad use of cheap intelligence are gonna be great, I can give you so many reasons, right? So Andre Karpathy just started a company around education, and, ah, like, the, the, the fields that have been super resistant to cost improvement, basically healthcare, the government, and education, like, I, I think this will actually move the needle on some of the domains that matter a lot to all of us humans. Um, and I, I think like when, when people talk about like the doom scenario, it's really fun and scary to talk about the dystopian doom scenario. But, uh, I think the opportunity cost of not exploring the ways in which like there, you know, you can have an economy of abundance. We need to talk about that. And that is really what I focus o…

AI assessment note: “if you ask me, like, what are the reasons in which, ah, broad use”

Q Okay. What are the other ones that are interesting? So there's character AI that has like some absurd amount of traffic, but I've also heard some things about like, I don't know if this is all legit traffic or what, but there's character AI. What are the other ones that are interesting or what do you, what do you, who's doing well?

A I want to touch on character for a second, because I, I think like, you know, when you look at consumer companies, one of the things that I learned was that The behavior patterns, like when something just really, really stands out from all other products in their category or previous categories, that's when you pay attention. It's like the, you know, dumbest metric, but it is really clear when something has special consumer behavior around. And the thing that is really interesting to me about Character or the companion apps that work really well is like people spend hours with them, right? Like, you know, in terms of the number of products, like, How many products do you spend hours with every day? Not a lot, like social media.

AI assessment note: “I want to touch on character for a second”