David Waxman of TenOneTen

Welcome to the Puck, Venture Capital and Beyond, a show that explores the evolving landscape of the venture capital world. We'll have candid conversations with today's VCs and entrepreneurs who are shaping those changes. I'm Jim Bier, managing partner of Bier and Troff and president of CMBG Advisors. This podcast brings change makers to the table to discover the inner workings behind decision making strategies and ultimately how they got got to where they are today. Francis Bacon said knowledge is power, and I'm sure he was the first person to say something like that.

You know, companies have always run on information. What we're seeing now is that there's all these big industries or verticals that are getting completely turned over with the ability to harness data and use it. Today, I have the pleasure of sitting down with David Waxman, managing partner at ten one ten Ventures. David shares his journey from successful tech entrepreneur with companies like Firefly and People PC to leading investor whose exacting eye is helping to shape the future of data and the innovation of businesses around the world. Welcome, David.

Thank you. Before we jump into some intense discussion about AI and data and stuff, why don't you take a couple minutes and tell us how you got to where we are today? Like, what's interesting about your background, where you came from, kind of the companies that informed your viewpoint on the world today and so forth, and let's go from there. So Sure. I mean, one of the more unusual things about me is that I have a music degree.

There's a couple other VCs out there with music degrees, like David Hornick from August Capital, but most don't. I went to undergraduate at Cal, Berkeley. And I was always fascinated by computers and music. And I thought computers would be the way that I would make sort of side money, which I was doing as a programmer for a game company. And that music was really the thing that I wanted to do as a passion, and maybe a career.

And so, in college, I figured out that I didn't have to be exclusive to one, and ended up getting a degree in computer music, which was a way for me to blend these two passions. And I continued with that for a while before ultimately starting companies and then becoming an investor. So, I got the degree in computer music. I went and worked in a research facility in France for several years that was dedicated to the studies of how computers and music and electronics and music could coexist, called IRCAM. And while I was there, I met a guy from MIT, from the media lab, who said, you should come and continue your your work with me, and that professor, Todd Machover, took me over to the lab and spent two years there.

And I didn't really expect to start a company. I didn't know what venture capital was. I didn't know much about starting a company. To me, starting a company would be some kind of consultancy, But I met a kid from Harvard Business School who knew a lot about starting companies, or or at least he had been educated in them. And he and I and a couple other students from the media lab started our first company called Firefly.

And from that point on, I was hooked to doing startups. You know, we we knew very little about the internet. Nobody really did know anything about the internet, and this was 1995. Wow. But I knew it was exciting, and I knew it was gonna change everything.

And rather than stay for a PhD at MIT, I decided to go do this startup. I I actually ended up working with that cofounder for three companies. So I ended up doing three venture backed companies. We sold the first one to Microsoft in 1998. Then we started a second company called People PC, which we took public in 2000, and then sold to EarthLink here in LA.

And and then had another company which brought me permanently to LA called Spot Runner. That was a career of about eighteen years of starting and running companies before I decided to become an investor. Interesting. One of the things you talked about, you were saying that not a lot of people have a music background. They kind of go on a venture, and I think that's obviously true.

But in talking to some of my friends that are into music and they're also good at math, there is a correlation between music and math. I'm not good at music and I'm not good at math. But from what I understand, at least, from those people whose brains understand music and math, that there is a correlation between it because music is apparently very mathematical. Well, well, first of all, let me go back and correct something you said. Okay.

I didn't say that most VCs didn't have a musical interest. Okay. There's a lot of VCs who are in bands. Right. It's a pretty good band.

So you've got, Tim Chang at Mayfield plays in a band. You've got the guys from Foundry. Several of them play in a band. So there are a lot of VC bands out there and a lot of VCs who like music. I'm just not sure how many of them pursued the role professionally for a while.

While. And I think what you said is also true. I've seen it to be true that that a lot of musicians are also mathy. Not all. But MIT, for example, had a really good orchestra and really good music program, and a lot of people who went there were instrumentalists or composers.

So moving from starting and running companies to founding ten one ten and actually getting on the other side and backing these companies, what what led to that transition? Well, I'd always thought that that would be an interesting thing to do. I was too addicted to doing the startups for the three companies that I did to to do anything else. But after this third one, I started thinking about different ways that I could participate in the same ecosystem and the same kind of world that I'd gotten used to. Investing seemed like a pretty natural fit.

