WEBVTT - Garry Tan Decides Who Gets to Build the Future 0:00:09.760 --> 0:00:14.100 The Builders is a production of Kaleidoscope and iHeart, presented 0:00:14.260 --> 0:00:18.860 by Ramp. Ramp is a financial operations platform that helps 0:00:18.960 --> 0:00:23.120 ambitious companies manage their finances so their teams can spend 0:00:23.180 --> 0:00:26.980 less time chasing receipts and approvals and more time bringing 0:00:27.120 --> 0:00:32.919 extraordinary ideas to life. I'm Walter Isaacson. Welcome to The Builders, 0:00:33.260 --> 0:00:37.699 my series of conversations with the most consequential minds shaping 0:00:37.780 --> 0:00:45.080 our future. Today, I'm speaking with Gary Tan. Y Combinator 0:00:45.540 --> 0:00:49.520 has become one of the most powerful and important brands 0:00:49.850 --> 0:00:58.490 in Silicon Valley. The companies that emerged from YC include Airbnb, Stripe, DoorDash, 0:00:58.910 --> 0:01:05.230 and Dropbox. Together, they have transformed how we travel, spend money, work, 0:01:05.470 --> 0:01:11.140 and communicate. Now Gary Tan is deciding which builders come next. 0:01:11.959 --> 0:01:16.010 Gary first entered Y Combinator as a founder. He became 0:01:16.050 --> 0:01:20.890 an early investor in Coinbase and Instacart. And today, he 0:01:21.030 --> 0:01:26.840 leads YC itself. And so, in some ways, he's helping 0:01:27.020 --> 0:01:31.780 shape those who shape the future. Here's my interview with 0:01:31.840 --> 0:01:42.840 Gary Tan. Gary Tan, welcome to The Builders. 0:01:43.680 --> 0:01:44.300 Thanks for having me. 0:01:44.959 --> 0:01:48.680 As a leader of Y Combinator, you have to pick 0:01:49.060 --> 0:01:52.280 founders who are going to make it. Do you look 0:01:52.300 --> 0:01:55.430 for one particular thing or are there like 20 different 0:01:55.470 --> 0:01:57.350 models of founders that can work? 0:01:58.230 --> 0:02:00.790 Well, so the wild thing about Y Combinator is that 0:02:00.930 --> 0:02:05.560 we got 80, maybe 90,000 applications a year now. for 0:02:05.780 --> 0:02:08.889 about 800 spots. So it is less than a 1% 0:02:08.889 --> 0:02:13.029 acceptance rate. Maybe 8 to 10% of those people would 0:02:13.070 --> 0:02:16.950 get an interview, sometimes less now. And the wild thing 0:02:17.010 --> 0:02:19.390 is we actually spend only 10 minutes with them for 0:02:19.430 --> 0:02:23.020 that final stage. But the interesting thing is, Paul Graham, 0:02:23.060 --> 0:02:25.820 the founder of YC, figured out that you didn't need 0:02:25.860 --> 0:02:26.400 half an hour. 0:02:26.580 --> 0:02:27.640 You don't need an hour. 0:02:28.180 --> 0:02:31.220 You could probably actually do it in about five minutes, 0:02:31.340 --> 0:02:34.910 but five minutes feels rude. So that's how YC ended 0:02:34.950 --> 0:02:35.690 up at 10 minutes. 0:02:35.970 --> 0:02:38.370 So what did Paul look for and what do you 0:02:38.450 --> 0:02:40.329 look for in those 10 minutes? 0:02:40.990 --> 0:02:44.250 I mean, it sounds almost too simple, but 10 minutes 0:02:44.310 --> 0:02:48.480 is a wonderful constraint. Kind of like how 140 characters 0:02:48.540 --> 0:02:51.020 for Twitter way back in the day was a wonderful constraint. 0:02:51.480 --> 0:02:55.240 If you can say something very simply that hits us, like, 0:02:55.260 --> 0:02:58.169 you know, here's a user, here's a real person who, 0:02:59.690 --> 0:03:02.290 has a problem, and here's how the technology that we 0:03:02.310 --> 0:03:04.910 build actually solves that. We just want to know what 0:03:04.950 --> 0:03:07.630 it is and why it's awesome. And then if we 0:03:07.669 --> 0:03:09.750 have a little bit of extra time, we want to 0:03:09.770 --> 0:03:12.639 know why that founder is the person to go and 0:03:12.880 --> 0:03:15.120 build that thing. And you can do all of that 0:03:15.180 --> 0:03:18.380 in 10 minutes, and it's an up-down decision. Something like 10% 0:03:18.380 --> 0:03:22.520 to 15% of all self-made billionaires as of present time, 0:03:22.540 --> 0:03:26.180 age 50 and under, are YC alums. 