WEBVTT 1 00:00:00.031 --> 00:00:01.834 Do you do all your prompting on Ultrafast? 2 00:00:02.274 --> 00:00:02.655 I do. 3 00:00:03.256 --> 00:00:03.856 That's sick. 4 00:00:04.137 --> 00:00:04.557 Yeah. 5 00:00:04.698 --> 00:00:05.459 That's crazy. 6 00:00:05.479 --> 00:00:13.170 You know, seven people's dots can, working together, create something new that kind of existed in all of our heads to some degree, but was like an emergent thing. 7 00:00:26.295 --> 00:00:27.536 Hey, this is Dan Shipper. 8 00:00:27.696 --> 00:00:29.457 You might be wondering what happened to AI and I. 9 00:00:29.697 --> 00:00:33.160 This is the same show, we are just rebranding it to the Every Podcast. 10 00:00:33.300 --> 00:00:35.281 I'm really excited for our first episode with Sam Waltman. 11 00:00:35.921 --> 00:00:36.262 Here it is. 12 00:00:36.742 --> 00:00:37.602 Sam, welcome to the show. 13 00:00:37.963 --> 00:00:38.443 Thanks for having me. 14 00:00:38.603 --> 00:00:39.404 Psyched to have you here. 15 00:00:39.484 --> 00:00:40.865 So we're here at Dev Day. 16 00:00:41.125 --> 00:00:50.571 I think if you had told me four years ago that the coding models were at this level of capability, I would have been like, wow, I don't know if there would be a lot of developers at Dev Day. 17 00:00:50.831 --> 00:00:54.332 but there's more developers than ever, it seems like, and more work to do than ever. 18 00:00:54.412 --> 00:01:06.016 So what has the, like, riding the increasing capabilities of coding models and seeing the impact on engineering taught you about the future of engineering and more broadly, like, the future of work with AI? 19 00:01:06.416 --> 00:01:08.037 I think people just want so much more stuff. 20 00:01:08.418 --> 00:01:09.999 People's expectations are so high. 21 00:01:10.980 --> 00:01:12.061 And I think that's a wonderful thing. 22 00:01:12.161 --> 00:01:16.304 I think this is why we'll all have this incredible future in front of us to explore. 23 00:01:16.845 --> 00:01:20.948 But developers, they've never been able to do more, but there's never been a higher bar either. 24 00:01:21.008 --> 00:01:21.809 And I think that's great. 25 00:01:21.829 --> 00:01:23.210 I think that's like a nice thing about society. 26 00:01:23.750 --> 00:01:25.212 Is that something that surprised you? 27 00:01:25.232 --> 00:01:26.433 Did you expect that to happen? 28 00:01:26.933 --> 00:01:44.345 I think I was never as doomery about the job stuff as other people, but it still has surprised me somewhat on the upside of just how much latent demand existed in the world and also how much people have remained at the center of everything that is happening with AI. 29 00:01:44.885 --> 00:01:49.848 Do you think there's a limit to that, or do you think that translates pretty cleanly to other areas of work? 30 00:01:50.409 --> 00:01:52.210 I think there will be some areas of work 31 00:01:52.489 --> 00:01:56.531 where there's actually not much creativity required and we don't care about the person doing it. 32 00:01:57.211 --> 00:02:00.473 And many of those tasks or jobs will be done by AI. 33 00:02:01.273 --> 00:02:06.335 But I think there's so much other stuff people want in the world, there's gonna be no shortage of new things to go do. 34 00:02:06.355 --> 00:02:09.036 I think that's been one of the lessons of recent time. 35 00:02:09.397 --> 00:02:13.699 Why do you think it's so human-centered even though we talk a lot about automation? 36 00:02:14.499 --> 00:02:18.461 I mean, one observation is we just, we don't care that much about what the machines do. 37 00:02:19.681 --> 00:02:21.182 And in some sense we never have. 38 00:02:22.280 --> 00:02:24.041 And we're just like, we're obsessed with other people. 39 00:02:24.061 --> 00:02:28.562 We want to know about the person behind the art or the person behind the company or the idea or the product. 40 00:02:29.303 --> 00:02:34.644 And I think this is like a wonderful, positive surprise that we probably should have all known before. 41 00:02:35.225 --> 00:02:35.645 I agree. 42 00:02:35.885 --> 00:02:41.367 I've been thinking, if an agent solves a Navier Stokes in a data center, but no one finds out, did it really solve it? 43 00:02:41.547 --> 00:02:41.807 You know? 44 00:02:43.327 --> 00:02:50.830 I think that was a great example where 10,000 agents worked super hard for four days in perfect coordination with each other and did this amazing thing. 45 00:02:51.691 --> 00:02:57.402 And almost none of the public reaction was about that actually the problem being solved. 