00:37:51 · Markie Wagner
[00:37:51] S1: Yeah. So Delphi is a very fun story. I was always working while I was in school, pretty much. From kind of the early days of freshman year, I had gotten to know a lot of founders because I loved startups and IBM companies. And I think that the early days are some of the most interesting fun stories in the world. And so I got to know a lot of founders, seed, series A, series B, series C founders who would really struggle to stand up AI teams for the first time and kind of build any early AI product, often external AI products. So someone come to me and say, hey, we're looking at this $250,000 ARR contract. And to close it, this VP or this buyer really wants recommendations or they really want some kind of like little feature that they requested that somehow involved AI. And they said, frankly, like, I have no idea how to do that. Nobody else on the team does. Could you come and help us figure out how to get that going quickly and also just to make it a competency in house? Because I have been doing AI stuff kind of throughout high school. I started doing that early freshman year and a lot of those pilots and early MEP's went well and ended up providing a lot of value to the customers, to the folks that I would work with. And yeah, I mean, really appreciate a lot of the founders took a bet on me even though I was disproportionately young. They let me go deep into the systems and stand up stuff in prod all the time. So I started doing that at an increasingly high cadence with increasingly large companies. And by the time that I had dropped out of Stanford, I was helping public companies, folks that were about to go public as they set up AI teams for the first time and built out their whole AI orgs, which is really fun. I had been, even the Google and my work bled into like time when I was in school. So there was never really school time and then time for building within the context of companies within Delphi. It was very much all happening at the same time. And so Delphi kind of exploded during the pandemic and during the time where Stanford went online. And so I knew that these were opportunities and products that wouldn't wait for me to go and finish up school. And frankly, I had blasted through a lot of the AI classes that I wanted to take at Stanford. And so it just made sense to make the leap. There's a lot of people working on Delphi too.
00:40:04 · Markie Wagner, Nikhyl Singhal
Expanded. And I went from like fake doing classes to just not doing classes. But my days were still eighteen hours. So it was all the same anyways. And it was just very intense and really special. I think some of my favorite times of my life have been working with the team at Delphi on their very intense sprints. But that was the natural thing to do. It was kind of not really a question when it happened. [00:40:27] S0: Yeah, I'm curious, because AI is the hottest of all hot words that we ever see today. I think everyone, every company, or even my dad asked me like, hey, how do I get into this AI thing? He's in his 80s. And so I remember that happening when web programming was big, mobile programming was big. So we're clearly seeing another wave. You are maybe one of the most serious and seasoned executives from AI, and you've been with it from the beginning of this sort of current wave. And I'm curious just sort of maybe an open question around what's your current take? There's always a hype that may be bigger than the reality, but I'm curious as to... Is AI as transformative as we talk about in the media, and are you as excited about what the next ten years looks like for changes in transforming society with AI?
00:41:24 · Markie Wagner, Nikhyl Singhal
[00:41:24] S1: I am. And so I've obviously been around in AI since even during eras where not much happened. But I do think that large language models in the advent of these large language models and these capabilities that come with having scaled these foundation models to massive scales, it's one of the most important things that has happened in tech since the iPhone, I think. I think it's actually important. I think that while there's a lot of chatter about it and ChatGPT obviously made a massive splash, that doesn't mean, oh, there's so much hype. There must be nothing there. It's like, there's a lot of hype and there is something there. And people might dream beyond its capabilities, but something very important has happened before these language models and after in terms of the capabilities of what people can do. [00:42:08] S0: I'd be curious as to, from your honest opinion, from being in the trenches, what are things that we may think is magic today, like how, say, maps worked back in the day where you had a book that you would look up for an address, and now how everyone assumes you can get there, you don't even give directions. What's like magic today but assuming as standard routine tomorrow? I'm curious as to where is AI going to impact our world and our society?
00:42:42 · Markie Wagner, Nikhyl Singhal
[00:42:42] S1: Yeah. When I think about what my kids will be astonished that I do today, which is the equivalent of my mom showing me the maps that she used to keep in her car back in the 90s or something like this. I think a lot of that will come in reflecting on how people spend their time, like their working hours today. And in the future, very large proportions of the work that even executives, ICs, managers, kind of everywhere up and down the pyramid spend their time on will be abstracted away and almost delegated to AI in a lot of ways. And so things that are routine, say I do this workflow once, I think, you know, we could say, were you actually doing on your computer on the days of Delphi? Say, okay, was copying code from here to there, or I would do outbound and I would go all over LinkedIn and spam these things. And those are what I was spending my hours on. I think they'll say, wait, what? Like you didn't have someone do that for you? It's like, well, no, like it costs a lot of money to hire someone to do something for me. And they'll say, oh, because they would just imagine that I would be kind of delegating these processes and workflows to software. And I think it'll be a crazy thing to kids that people were doing large sections of the workflows that they do today in fifteen years. [00:43:58] S0: Yeah, I almost had this vision of every human will have an AI companion who essentially will do their bidding on anything that's routine. [00:44:09] S1: AI Jarvis, right?
00:44:09 · Markie Wagner, Nikhyl Singhal
[00:44:09] S1: AI Jarvis, right? [00:44:10] S0: Yeah, AI Jarvis and that will allow us to sort of shift our focus from all of these things that we were burning energy into something that I think using your language could be potentially very powerful and very good quest like. Now I think being a career podcast, the natural question is, how does this then impact those that are starting their career or in mid career? That statement can be empowering, and that statement can be quite scary. And I'm curious as to what's your career guidance for those that are in mid career and they're worried about what AI might do for them. You know, there are certainly things that can help, but then there's people that might be fearful of what this could mean career wise. What's your take on that?
00:44:57 · Markie Wagner
[00:44:57] S1: My guess is that we will become orchestrators of new forms of capital and leverage that are very AI driven. So we might have AI specialists. We might have AI interns. Know, whatever you want to call it. It's kind of behind the scenes AI workflow fabric that can use applications and take actions on our behalf that we direct. And so you become an orchestrator. And of course, you're also doing maybe some IC work, but that will probably trend away over time. Rather than writing code to do code reviews. Or rather than sending the outbound, you review the outbound, but respond back or something like this to that effect. And so I think in a lot of ways you effectively end up moving up the management letter, but you're managing AI resources in a lot of way. And so I think more so than ever before, getting the experience and doing something yourself before you delegate, it's very important. So obviously doing the IC work today, but also doing so much IC work that you then end up scaling through the ladder and end up managing, I think human resources could be helpful, ended up understanding how to communicate things that you want done and kind of learn how to be an orchestrator of human capital today. We'll probably have lots of transferable learnings to managing AI capital in the future. And so I think that might be an important story. But you can't just say, oh, I want to be a manager, so I'm going to start managing before I even know how to do the work. I don't think a shortcut is the answer here either. Learn how to do the work, Learn how to do all the types of work in your part of the org before you then transition to that next layer.