0:03:26.880 --> 0:03:29.130 Let me ask you a real quick question then. Suppose 0:03:29.250 --> 0:03:32.310 Elon Musk walked in and you had 10 minutes with him. 0:03:32.669 --> 0:03:35.750 What do you think the percentage chance is that you 0:03:35.790 --> 0:03:36.990 would actually fund him? 0:03:38.010 --> 0:03:41.500 Oh, 100%. I mean, in current form, 100%. No, no, no, no. 0:03:41.570 --> 0:03:44.660 I'm saying you saw him. You didn't know he was successful. 0:03:45.060 --> 0:03:48.660 He was a guy who'd just been, say, ousted from PayPal. 0:03:49.040 --> 0:03:51.120 I mean, I had a managing partner meeting the other day, 0:03:51.240 --> 0:03:53.940 and that's actually one of the example questions we ask 0:03:54.000 --> 0:03:56.780 ourselves to figure out whether we're on the right track, right? 0:03:57.260 --> 0:04:01.370 Like by definition, if YC is operating properly, like if 0:04:01.450 --> 0:04:05.070 an Elon or a Bill Gates or a Steve Jobs or, 0:04:05.290 --> 0:04:07.990 you know, a Drew Houston from Dropbox, a Brian Chesky 0:04:08.010 --> 0:04:12.170 from Airbnb, Brian Armstrong from Coinbase, Tony Hsu from DoorDash. 0:04:12.250 --> 0:04:16.030 Like if someone like that walks into our office, can 0:04:16.060 --> 0:04:18.200 we say yes to them in 10 minutes? And if 0:04:18.240 --> 0:04:21.479 that answer is no, actually, I'd have to rewrite everything. 0:04:21.620 --> 0:04:24.210 I would have to retool everything, right? That is actually 0:04:24.390 --> 0:04:26.580 almost the only question that really matters when it comes 0:04:26.620 --> 0:04:28.200 to whether YC is doing well or not. 0:04:28.500 --> 0:04:31.820 And if you had walked in years ago, would you 0:04:31.860 --> 0:04:32.700 have accepted you? 0:04:32.720 --> 0:04:35.860 I think so. 0:04:36.700 --> 0:04:39.580 My startup, Posterous, it was Dead Simple Blogs by email. 0:04:39.900 --> 0:04:42.390 And I mean, I had a pretty banger demo. So 0:04:43.730 --> 0:04:46.810 we had something that I think Paul and Jessica and 0:04:46.870 --> 0:04:50.589 Trevor Blackwell, the founders of YC did. really got immediately. 0:04:50.770 --> 0:04:52.669 They didn't grill us at all. Like we came on 0:04:52.710 --> 0:04:55.409 with a really good demo and they said, well, that 0:04:55.450 --> 0:04:58.190 makes sense. And then we got funded. So sometimes it's 0:04:58.210 --> 0:04:58.820 that simple though. 0:04:59.240 --> 0:05:03.279 And I was creating that company, running that company for 0:05:03.300 --> 0:05:08.419 a while. How fundamental was that to your ability to 0:05:08.500 --> 0:05:13.960 be at Y Combinator and to be making venture capital decisions? 0:05:15.190 --> 0:05:17.810 Oh, I think it's quite important. I mean, There's a 0:05:17.850 --> 0:05:20.409 Jay-Z line that I love. One of his songs, he goes, 0:05:20.810 --> 0:05:22.490 everybody wants to tell you how to do it, and 0:05:22.510 --> 0:05:26.190 they never did it. And so, I mean, the entire 0:05:26.310 --> 0:05:29.810 act of being an investor in technology is actually bewildering 0:05:29.850 --> 0:05:32.229 and strange if you really think about it too long, 0:05:32.310 --> 0:05:36.650 because what we're doing is selling money. Like, tech has 0:05:36.690 --> 0:05:40.610 now become such a fundamental force in the financial world 0:05:40.870 --> 0:05:44.779 of society that that there's way more money that wants 0:05:44.820 --> 0:05:47.480 to go into tech than good opportunities. 