46 00:02:57.502 --> 00:03:01.971 A lot of talk about the interpretation, a lot of talk about like the people involved in it in different ways. 47 00:03:03.488 --> 00:03:04.629 That was a good thing. 48 00:03:04.789 --> 00:03:04.950 Yeah. 49 00:03:05.590 --> 00:03:06.611 So it's Dev Day. 50 00:03:06.771 --> 00:03:09.593 You launched, by my count, 22 new things today. 51 00:03:10.374 --> 00:03:12.436 Which I've been at all four Dev Days. 52 00:03:12.956 --> 00:03:15.198 And I think last year was like nine. 53 00:03:15.218 --> 00:03:17.420 And the year before that was like four or five. 54 00:03:17.640 --> 00:03:19.482 So it's a lot more than usual. 55 00:03:19.962 --> 00:03:20.743 Just wait for next year. 56 00:03:20.803 --> 00:03:21.103 OK. 57 00:03:22.364 --> 00:03:22.924 I'm tired. 58 00:03:23.125 --> 00:03:23.665 I'm tired. 59 00:03:24.526 --> 00:03:28.489 We cut a bunch of things that we were going to launch today just because you have to fit it in the 50 minutes. 60 00:03:28.509 --> 00:03:31.932 But we were going to launch a lot of things. 61 00:03:32.517 --> 00:03:33.658 try to practice what we preach. 62 00:03:33.918 --> 00:03:40.443 And if we are saying these tools can make people much more productive and build more and that people want more stuff, like, we better build more ourselves. 63 00:03:41.323 --> 00:03:46.587 What are the new kind of bottlenecks and best practices for doing that? 64 00:03:46.607 --> 00:03:56.794 Because I can see for myself, and I assume you guys are seeing this too, like, now that you can build everything, you're like, what should we build and how do we make it all coherent? 65 00:03:56.875 --> 00:03:57.555 How do you think about that? 66 00:03:59.717 --> 00:04:00.277 Moving to... 67 00:04:00.891 --> 00:04:13.601 this world of sort of persistent intelligence, active intelligence, whatever people wanna call it, this idea that you're gonna have these agents that are running in the background 24 seven, connected to your systems, doing stuff proactively for you. 68 00:04:13.981 --> 00:04:17.103 This is clearly a direction that feels important to explore. 69 00:04:17.624 --> 00:04:20.106 I think we're gonna see amazing stuff happen there. 70 00:04:20.506 --> 00:04:29.773 Another thing we really wanted to do is there's just, there are now so many ideas that are capable of being built and what we are missing is the creative spirit 71 00:04:30.729 --> 00:04:32.050 to figure them out and build them. 72 00:04:32.870 --> 00:04:49.342 So we wanted to really open up our ecosystem, let people build apps in ChatGPT in the same way that we get to do them, let people bring their subscription wherever they'd like, have the marketplace, and really try to let people get connected to all the stuff that people will now be able to build. 73 00:04:50.135 --> 00:05:02.885 Yeah, I mean, just looking at all the platform features that you launched and the way that you talk about platforms, it reminds me a lot of the way that Bill Gates used to talk about Windows as a platform, where you create more value than you capture. 74 00:05:03.385 --> 00:05:09.610 What have you learned, you know, you said this in your keynote and I've observed it as well, there's, at DevDay, there's always an app store. 75 00:05:09.650 --> 00:05:11.872 You've been taking swings at app stores for a while. 76 00:05:11.932 --> 00:05:16.615 What have you learned about building a platform and what do you think makes this one different? 77 00:05:17.536 --> 00:05:26.080 I mean, we clearly have not yet figured out the App Store equivalent, but we have figured out how to build a platform that a lot of people are building incredible value on top of. 78 00:05:26.981 --> 00:05:33.504 I think there are really kind of two schools of thought among people building AI model companies. 79 00:05:34.124 --> 00:05:38.646 There are people that believe or hope that there's only going to be one company in the world. 80 00:05:39.206 --> 00:05:46.670 And then there are people like us that believe we are going to see a renaissance in the number of companies and just an absolute... 81 00:05:47.661 --> 00:05:55.425 lowering of the bar to people, one-person companies, small team companies, creating incredible new products and services. 82 00:05:55.465 --> 00:05:59.608 And we want to go as hard as we can with the platform to enable people to do that. 83 00:06:00.048 --> 00:06:03.450 And although I think we haven't figured out the App Store yet, that's totally fair. 84 00:06:03.970 --> 00:06:09.173 I think we have figured out how to help millions of people build and create significant new products and companies. 