I had already started advising and angel investing, and it's a little bit of a slippery slope. Gil, my partner, has done a lot of angel investing as well. So I found it this is kind of a scary analogy to put out there because it makes me feel old, but it's kinda like going from being a parent to being a grandparent. You have a different relationship with startups. You have influence, but you don't have control.

Right. You get to play with the baby, but you can give the baby back, which is pretty awesome. And you care a lot. You care deeply, but it's a slightly different relationship than something you have with your own. Sort of one level emotionally removed from it being your own company.

So it's been great. I feel like I could share everything I've learned and everything I'm still learning, and do it amongst a lot of different companies. And it's, frankly, a little less insane than starting a company, which is super, super hard. Right. Tangentially, why Southern California?

I love Southern California, for instance. But is there anything that particularly draws you to here as opposed to, for instance, places like the Bay Area? You know, I like both places. I actually grew up I was born in Santa Monica Okay. At Saint John's, but I grew up in Berkeley.

Oh, okay. And really thought of myself as Berkeley being my hometown. Still kinda do. So I I've always had family and affection for both places. The thing that brought me to LA was my last company, Spot Runner.

My partner, I explained, I had I had already worked with the same partner for quite a few companies, and and he got the third one sort sort of started here. And I wanted to come join him and do another one. So, really, that company, Spaw Runner, brought me to Los Angeles, and then I learned to really love the place. It grew on me very fast. And it's been amazing to see this place grow.

So in 2005, when we started Spar Runner, it was very, very hard to put together a good engineering team. We did, but it was a tremendous amount of work. And it was nearly impossible to get people to relocate to Los Angeles, particularly from the Bay Area. So we've had this phenomenon where we had then and we have now all these great universities that create engineering grads and folks that would be great as part of the startup ecosystem, but they were leaving. They weren't staying here.

They were going to San Francisco, in the Bay Area, or New York, or other places. Since that time, we've gotten enough, sort of, critical mass that people are coming here, and people are relocating their companies down here from Northern California or other places, because it's known to be a great place to have a start up. And I think the key to that, ultimately, is the talent, is the actual employees that make things go. Now you can wire money from up north. The lawyers will be there if you need legal help and the accountants and all the things that are around the start up ecosystem, but it takes a while to get the talent liquidity that's required to really have a a vibrant startup ecosystem.

So I remember meeting someone, really, in the early days of Snap who came down from the Bay Area, and he told me, well, I knew I had to come down. I see the trajectory of this company. I know it's an important career move. But if it doesn't work out, I'll just have to go back to the Bay Area. That was the way people thought.

Like Right. You know, this wasn't that long ago. That was maybe 2012 Right. Where LA was like this island. And if your job on the island doesn't work out, you go to the mainland.

It's not at all like that anymore. So with Snap growing and Tinder and Cornerstone and I could name a 100 companies, Drew Carr, with the critical mass. Now here, people know that they can come to LA and make their career in tech. And so it's changed in a wonderful way. Yeah.

That's the thing I've seen. Because I started doing two things as a tech lawyer. And when tech companies would get in trouble, they would come to me as the cleanup guy. And I always told people back in 2001, the good news was I did more VC deals than anybody, but the bad news was it was LA. And there just wasn't enough infrastructure down here.

So then, when we had a slowdown, it was kinda like you had to go back to your day job. Whereas now, and one of the reasons we're doing the Puck is to get the word out there that, hey, LA is really on the map. Southern California has all this dynamic stuff going on. And we don't need to kind of be the the stepchild, so to speak. And it's it's exciting to see that we have unique things.

And as you mentioned, the universities, we have this infrastructure down there that people weren't aware of. We're we're not even focused on it because as you said, they would leave. So I think it's an exciting dynamic time for Southern California and people like you who are building it. For me, it's very exciting having kind of lived through that. So let's talk about where the world is going and things that I don't understand well.

But you've got data and you have AI. And there's a relationship between the two. But can you help us understand kinda what intrigues you about where AI is going and and the relationship between AI and data? Sure. And they're yes.

They're very, very closely linked. So let's first just talk about data generally. It's not a new idea. Right? Francis Bacon said knowledge is power.