0:05:48.120 --> 0:05:49.979 And that's really what YC is. 0:05:50.020 --> 0:05:52.790 And if you look at most venture capital, it's crazy 0:05:52.820 --> 0:05:55.130 that it's grown so much. The S & P 500 0:05:55.130 --> 0:05:58.469 is almost entirely driven by tech. And then the wild 0:05:58.529 --> 0:06:01.810 thing that Paul Graham figured out and Jessica Livingston figured 0:06:01.850 --> 0:06:04.570 out when they first started was, when they started in 2005, 0:06:04.570 --> 0:06:08.020 the type of people who would get funding, it would 0:06:08.060 --> 0:06:11.540 be your buddy at Stanford Business School, your golf buddy, 0:06:11.720 --> 0:06:15.179 like maybe someone who your son went to school with 0:06:15.360 --> 0:06:18.060 at like some prep school at Harvard Westlake or what, 0:06:18.080 --> 0:06:20.870 you know, I mean, it was like access, right? 0:06:21.170 --> 0:06:23.589 As it was for a few millennia. 0:06:23.730 --> 0:06:25.050 Time immemorial, right? 0:06:25.290 --> 0:06:26.970 But do you all think you broke the back of 0:06:27.029 --> 0:06:30.229 that or is being a startup junkie or a startup 0:06:30.310 --> 0:06:33.270 entrepreneur something that you still have to be pretty privileged 0:06:33.310 --> 0:06:34.650 and pretty connected to do? 0:06:34.670 --> 0:06:36.950 I mean, it would be a mistake to say that 0:06:37.000 --> 0:06:40.390 we were actually a meritocracy, but I will say and 0:06:40.570 --> 0:06:43.610 affirm loudly that we aspire to be. 0:06:44.010 --> 0:06:46.270 We will never be. You know, I think. 0:06:46.130 --> 0:06:49.050 That the woke age of sort of like five or 0:06:49.050 --> 0:06:51.270 10 years ago, we sort of lost sight of that, 0:06:51.490 --> 0:06:55.070 that like meritocracy could never be achieved, but it is 0:06:55.120 --> 0:06:57.000 a virtue to attempt to achieve it. 0:06:57.680 --> 0:07:00.180 Well, the reason I ask about that in terms of 0:07:00.220 --> 0:07:03.740 a startup culture, which is what you very much nurture, 0:07:04.320 --> 0:07:07.860 is that it used to be somebody coming out of college, 0:07:07.940 --> 0:07:11.060 take a Steve case, for example, could go work at 0:07:11.100 --> 0:07:16.760 a company like a Procter & Gamble and learn a business. Nowadays, 0:07:16.840 --> 0:07:20.180 it's very hard to sort of get a steady job. 0:07:20.560 --> 0:07:23.610 And the economy for young people tends to depend more 0:07:23.650 --> 0:07:27.050 and more on startups. And yet it seems to me 0:07:27.870 --> 0:07:31.170 that unless you got a safety net, unless you're somebody 0:07:31.190 --> 0:07:34.330 who doesn't have to make sure your next meal can 0:07:34.370 --> 0:07:37.760 be paid for, you can't be in that game, and 0:07:37.800 --> 0:07:39.360 it leaves a lot of people out. 0:07:39.940 --> 0:07:43.830 I mean, that's why YC gives $ 500, 000, for instance. When 0:07:43.870 --> 0:07:46.870 I did YC, it was only $ 18, 000. Like today, you 0:07:46.910 --> 0:07:50.510 hear about seed funds and Stanford kids raising $ 30 million 0:07:50.570 --> 0:07:54.310 on $ 300 million post-money evaluation. It's like literally, like it's 0:07:54.350 --> 0:07:57.800 just an idea and one person, and the company is 0:07:57.830 --> 0:08:01.140 somehow worth $ 300 million. But when we were starting many 0:08:01.180 --> 0:08:03.470 years ago, It was a very small amount of money. 0:08:03.650 --> 0:08:05.970 You had to show that you had something that was great. 0:08:06.690 --> 0:08:09.670 And so that's why YC didn't stay. 0:08:09.870 --> 0:08:15.990 And we easily could have just stayed at $ 12, 000. 0:08:13.470 --> 0:08:15.270 Or $ 15, 000. 0:08:14.270 --> 0:08:16.710 But I think there is a recognition that if we 0:08:16.750 --> 0:08:19.470 want to be a thing that allows people to participate, 0:08:20.170 --> 0:08:22.750 you don't have to have rich parents. The half a 0:08:22.770 --> 0:08:25.770 million dollars getting into YC, you could just have an 0:08:25.830 --> 0:08:28.620 idea and you could come in and meet us for 0:08:28.620 --> 0:08:29.180 10 minutes. 0:08:30.540 --> 0:08:33.559 You were one of the earliest employees at Palantir. You 0:08:33.600 --> 0:08:36.820 got to see how Peter Thiel and Alex Karp in 0:08:36.880 --> 0:08:41.000 particular really worked. What did you learn from their style? 