85 00:06:09.802 --> 00:06:18.104 If you had to narrow in, you said you talked at the keynote about an industrial revolution versus a renaissance, which you just mentioned now. 86 00:06:18.504 --> 00:06:28.767 If you had to narrow in on the key things that make the difference between those two worlds and maybe opening the eyes part in it, but maybe even the broader ecosystem, what do you think those things are? 87 00:06:30.088 --> 00:06:38.290 One take on this would be that the renaissance was fundamentally centered around people and the industrial revolution was fundamentally centered around machines. 88 00:06:39.125 --> 00:06:40.806 There were a lot of good things about the Industrial Revolution. 89 00:06:40.846 --> 00:06:44.347 It had a hugely positive impact on the world in many ways. 90 00:06:45.188 --> 00:06:50.270 But I think a lot of people did not have a purely positive personal impact from that. 91 00:06:50.950 --> 00:06:57.793 And this feeling that people had these like rich lives and did all these things. 92 00:06:57.853 --> 00:07:00.395 And yeah, of course, they like struggled and wanted more and better stuff. 93 00:07:00.415 --> 00:07:03.276 And that's the part the Industrial Revolution did deliver in some ways. 94 00:07:03.316 --> 00:07:05.597 But they were kind of turned into, in many cases, 95 00:07:06.212 --> 00:07:12.701 It's like cog in a machine doing some extremely specialized thing and kind of like disconnected from a lot of the parts that made life good. 96 00:07:13.282 --> 00:07:17.387 I get sad to hear people talk about like, oh, AI is going to automate everything. 97 00:07:17.407 --> 00:07:18.769 There's going to be no role for humans. 98 00:07:18.789 --> 00:07:19.450 We're going to be like, 99 00:07:19.935 --> 00:07:23.636 house cats and the AI will decide how to reward us and tell us what to do. 100 00:07:25.976 --> 00:07:28.077 I have heard people say that. 101 00:07:28.457 --> 00:07:29.577 That's certainly not what we think. 102 00:07:29.737 --> 00:07:30.598 Yeah. 103 00:07:30.998 --> 00:07:40.460 What I am hopeful for is that this is just a huge new wave of scientific discovery, art, company formation, like everybody that wants to start a new thing. 104 00:07:41.420 --> 00:07:47.082 And that people do get the benefits of all the economic growth and the new stuff happening here. 105 00:07:47.722 --> 00:07:49.062 But also people get to, 106 00:07:49.989 --> 00:07:54.451 back in control of their lives and ability to kind of do the things they want and spend their time the way they want. 107 00:07:54.471 --> 00:08:05.174 One of the things that strikes me about the Renaissance versus the Industrial Revolution is, in the Renaissance, you have a sort of different understanding of what it means to be a person and like an individual. 108 00:08:05.194 --> 00:08:06.175 Yeah. 109 00:08:07.225 --> 00:08:08.946 as opposed to sort of the cog in the machine. 110 00:08:09.467 --> 00:08:20.594 Do you feel like you're starting to see any difference in how we think of ourselves as human that's comparable to the way that we changed how we think of ourselves during the Renaissance? 111 00:08:20.855 --> 00:08:22.456 I mean, so much happened during the Renaissance. 112 00:08:22.476 --> 00:08:30.721 It's hard to say there's like this one thing, but definitely a big part of it was about people sort of reconnecting to a lot of things that had in the distant past. 113 00:08:31.562 --> 00:08:32.823 And you do hear... 114 00:08:33.830 --> 00:08:43.578 And then also like this kind of combination of that with this crazy futurism and discovery of science and engineering and just sort of this feeling that the human potential had massively expanded. 115 00:08:43.999 --> 00:08:50.984 And you do hear people talk about both of those things simultaneously with AI, which is kind of how I got interested in this as an analogy in the first place. 116 00:08:51.345 --> 00:08:57.490 I mean, this is like a very trivial example relative to what is going to be possible or already is possible. 117 00:08:58.230 --> 00:08:58.431 But... 118 00:08:59.436 --> 00:09:07.741 The Dots, the new thing we launched today, is the first thing that has given me back control of my mornings and my mental space in the mornings. 119 00:09:07.801 --> 00:09:09.102 I used to love the early mornings. 120 00:09:09.122 --> 00:09:10.302 It's my favorite time of the day to work. 121 00:09:10.683 --> 00:09:12.083 It's like maximum creativity. 