It's been around for a while. Right. And I'm sure he wasn't the first person to to say something like that. I think he's just the first person credited with it. You know, companies have always run on information, but information often very poorly transmitted or inefficiently transmitted.

You know, when one person has a Rolodex, they go and have lunch with somebody else, there's no record of the conversation. There's much that anyone could do with that as, like, an asset, the information that was traded in that meeting. What we're seeing now is that there's more ways than ever for companies both to collect in a more structured way and to harness data so that they can use it again and use it for different things. So the comparison to that sales meeting that had no record before, maybe that person is now using or a similar person is using Salesforce, and that has a a complete record of what happened and what's supposed to happen next and how much contract value was discussed and all those things. And I'm using that example because I think it's everywhere.

It's very pervasive that collecting data in a way that it can be used is starting to permeate pretty much every industry. And if it hasn't yet, there are still a few holdouts where data isn't being used very well. It's going too soon. And now, you know, that Salesforce example is on a really small scale, but imagine that every interaction with every customer is now recorded in some way. And now you can look at trends, and you can look at problems that might not have been apparent with your product or ways that you can add a feature that your customers really want, but it wasn't getting through to the company before.

And then imagine on top of that with AI, and we'll talk a little bit more about that, that you can start to automate some of those things. So you don't need necessarily a person to glean those insights from looking at the data. You can have a computer show you, hey. You're missing this thing or fill in some of those blanks so that the salesperson doesn't have to or make other inferences that that will help the business go. And we're seeing that across industries and in the most important industries.

So the way we eat through ag tech and the way we get treated for disease through health tech, the way we entertain ourselves. There's all these big, big industries or verticals that are getting completely turned over with the ability to harness data and and use it. So how using AI as an example, can you give us some examples of where AI is helping the typical business or could be helping the typical business and where again, where it is today and where you think it'll be in five years, for instance? I kinda bristle at the term AI. Maybe we should have talked about this before we started the podcast.

But it's a big umbrella. Right? Okay. And it means different things to different people. For some, AI means, you know, the the old Turing test model of, can a computer act like a human to the point that it can fool a human?

And I don't think that that's a really good model for what computers can do. I think what people are mostly referring to is that computers can start to do things that they weren't able to do before, that only humans were able to do. But often, they do it at a different scale and a different speed, which makes them better than humans at the thing that they're doing. So, for example, humans are very they have an uncanny ability to identify and classify objects. Right?

A little kid can see dogs, a few dogs, not not a lot of dogs, and start to get the concept of dogginess in their head somehow. I'm not sure that's really understood how this works, but they can meet a few poodles and a Shar Pei, and then they see a Great Dane, and they know it's a dog. Like, that's that's a hard thing. Right. If you think about that, they don't look a lot like each other, but there's some essential dogginess that's understood.

And it's not, quote, unquote, big data that feeds it. It's a uniquely human ability. Now, computers do it a slightly different way. You can train a model with lots and lots and lots and lots and lots of pictures of dogs, and it can also get what's basically an idea of what dogginess means. And once you have that information, unlike a human who's limited to how many dogs can you see in a day and classify as dog, Quite a few, but you can't do a million or 10,000,000, and that's the kind of thing that a computer can do.

So I I guess we strayed a little bit from businesses. Well, using it, let's look at something like you mentioned health care, for instance. I know there's this whole movement towards electronic data, and there's HIPAA issues and privacy issues. But one of the things is that people have health records. And they'll go to a doctor, and they'll be written up that they have diabetes or they have some particular issue.

And And then they go to the hospital and then they get treated for that. And then they go to their chiropractor. But people code and write these things up in different ways. And yet, there's this whole movement towards being able to access these different records and understand them. Do you see, for instance, computers' ability to kind of correlate that data and standardize it as something that, for instance, would help in the health care field?

Yeah. I I think humans need to do a lot first, and it's about legislation and the way things are done. I think health care is a great example of where AI can help in a whole lot of ways. But as you say, the data are kind of in different silos. The industry is sticky because there are serious privacy concerns and lots of things that have been done a certain way for a very long time.

So it's quite hard to do, but it seems inevitable that it will be done, that a computer will be able to look at your records and your symptoms and all the literature, and be able to suggest to your doctor diagnoses for your disease. That seems inevitable to me. This is not getting rid of doctors. I think doctors can do a lot Right. With just talking to a patient and being the the interface, you know, that they can understand the patient and what they're thinking and catch things that might have been missed.