0:08:41.780 --> 0:08:44.240 I realize now, like, if I hadn't rage quit, I 0:08:44.260 --> 0:08:46.940 would have been, you know, sort of a peer to 0:08:47.000 --> 0:08:50.100 some of the, you know, real legends who really built Palantir. 0:08:50.160 --> 0:08:51.880 You know, I would go on walks with Karp and 0:08:52.520 --> 0:08:54.660 he would sort of, I mean, he was a philosopher. 0:08:54.740 --> 0:08:56.900 You know, he was not a technologist, I would say, 0:08:57.120 --> 0:09:00.929 but he was incredibly knowledgeable about the world. One time, 0:09:00.950 --> 0:09:05.630 I was sent to DC to do some demos for customers. 0:09:06.290 --> 0:09:09.470 And I remember being like a 25-year-old, and I was 0:09:09.530 --> 0:09:13.750 in my ripped jeans and band t-shirt and getting sent 0:09:13.790 --> 0:09:18.790 to stay at the Mandarin Oriental in DC. And we 0:09:18.809 --> 0:09:23.030 would wonder, Alex, why are we staying at the Mandarin Oriental? 0:09:23.070 --> 0:09:23.470 Look at us. 0:09:23.510 --> 0:09:26.500 We're just kids. We don't have to do that. Just 0:09:26.530 --> 0:09:27.750 set me up in the Motel 6. 0:09:27.750 --> 0:09:30.760 We're a startup. Alex looked at. 0:09:30.740 --> 0:09:33.900 Me and my other early engineers for the Palantir Finance 0:09:33.940 --> 0:09:37.800 product and he said, guys, you're going to go and 0:09:37.840 --> 0:09:41.020 talk to such and such general or such and such 0:09:41.280 --> 0:09:45.030 VP of some Fortune 500 and you're going to show 0:09:45.090 --> 0:09:48.189 this incredible software that you've built. And then as you 0:09:48.230 --> 0:09:50.270 go out, you're going to make small talk and they're 0:09:50.290 --> 0:09:52.070 going to ask, well, how long are you around and 0:09:52.090 --> 0:09:54.930 where are you staying? And if you say Motel 6, 0:09:54.929 --> 0:09:57.250 they're going to think that they're buying Motel 6 software 0:09:57.770 --> 0:10:00.770 and Palantir is Mandarin-oriented software. 0:10:01.570 --> 0:10:05.500 I'm like, OK, now I get it. And so, you know, yeah, 0:10:05.580 --> 0:10:05.860 I don't know. 0:10:06.050 --> 0:10:07.880 I think embedded in that story is just a lot 0:10:07.960 --> 0:10:10.380 of different things. And I learned a lot from Alex 0:10:10.420 --> 0:10:10.740 from that. 0:10:11.420 --> 0:10:14.740 Don't go anywhere. There's a lot more right after this break. 0:10:23.710 --> 0:10:28.110 I know you've been fighting this antitrust battle. How important 0:10:28.170 --> 0:10:31.700 is that in order to keep innovation and startup culture? 0:10:32.380 --> 0:10:35.500 I think true free market enterprise has never been tried. 0:10:36.440 --> 0:10:38.970 I think that in tech in particular, we have very 0:10:39.020 --> 0:10:41.989 short memories. I would argue that Google itself, one of 0:10:42.010 --> 0:10:45.309 the big tech companies, was able to exist and thrive 0:10:45.370 --> 0:10:48.530 and become a trillion-dollar company because of the actions of 0:10:48.570 --> 0:10:52.030 the DOJ with the consent decree on Microsoft. Like, Microsoft 0:10:52.050 --> 0:10:55.110 was ready to take over the web, and it used 0:10:55.200 --> 0:10:59.640 its platform advantages in Windows to default install and take 0:10:59.760 --> 0:11:01.320 over the browser market. 0:11:01.600 --> 0:11:04.040 And do you think Google is now doing the same thing? 0:11:04.929 --> 0:11:05.780 I would argue yes. 0:11:06.120 --> 0:11:09.360 The Delaware C Corp sort of actually insists on it. 0:11:09.500 --> 0:11:13.559 Like the governance structure itself actually insists on you're supposed 0:11:13.600 --> 0:11:16.000 to do your duty and your duty is to maximize 0:11:16.020 --> 0:11:19.630 shareholder value. So people are following the letter of the 0:11:19.690 --> 0:11:22.850 law and the spirit of like their Delaware C Corp 0:11:22.870 --> 0:11:25.830 to close platforms. Like it's less and less percentage of 0:11:25.890 --> 0:11:28.069 the screen real estate if you do a Google search 0:11:28.630 --> 0:11:30.910 is even non-Google properties, right? 0:11:31.590 --> 0:11:32.770 And so that's self-preferencing. 0:11:32.790 --> 0:11:32.929 So.... 0:11:34.390 --> 0:11:37.050 A lot of your tech bro friends are very much 0:11:37.150 --> 0:11:41.590 against antitrust enforcement. How do you have a startup culture 0:11:42.010 --> 0:11:44.770 if you're going to allow big companies to self-preference? 0:11:45.410 --> 0:11:48.950 Well, I mean, my view is that we shouldn't allow self-preferencing. 0:11:49.190 --> 0:11:52.309 And then my super pro startup stance is we should 0:11:52.370 --> 0:11:56.319 stop self-preferencing. We should require open markets. Like I think 0:11:56.360 --> 0:11:59.920 the DMA in Europe, for instance, is a great example 0:11:59.980 --> 0:12:03.740 of forcing big tech to open its own platforms. And, 0:12:03.840 --> 0:12:06.300 you know, not just data transparency and privacy, but also 0:12:06.340 --> 0:12:09.699 portability and reverse engineering. Like these are sort of things 0:12:09.780 --> 0:12:13.880 that allow for open markets and open access to startups. 0:12:14.280 --> 0:12:16.720 And if you ask me, is big tech too powerful? 0:12:16.760 --> 0:12:19.900 Like I say, absolutely. And the answer and the antidote 0:12:20.020 --> 0:12:23.660 is little tech. I think it's not that unlike robber 0:12:23.700 --> 0:12:26.540 baron times in that, you know, if you own the 0:12:26.600 --> 0:12:29.350 railroads and there was a monopoly on the railroad and 0:12:29.750 --> 0:12:33.150 you could set the price on literally everything else. And 0:12:33.230 --> 0:12:36.210 so one thing would lead to another thing, and you 0:12:36.250 --> 0:12:40.730 would have multiple monopolies across both commodities and the railroad 0:12:40.770 --> 0:12:42.050 the commodities travel over. 0:12:42.309 --> 0:12:44.410 Well, wait, let's look at AI. I'm going to push 0:12:44.450 --> 0:12:48.900 back a bit. We get a whole lot of new companies, Anthropic, OpenAI, 0:12:48.940 --> 0:12:53.020 they're all popping up. Somehow Apple wasn't able to self-preference 0:12:53.440 --> 0:12:56.740 and embed its own system in its own phones. Even 0:12:56.820 --> 0:12:59.540 Google is trying to do that where it almost forces 0:12:59.600 --> 0:13:03.620 you to use Gemini when you're using Google search. And 0:13:03.679 --> 0:13:05.700 yet these other companies are doing well. 0:13:06.520 --> 0:13:07.179 Yeah, that's right. 0:13:07.280 --> 0:13:10.870 I mean, I think what's funny about the AI companies 0:13:10.900 --> 0:13:12.970 is like they're almost certainly, I mean, I think you 0:13:13.010 --> 0:13:15.270 can fairly say that they are the new big tech. 0:13:16.170 --> 0:13:19.950 And so I think I have concerns there as well, right? 0:13:20.610 --> 0:13:23.270 Critics of some of the AI companies would say that 0:13:23.790 --> 0:13:27.140 It's possible that as the AI wars progress, there might 0:13:27.200 --> 0:13:31.160 not be open access to the latest and best frontier 0:13:31.200 --> 0:13:33.800 models that those companies may gatekeep. 0:13:34.260 --> 0:13:37.660 You think they should be open, that all the AI 0:13:37.730 --> 0:13:38.679 models should be open? 0:13:39.309 --> 0:13:41.810 I mean, that would be certainly the thing that supports 0:13:41.850 --> 0:13:44.790 startups and the ability for new people to create new 0:13:44.870 --> 0:13:48.370 companies and, frankly, new wealth and to give consumers choice. 