But there's no reason that a doctor should have to have read the entire medical literature. Right? And it's it's not really possible to do so. So why not have that help? So the doctor can say, here's what I'm seeing.

Here's what everybody else has seen about this patient over time. There's some obvious diagnoses and things that seem likely, but why don't you tell me about everything else and maybe I'll catch something that I missed. So one of the questions that I have is we hear the term AI, which, as you said, is probably overused and so forth. But as we're sitting here today in our everyday lives and we're using voice recognition or using our devices more and more, can you give us some practical examples of where it's being used in the real world right now that we may not even be aware of in terms of products that we're using and so forth? So, let's take Amazon Alexa where there's a bunch of layers of AI.

So, the first thing, when you talk to Amazon Alexa, you it turns your voice into words. Right? And that's something that traditionally humans have done really well, but computers have not done so well. And that's using AI or some kind of machine learning. Then to turn those words that you're saying into some kind of semantic understanding of what you said is also AI.

So if you say, I want to buy more laundry soap, there's turning that signal of the sound into those words, I want to buy more laundry soap. And then there's understanding that sentence of, I want to buy more laundry soap, which is a harder thing to do. And it's something that can be done with machine learning. Then once you make that order, the system might be smart enough to say, he wants to buy more laundry soap. Let me help with that.

Would you like to buy more laundry soap now? And you could say yes, and you have a little interaction with the computer, all of that would be based on AI. And then on the whole back end of Amazon, the fact that they have the laundry soap you need in the distribution center that's near you is probably using a lot of machine learning. The fact that their whole server farm that is used to run the systems, there's probably a ton of AI used to make that more efficient. When you get to the distribution center that has that laundry detergent that you want, there are probably robots there that are moving that laundry soap from where it's stocked to where it needs to be to get packaged and sent out to you.

So, all of those things are probably using some element of AI, and you won't necessarily know what's back there, But there's no way that that process of saying that you want more laundry soap, ordering it, getting to you, and all on Amazon time, which might even be the same day, could possibly happen without machines. Interesting. So you were talking about people having sales meetings. And in the old days, there was a book, I think, that John McKay wrote, like, Swimming with Sharks. But he would talk about how if he'd go out and have a prospect, he'd come back and he'd write down their birthday, and he'd write down what sports teams they liked, and how many kids they had.

You know, he'd physically write this down and then he'd put it in his Rolodex or he'd write it on a note card and so forth. In the world of sophisticated sales and business, can you give us some examples of how AI is facilitating that process? Sure. So if you had all that data in a system, birthday, what was said, all those parameters, and you could look at a bunch of instances of that, maybe across your whole company, everybody's conversation. You might be able to figure out or an AI might be able to figure out, okay.

When should I use one of these opportunities to reach out? What's more efficient? Should I call them on their birthday, or should I write them when they've got a new publication out? And maybe what should I write? So there's a lot of sales tools now.

One just got funded by Andreessen Horowitz, People dot ai, in that space that's helping salespeople do that kind of automation. So it's not replacing you, but it's making you a heck of a lot more efficient. Yeah. That's a great example. I mean, for instance, let's say I'm interested in David Waxman and I'm interested in your career path and I find what you talk about interesting.

Are there programs, for instance, where I could say, hey, I'm anytime David writes an article, for instance, or he's on a podcast, I'd like to track that. I mean, AI would be a tool for doing that. Right? Because it would be searching the Internet and kind of keeping up with and knowing to look for different things where you come up. And then, is this the right David Waxman or not the right David Waxman?

I mean, is that an example of things that people are doing now? Sure. I I don't know if for that basic thing of just looking for instances of me appearing on the Internet, you need a very sophisticated Right. AI. But as you said, as soon as you have two David Waxmans and you wanna figure out which one is the David Wachman you care about Right.

That's a problem that where machine learning can help. It's funny. Amazon seems to be taking over the world in a lot of ways. And one of the things I always wondered was, for instance, why it's not sending you out reminders saying, here's all your friends, here's their birthdays, here's what they're interested in terms of, like, what you should be buying them for your their birthday presents or their anniversaries or otherwise. And maybe that's being done and I'm just not even aware of it.