0:13:48.530 --> 0:13:51.160 So what is the role that Y Combinator wants to 0:13:51.240 --> 0:13:52.070 play in AI? 0:13:53.809 --> 0:13:56.940 What's funny is I think of YC almost as like 0:13:57.760 --> 0:14:01.460 we're not a union, but we have union-like characteristics. There's 0:14:01.520 --> 0:14:05.120 not really a group out there that can speak full-throatedly 0:14:05.480 --> 0:14:11.630 about policies and, frankly, actions that both governments and companies 0:14:11.690 --> 0:14:14.750 should take to make sure that the golden goose can 0:14:14.790 --> 0:14:19.810 still exist. Whether that's the billionaire tax in California or 0:14:19.950 --> 0:14:23.750 antitrust laws in D.C. or in Europe or we're going 0:14:23.790 --> 0:14:26.710 to start talking about it and just having a seat 0:14:26.730 --> 0:14:28.890 at the table. One of the things I discovered as 0:14:29.110 --> 0:14:33.450 I started examining politics, most tech people are not interested 0:14:33.510 --> 0:14:36.710 in politics, but politics sure is interested in us. And 0:14:36.770 --> 0:14:39.990 so I decided like three or four years ago, like, oh, well, 0:14:40.050 --> 0:14:43.100 I should probably spend some time and reciprocate. Like, let 0:14:43.130 --> 0:14:45.700 me understand what's going on. And what I discovered was 0:14:45.740 --> 0:14:48.859 pretty wild. I mean, millions of dollars in AstroTurf money, 0:14:49.570 --> 0:14:53.850 going to groups that would take around basically almost hired actors. 0:14:54.190 --> 0:14:57.150 They would claim to speak for small software companies or, 0:14:57.170 --> 0:15:02.030 you know, little tech. And they were just like paid chills, actually. 0:15:02.450 --> 0:15:04.690 And they would go to the halls of power, whether 0:15:04.710 --> 0:15:08.890 it's Sacramento or D.C. And they would say, no, no, 0:15:08.990 --> 0:15:12.030 like this thing that actually mainly helps big tech. We, 0:15:12.070 --> 0:15:14.490 the little guy, really wants that. And I was like, oh, wow, 0:15:14.510 --> 0:15:16.910 this is crazy. Like no real startups actually go out 0:15:16.950 --> 0:15:20.170 there and talk to real people in D.C. And I 0:15:20.230 --> 0:15:23.210 saw a system that was broken, and I wanted to 0:15:23.250 --> 0:15:25.970 fix it. And so that, you know, we do very 0:15:26.010 --> 0:15:28.940 little things. Like, we have a tiny budget. It's, you know, 0:15:29.380 --> 0:15:32.720 probably one one-hundredth that of, like, even any given single 0:15:32.760 --> 0:15:36.200 big tech firm spends on policy. We have one lone 0:15:36.240 --> 0:15:40.080 full-time hire in D.C. in a man named Luther Lowe. 0:15:40.880 --> 0:15:42.870 And we run it on a shoestring. But on the 0:15:42.890 --> 0:15:46.610 other hand, like, we're people-powered. I see that like a 0:15:46.670 --> 0:15:50.890 repeating pattern, like there are broken things in society. There 0:15:50.910 --> 0:15:53.970 are systems out there that, you know, have money, like 0:15:54.030 --> 0:15:56.000 just trace the money and like, you know, where's the 0:15:56.040 --> 0:15:57.900 money coming from and who benefits, right? 0:15:58.460 --> 0:15:59.600 And then the heartening thing. 0:16:00.060 --> 0:16:01.480 You know, I think a lot of people talk about 0:16:01.520 --> 0:16:03.920 social media in a very negative way. They think, you know, 0:16:03.980 --> 0:16:06.660 it was Facebook that ruined democracy. And it's like, I 0:16:06.710 --> 0:16:09.109 don't really, I don't think I agree. Like, I think 0:16:09.170 --> 0:16:12.350 that It's the little guy not realizing he has a voice. 0:16:12.450 --> 0:16:14.870 He and she do have a voice. And then, you know, 0:16:15.710 --> 0:16:18.750 when I look at YC, like, our job one is 0:16:18.790 --> 0:16:21.930 to find the next great founders. But I think right 0:16:21.970 --> 0:16:24.070 under that, it's like, how do we make sure that 0:16:24.090 --> 0:16:26.790 those people can have access to markets? How do we 0:16:27.270 --> 0:16:29.740 help people who would not have a voice, would not 0:16:29.770 --> 0:16:32.000 be able to go to D.C.? How do we give 0:16:32.040 --> 0:16:32.940 them that megaphone? 0:16:33.100 --> 0:16:34.960 And then YC itself now has a megaphone. 0:16:34.980 --> 0:16:37.680 You know, we have more than 2 million subscribers on YouTube. 0:16:37.920 --> 0:16:41.480 We can sort of speak in aggregates. And I'm really 0:16:41.520 --> 0:16:45.120 excited about that because I think that that's us trying 0:16:45.180 --> 0:16:47.380 to speak truth to power. I don't want to deny 0:16:47.420 --> 0:16:49.860 that YC is also a power in and of itself, 0:16:50.520 --> 0:16:53.250 but the world is full of systems. These are all 0:16:53.290 --> 0:16:56.590 systems that were created and constructed by, as Steve said it, 0:16:56.790 --> 0:17:00.770 people no smarter or better than you or I. And 0:17:00.970 --> 0:17:02.830 we can mold it and change it. And once you 0:17:02.870 --> 0:17:04.600 realize that, you'll never be the same. 0:17:05.440 --> 0:17:09.590 You were born in Canada, Chinese background. Tell me how 0:17:09.690 --> 0:17:13.910 important you think immigration is to the United States. 0:17:14.890 --> 0:17:18.359 I mean, I shout to everyone who will hear that 0:17:18.560 --> 0:17:21.879 about 50% of all the startups that really make it 0:17:22.060 --> 0:17:25.880 have at least one immigrant co-founder. And what we realize 0:17:25.940 --> 0:17:28.780 is that's one of the things that makes America incredibly unique. 0:17:29.320 --> 0:17:32.160 People from around the world want to come here. And 0:17:32.260 --> 0:17:33.980 in tech, they want to come and build the future. 0:17:34.520 --> 0:17:39.410 That's certainly my story. My mom's family came over from Burma. refugees, 0:17:40.210 --> 0:17:43.330 and my dad came here from Singapore. 0:17:43.550 --> 0:17:46.450 Well, you talk about your dad, and we biographers tend 0:17:46.490 --> 0:17:50.879 to believe it's sometimes about dad, especially when somebody's successful. 0:17:51.100 --> 0:17:53.040 You had a pretty rough time, though. I mean, your 0:17:53.080 --> 0:17:57.560 father was at times unemployed, at times an alcoholic. Tell 0:17:57.600 --> 0:17:59.540 me how you had to deal with all that. 0:18:00.480 --> 0:18:02.240 I mean, it really made me who I am today. 0:18:02.260 --> 0:18:05.350 I figured out that I can actually put myself back 0:18:05.570 --> 0:18:11.900 into let's say 1992, I'm 11 years old. And basically, 0:18:11.920 --> 0:18:13.780 I don't know when my dad would come home, like 0:18:13.820 --> 0:18:16.739 maybe around five or six, but I wouldn't know how 0:18:16.760 --> 0:18:20.000 many beers he would have already had. And I realized 0:18:20.020 --> 0:18:23.219 there's sort of this scanner that happens where if you 0:18:23.240 --> 0:18:26.690 don't know what father is coming home that day, then 0:18:27.450 --> 0:18:32.149 you get very attuned to other people and their interstates. 0:18:32.310 --> 0:18:34.899 And like, how to sort of react to that. And 0:18:34.920 --> 0:18:37.950 then it's interesting because now I realize like that's a 0:18:37.970 --> 0:18:39.770 scanner that I use all the time. Like I can 0:18:39.810 --> 0:18:42.830 sort of sit down with someone and it doesn't happen 0:18:42.869 --> 0:18:45.830 every time, but it happens relatively easily for me where 0:18:45.850 --> 0:18:48.690 I can sit down and just kind of feel what's 0:18:48.710 --> 0:18:51.169 going on for other people as someone who sort of 0:18:51.230 --> 0:18:55.380 spends time with hundreds of startup founders every single year. 0:18:55.480 --> 0:18:58.300 Like that's actually a pretty powerful superpower. 0:18:59.480 --> 0:19:01.969 Let me talk about what may be the next step 0:19:02.109 --> 0:19:06.710 in AI, which is AI agents. And you've said that 0:19:06.770 --> 0:19:09.810 you don't want to fund companies that look at AI 0:19:09.930 --> 0:19:12.570 agents as a feature. You want them to look at 0:19:12.609 --> 0:19:16.190 it as the core operating system. Explain how you think 0:19:16.210 --> 0:19:20.020 that changes what we think about when we talk about AI. 0:19:20.900 --> 0:19:21.740 Yeah, absolutely. 