But it's it seems to me again, like, if you could organize that data, you could use it for knowing people's restaurants, what to buy them. I'll, like, try to send a client a bottle of wine. But I mean, for instance, organizing what wine they like or different things like that, it takes a lot of time. And I'm wondering if there are tools, for instance, that are developing that help us better understand trends and what people like and how to keep track of all that. Yeah.

I'm sure that they could do that today. I think where that gets really sticky is around privacy. Right? Right. And that's the other side of collecting a lot of data, particularly about people, is that it can feel or be invasive.

Like, maybe I don't want you to know that I drink at all. Right. Or maybe my hobby is archery, and I think it's too weird for you to know about. Right. I don't necessarily wanna share that information advertising targeting that happens on the web, those kind of recommendations are already being made to you.

Right. It's just companies have to be a little bit subtle. They're not always successful with that, but they do have to be a little bit subtle about how much they know. Because if you actually saw how much they knew, it might freak you out. Right.

Yeah. And it seems like it's growing exponentially because, you know, you'll be on the web and you'll have ordered something from Amazon or you've searched for something on Google. And all of a sudden, you're on Facebook and you're getting inundated with similar products or vacation recommendations and crazy stuff. And it's like, big brother is clearly watching. So Yeah.

Including, by the way, I know because I just recently bought a new Audi. It's amazing how all of a sudden, I'm getting all these car advertisements and crazy things relating to cars. So it's out there and God only knows how, but it it's people are definitely aware of it. Yeah. Once again, if I were able to print out all the data that Facebook and Amazon and Google, just those three, have about you, you would be alarmed.

Yeah. I'm pretty sure. You know, and that's a trade that that a lot of people are willing to make, and we all sort of are implicitly making it by interacting with these services. No one's reading the the privacy policies or the terms of service, but what we're all agreeing to is that they can have our data in order to provide us more convenience. Right.

And is tracking that data and, again, using that data to make money and to make recommendations. That's all controlled by AI. That's that's an example, I would assume, between how data and AI are working together. I would go back to using the term machine learning for that. That's great.

But for sure, all the inputs to that problem. Right. Are, you know, the problem of knowing what ad to serve you. Right. All the inputs are data.

Right? It's who are you? How old are you? Where do you live? You know, where have you been?

What's your favorite color? Like, all these things. And then maybe a whole lot of data about people who are similar to you or dissimilar to you so that that you can be put on a big multidimensional map of the kind of shopper that you are. There is no AI without data. That's how it works.

And it seems to me like you can collect data, And if you don't have machine learning to help you utilize that data in a productive manner, it becomes a nightmare because there's so much of it. How do you know how to sort through it in any kind of efficient way? I mean, just news sources right now, in terms of what news to serve up to you, what you're interested in, what authors you're interested in. I mean, it's a morass for people that don't have the time to really kinda sort through that. And I I would assume machine learning helps all of us attract the kinda news stories, the podcast we're interested in.

I mean, again, going back to my car example, I was I was driving the other day and all of a sudden, there were recommendations of podcasts, including, by the way, the Puck, which I was, like, fascinated by. Like, it's recommending different podcasts for me to listen to and it's got my own there. God only knows why. But it it somehow, it put me together with that podcast. Yeah.

Well, that's a really interesting we could go down a whole other avenue about that because there's a a real danger, I think, of people getting into their own filter bubbles. Right? Absolutely. That that the more and more that news gets hyper personalized, the less and less you're seeing what the other guy is thinking and the other person is reading. And I think that's part of the mess we've got with media today and particularly information.

Like, I've never seen so much argument about fact before these last few years. And and I think a lot of it is that people are getting their news from a very, very tailored source. So one of the dangers or one of the challenges is that we are being served the things that we are interested in or that will appeal to us. And, like, I like the show left, right, and center, which gives you the different perspectives. But, I think I know three people that listen to the show, for instance.

And instance. And most of us are being served up news stories, you know, whether or not it's watching CNN or it's watching Fox News, or even on, like, on our iPhones when we're being given news stories. So is that some I mean, that's clearly a challenge that you were addressing in terms of the fact that we are being kept in these narrow lanes. But do you see machine learning and or any of these companies which obviously have a lot of power now from a social conscious perspective thinking through, like, their contribution to these challenges and polarization, what and what maybe they can do to kind of combat that a little? Well, I think there are definitely a lot of people talking about this question, particularly around news, but also just around bias in AI.