0:19:21.820 --> 0:19:24.050 I mean, We're going to get a little sci-fi here, 0:19:24.130 --> 0:19:26.570 but this is something that we're starting to see ourselves, 0:19:26.590 --> 0:19:29.330 you know, rebuilding the way we operate at YC. I 0:19:29.609 --> 0:19:32.460 think the number one thing that people get wrong is 0:19:32.580 --> 0:19:34.520 they just think that AI is just chat. 0:19:35.060 --> 0:19:36.900 And then it is so much more than that. 0:19:37.119 --> 0:19:40.120 We now have an internal AI agent, for instance, named 0:19:40.180 --> 0:19:43.670 Straylight that kind of has a personality and then it 0:19:44.050 --> 0:19:47.530 has a company brain. Like it actually has the total 0:19:47.570 --> 0:19:51.920 context of every foundation we've ever backed, anyone who's ever 0:19:52.060 --> 0:19:55.140 emailed us, like all the applications. The way I think 0:19:55.180 --> 0:19:58.520 about it from an interesting tech perspective is like a 0:19:58.580 --> 0:20:01.810 human can only keep track in their heads, like seven 0:20:02.070 --> 0:20:05.030 plus or minus three things. And what an AI agent is, 0:20:05.390 --> 0:20:08.070 is already super intelligent. It can keep, you know, I 0:20:08.090 --> 0:20:10.530 was explaining this to my 11 year old. What an 0:20:10.650 --> 0:20:13.320 AI can do is it can keep a stack of 0:20:13.320 --> 0:20:17.900 1,000 pages, think like three Harry Potter books. It can 0:20:18.160 --> 0:20:21.199 needle in a haystack any page, any fact, anything that 0:20:21.240 --> 0:20:26.850 happened in those three Harry Potter books. It can just know, synthesize, 0:20:27.230 --> 0:20:32.209 work with. The working memory is absolutely gigantic. The thing is, 0:20:32.350 --> 0:20:34.670 in another year, it won't be three Harry Potter books 0:20:34.770 --> 0:20:37.869 in the brain of an AI agent. It'll be five, 0:20:37.910 --> 0:20:40.450 or it'll be 10, or it'll be 20. And then 0:20:40.609 --> 0:20:43.270 the set of things that it can operate on is 0:20:43.410 --> 0:20:43.970 even bigger. 0:20:43.990 --> 0:20:46.489 Well, wait, give me a better example and make this 0:20:46.530 --> 0:20:50.190 more concrete for me, because data retrieval, even from three 0:20:50.230 --> 0:20:53.720 Howard E. Potter books, that's kind of cool, but that's not, 0:20:53.960 --> 0:20:57.780 I mean, that's just data retrieval. What makes something an agent? 0:20:58.400 --> 0:21:01.820 I mean, I would point to what Peter Steinberger of 0:21:02.060 --> 0:21:05.500 OpenClaw figured out, which is if you use Claude Code 0:21:05.680 --> 0:21:09.570 or Codex, I would say right now, They're so worried 0:21:09.609 --> 0:21:13.439 about sort of repeatability and safety that they actually trade 0:21:13.580 --> 0:21:16.820 off all of the beautiful parts of working with an 0:21:16.859 --> 0:21:21.179 AI agent, which like basically can be so similar to 0:21:21.240 --> 0:21:23.950 a human that like it sort of takes on human qualities. 0:21:24.230 --> 0:21:26.390 So tell me what you ask your agent to do. 0:21:27.010 --> 0:21:27.490 Oh, gosh. 0:21:27.730 --> 0:21:30.390 I mean... One of the things I did, it's read 0:21:30.530 --> 0:21:33.560 all of my email and all of my Dropbox. It's 0:21:33.640 --> 0:21:37.480 read all of my Apple Notes, even my therapy notes 0:21:37.520 --> 0:21:40.240 going back 20 years. I mean, for instance, I used 0:21:40.300 --> 0:21:41.980 it for prep to be able to talk to you. 0:21:43.080 --> 0:21:45.040 Not to get too meta about it, but in about 0:21:45.040 --> 0:21:48.540 15 minutes right before this talk, it was able to 0:21:49.440 --> 0:21:53.690 download and transcribe your top 10 podcasts you've ever done. 0:21:54.350 --> 0:21:57.170 And then it actually diarized them. So, you know, it 0:21:57.210 --> 0:22:00.129 had an LLM go like line by line, timestamp by 0:22:00.170 --> 0:22:04.129 timestamp to pull out like major themes. And that was 0:22:04.190 --> 0:22:06.090