So machines don't have political opinions, but they do take all of the inputs they've been given and create outputs based on those inputs. So let's say you get a list of all the venture capital investments that have ever been made ever and try and predict what the next great one might be, Really hard problem. You'd see a bunch of a certain type of founder. You'd see a bunch of a certain type of investor, maybe certain types of industries, and you'd be continued whatever biases there are, and there are plenty in VC, you'd see them continue on through the computer emulating what has been taught. So that's a a real problem.

There was a pretty well known case, and, unfortunately, I forget some of the details. But there was basically a bot that was talking to a bunch of pretty right wing folks, and it got really nasty really fast. The bot was starting to say offensive things. That's funny. And it was a big company that was experimenting with it and they had to take it down because all of a sudden they had this offensive rogue bot repeating what it had been heard or what it had heard and sort of developing answers that were consistent with what people were reinforcing it with.

Well, that one of the things I did not appreciate until recently was a lot of times now, as I understand it, when you ask a question on the Internet, like, where you type into a bot, you know, hey, I'm interested in finding this product or how do I use something? And you get a typed response back. In the old days, you'd think it was a human being typing that response back. But a lot of times, that's a computer talking back to you. It's interpreting your question and then going out and finding the answer.

And that's happening more and more as I understand it. Yeah. That's happening a lot. Sometimes, it's still with a human. So it'll be a understand it.

Yeah. That's happening a lot. Sometimes, it's still with a human. So it'll be a suggested response that someone just has to to let go. And sometimes, it's probably fully automated.

I have a client that is developing a bot where if you are going to your company's website and you're asking it questions in terms of, like, you've got information about your company. So, it'll say, you know, we're a law firm and we've done 10 VC deals in the healthcare field. And then, somebody will type in, well, have you done any areas in big data? And then, it somehow searches the web and it starts to expand and grow your website based on information it's learning about what the company's doing and stuff. So that, again, it's filling out your profile for you.

Again, my mind can't even comprehend the things people are starting to think about. That's part of what is interesting to me and what you're helping me understand and for our listeners is that there are things that are going on out there and and tools that are available that we're not even aware of. Mhmm. I mean, we're all aware of the voice recognition in terms of like driving and so forth. Right.

But, in terms of you you mentioned, like, in the food industry, in in in terms of how we eat, is there an examples where machine learning is helping in that area? Sure. There are there are watering systems that are sensing in agriculture how much precisely water needs to be given to a certain crop. I was actually just looking at a company recently that is using spectral imaging to try and figure out where plant disease is before it's visible to the human eye. There's a lot of tech being used in agriculture, and I think we need more of it because we're gonna have to feed a lot more people.

So it's being used in most every big industry. And it's I heard this once from someone else. I can't attribute the quote, but it's, in a few years from now, saying that your company uses AI is gonna be a bit like saying your company uses a database. It's like Right. It's a tool, and it's a really very powerful tool, but it's just being part of the greater tool set that people are using to create products for people.

In terms of and maybe this is an off the wall question, but in terms of, like, reliability for our autos or even things we take for granted like airplanes, is machine learning helping make things more reliable in any way in terms of the everyday products we use and things that we sometimes take for granted? Does machine learning have anything to do with how reliable things are, for instance? Well, sure. Let's talk about self driving cars. We haven't really talked about that yet.

Ton of machine learning and and AI used in a self driving car. They need to perceive the environment around them. They need to understand the controls of the car, the the conditions, and make really minute adjustments at very fast speeds with very high stakes. There you know, people can die if the car doesn't do the right thing. I would trust an AI driver of a car way more than I would trust a human.

Right? Because AIs don't get distracted. They don't get calls on their cell phones, and they're pretty quick, so they can tap on the brakes faster than than you or I can. Now there's still some work to be done to make them up to the level that they need to be, and it's gonna take a while for them to be adopted. But But I was kinda hoping that my children won't actually drive cars, that they'll have computers drive cars because they're probably a lot better at it.

Well and we have done some stuff on self driving cars, and that's a fascinating, fascinating area. And I'm wondering if there are other areas like that that are are less known, or that'll be the next thing after self driving cars in terms of helping us just think through whether or not it's helping remove make traffic better. I mean, I guess Waze is machine learning in terms of traffic. Absolutely. Yeah.

Right. The app literally saves me hours a month Right. As an LA driver. You know, it saves me time off of almost every trip I take, sometimes driving me in very weird ways. But that's another story in things like a computer.

But, yeah, that that's something that we're all using day to day is is nav and that kind of understanding of which way to to go to get somewhere fast, it's AI and it's big data. And that's an example, like, with Waze, for instance, which I love. It's changed my life dramatically as well. But what's fascinating is when it first started out, it would often take you places where it would tell you to cross Olympic where there was no red light or whatever. Right.

And I haven't even paid that much attention to it, but I assume it's getting better at that. Like, it's knowing, you know, if you're gonna have to cross a really busy road. And yes, it would technically be a lot quicker because there's no cars in that road. But oops, there's no stoplight there. You know, you're gonna get stuck.

It's fascinating. I mean, I I'm assuming those are the kind of things that people are thinking through in terms of how do they program that into the Yeah. I assume that that's getting worked on as well. I mean, that should become apparent in the data. Right?

If a lot of people go to, let's say, a crossing of Olympic Boulevard, and if you look at their signal, the car is going, going, going, stopped. Right. Stopped, stopped, waiting to cross, that's interpretable. That's that's data that that you can interpret and say, okay, that doesn't work. Even though there seems to be light traffic on Olympic, it's actually impossible to cross.

And I I hope it's getting better because I do run into some of those issues as well. But, yeah, I found the key to Waze is to try not to improvise on your way Because often, it'll take me on these really weird routes. And every time I say, yeah, this isn't possible. Like, I gotta turn over here. I can just turn right over there.

I end up getting caught in traffic. Right. So in other words, let let the machine take you or Yes. Yeah. Give way to Waze.

Right. But there are still these edge cases which are which are not working well yet. So just selfishly, and it is selfish, in the same way that somebody told me about Waze, because I didn't know about Waze. This was a few years ago. But for instance, are there cool things that you're seeing that are affecting your life, for instance, that some of us might not even be aware of?

One of the first AI applications, and and actually my first company was focused on this problem was, recommendations for things that I might like to see or read. Okay. And, Netflix and others are putting an enormous amount of money into helping me figure out what to see or read or see in their case. As you mentioned earlier in this conversation that there was sort of an overwhelming amount of data. Well, there's now an overwhelming amount of content at my fingertips, and I could look at just what's popular or what's popular amongst my friend group, but I might be missing a gem that would be perfect for me that none of my friends would happen to tell me about.

So I think recommendations are are something that they're around us in a lot of ways. People are recommending products. Amazon's recommending products for you to buy. Advertisers are using your data to try and tell you what you want. And content companies or content distributors are now using and have been using a lot of this kind of intelligence to help you find things you like.

That's a fascinating thing. Like, again, go back to, you've got Audible and books that you're listening to. You've got Netflix. You've got Spotify. You've got all these different things where we're consumers of different entertainment.

And, you know, if you watch a drama with a particular actor, do they recommend dramas or do they recommend shows with that particular actor or actress that you like or directors? As a consumer of this, I don't know how deep it goes. Mhmm. So on some level, it knows. It may start to know what I like more than I know what I like.

I mean, so, again, on Spotify, if you're listening to classical, it may recommend classical. But on the other hand, is it and you're a music person, like, does it go to the level of, you know, is it a fast paced song? Is it an uplifting song, like the rocking music or the Bond music? Where that starts and stops, God only knows. Well, a lot of these systems, these recommendation systems started with something, once again, that my first company worked on, which is collaborative filtering.

So it's not it doesn't need to classify this is classical rock, or this is fast, or this is slow. It might say, people like you, who probably like I bet you have a diverse taste in music. Yeah. Right? Absolutely.

Do you like some jazz? Sure. Do you like some classical? Sure. Do you like some R and B?

Right. If you take all of those data points and say, well, here's all the things that you like and find other people who like those same things, maybe the things that they also like that you don't know about yet Interesting. Would be interesting to you. Right. That's collaborative filtering.

And you see it on Amazon all the time when when they say you might be interested in, then they're probably using a lot of different techniques. But it's not necessarily about trying to label all those things. And that's one of the things that AI is generally good at. It doesn't need to to have a system that works and and gives you a recommendation that's meaningful to you, it doesn't need all that classification. You don't need to decide that something is classical or rock or r and b or jazz.

It can just hear enough music that you like and say, here's another piece of music that might be interesting to you. And there's a lot of techniques to do that, none of which necessitate having some kind of taxonomy of how we, people, label music. So, I mean, let's go to, like, something like OpenTable. OpenTable knows all the restaurants I go to. Mhmm.

Now if it starts recommending particular restaurants, for instance, it presumably would upset some of its clientele that are listing at OpenTable. On the other hand, I would find it very valuable to say, okay, I like a fish house, or I like this particular type of chef, you know, James Beard award winning chef, or whatever whatever it is, or I like a particular style of wine. And then I'm going to Chicago, or I'm in LA and there's these three new restaurants opening up. I mean, are there advertisers or apps that are tracking that information and starting to think about making recommendations in terms of what restaurants you should be going to? And Oh, absolutely.

And there have been for a while. Whether OpenTable or or does it or not is, as you say, it's a business decision. Right. They certainly have the information, the data to do Right. A good recommendation.

But, yeah, there have been hundreds, probably thousands of apps that have made recommendations for places to eat, places to go, movies to watch, all the entertainment and discretionary things that we do. Now, you you had mentioned something about helping people with related to drugs in terms of medical treatments and so forth like that. How is machine is that helping the doctors or actually helping us as patients in terms of do we take an Advo or do we take a In that case, I was talking about drug Development? Development and discovery. There's lots of techniques that can dramatically up the number of drugs that can be tested at a certain time.

There are companies that are creating models that are computer models of people so you can test drugs on the model before you even take the drug to an animal testing. So we're invested in a in a company that does that called Verisint Life. And throughout that process, there is a lot of sophisticated AI that's going into helping create more candidates for curing diseases. Interesting. You know, America has led in a lot of areas.

And we do live in a competitive world. And whether or not it was the race to the moon or the South China Seas, there's a lot of talk about this new cold world with China. And there's a lot of technology coming out of China. When you look at the machine learning that's going on in America versus the machine learning that's going on in places like China, how are we doing relative to places like China? And and what's, if anything, is coming out of China that's unique as opposed to what's going on here?

So by all accounts, the Chinese are investing a ton of money into AI, Chinese companies, Chinese government, and there are great researchers in China. There are great technologists. I think it's wishful thinking or kind of fairyland thinking to to imagine that they're not going to be able to develop software and and AI, at least as well as we can. One thing that's different is that the idea of privacy and what you can do with data is very different in China than it is here. They have a lot more latitude, or at least the government does, to use data however they want.

So my understanding is that there's pretty pervasive surveillance Right. With face recognition throughout the country in China. And that's something that would have a lot of political pushback here that they don't have. So I'm not, an expert in what's going on exactly in various companies in China. But from what I know and have heard, they're doing really well.

And they're probably on some fronts doing better than us. When you look at some of the kind of political things that are going on, like targeted ads and and foreign interference with our elections and so forth, is that a lot of machine learning as well, like, in terms of figuring out how to target people with different ads or other information that's gonna raise their blood pressure or get or influence them, for instance? Absolutely. Yeah. Absolutely.

And figuring out where to place those ads, when to place those ads, what's gonna work. Does ad a work better than ad b? There's very sophisticated work behind the scenes to make those interferences happen. I mean, somebody was saying even and and, again, completely keeping this nonpolitical. But when you look at, for instance, swing states or you look at swing districts, that in terms of how you influence an election, that if you have young, smart, technological people really drilling down in these particular close races and in these swing districts, that you really can, in a much more sophisticated way, target ads to those areas in a way that we couldn't even imagine doing ten or twenty years ago.

Yeah. As we said early in this conversation, and I hope you keep it, knowledge is power. Right. And when you can get information on a granular level about every voter in in a certain district or area of an election, figure out what messages work for them, who to contact, who not to bother with, who you might not want to encourage to go to the polls. It's incredibly powerful.

Yeah. And you can do it on a scale with computers that you can't do with however many humans you can scare up to look through that information. This was great. Well, thank you and, for being our guest today. My pleasure.

Please join us next time when we talk with Minnie Ingersoll, a VC with an impressive story. From helping to build Google in its early years to the co founding of Shift, one of the largest online marketing platforms for used cars, Minnie shares her perspective of the advantages for women in computer science and how to tackle the needs of consumers in an ever changing retail world.