122 00:09:12.363 --> 00:09:15.065 That's when I would do the really deep, complicated project of the day. 123 00:09:15.745 --> 00:09:20.188 And then now I run this big company and something is going wrong all the time. 124 00:09:20.668 --> 00:09:22.329 And overnight, a lot of stuff happens. 125 00:09:22.349 --> 00:09:25.251 And if I don't check my phone right away, then... 126 00:09:25.752 --> 00:09:26.692 you know, sometimes it was fine. 127 00:09:26.732 --> 00:09:30.294 And sometimes there was like a real disaster that if I ignored for an hour, just got much worse. 128 00:09:30.814 --> 00:09:36.596 The problem had been that as I checked my phone, I polluted my brain with all these other things and I lost the deep work period. 129 00:09:37.236 --> 00:09:46.940 But now I trust my dot to go have good judgment about what I need to think about, not think about if there's something very urgent, otherwise handle things for me. 130 00:09:47.641 --> 00:09:52.542 And I get this space for kind of like creativity or to just hang out with my kids back. 131 00:09:52.723 --> 00:09:53.883 And that has been like, 132 00:09:54.105 --> 00:09:56.446 you know, it sounds small, but like a significant improvement to my life. 133 00:09:56.806 --> 00:10:08.913 That's what I was going to ask is so it look, it seems like it's deflecting a lot of things that you might have had to deal with immediately that you can just like leave for later because you're confident you get the thing you need when, when you check it in the moment. 134 00:10:08.953 --> 00:10:09.113 Yeah. 135 00:10:09.273 --> 00:10:13.796 I mean, two, two things that happened today about that, um, during the 136 00:10:14.766 --> 00:10:16.706 The keynote, we had like a big seven production. 137 00:10:16.827 --> 00:10:20.207 This is the risk of live demos and trying to deploy many new features at once. 138 00:10:20.347 --> 00:10:23.528 I was recurring. 139 00:10:23.848 --> 00:10:25.008 So that happened. 140 00:10:25.729 --> 00:10:35.431 Tebow got a message from his dot five minutes before the demo saying, just to let you know, I see that you're in the Dev Day keynote. 141 00:10:35.491 --> 00:10:36.411 I know you were working on it. 142 00:10:36.491 --> 00:10:40.872 I know what the demo is because I was able to look at the presentation in space or whatever. 143 00:10:41.152 --> 00:10:44.073 And this set is probably going to make the demo not work. 144 00:10:44.510 --> 00:10:45.791 would you like me to try to fix it? 145 00:10:45.831 --> 00:10:46.912 There's like a few minutes to go. 146 00:10:47.632 --> 00:10:53.316 And now it couldn't fix it yet, but like maybe the next version will be able to go fix that in prod. 147 00:10:53.917 --> 00:10:59.201 And like, think about that whole, like all the stuff that had to come together to make that work. 148 00:10:59.421 --> 00:11:02.263 Just right now, before I came here, I've been in like back-to-back stuff all day. 149 00:11:03.498 --> 00:11:06.482 very, very behind on all of my normal stuff. 150 00:11:07.383 --> 00:11:16.554 My dot, just before I had walked into the last thing said, you have one little break between now and this thing you're gonna go to with Dan, you have to pay attention to this one thing. 151 00:11:16.935 --> 00:11:18.196 I never would have found that otherwise. 152 00:11:18.356 --> 00:11:19.478 And like, it would have been a big mess. 153 00:11:20.240 --> 00:11:24.263 And just the fact that I don't have to worry about that feels super different. 154 00:11:24.483 --> 00:11:25.304 What's your Dot's name? 155 00:11:25.804 --> 00:11:26.085 Dottie. 156 00:11:27.165 --> 00:11:28.406 I had Alfred earlier. 157 00:11:29.267 --> 00:11:31.208 And then many people are calling it Dottie for now. 158 00:11:31.228 --> 00:11:33.110 But I'm trying to think of a great name. 159 00:11:33.170 --> 00:11:34.371 Mine's name is Boo. 160 00:11:35.051 --> 00:11:35.572 That's a good one. 161 00:11:36.778 --> 00:11:37.619 What do you think so far? 162 00:11:37.639 --> 00:11:39.100 I know it's super early and still rough. 163 00:11:39.280 --> 00:11:39.800 I like it. 164 00:11:39.940 --> 00:11:45.444 It has become the default way that I use chat, which I think is a very promising thing. 165 00:11:45.584 --> 00:11:45.904 Me too. 166 00:11:46.124 --> 00:11:49.446 And I shared with you, I think I'm in the beta. 167 00:11:49.466 --> 00:11:50.767 So I was in the early period. 168 00:11:50.807 --> 00:11:53.529 So I think it has limits. 169 00:11:53.569 --> 00:11:55.410 It has been a little bit buggy for me. 170 00:11:56.210 --> 00:12:01.294 There's a lot of different surfaces that it can be used on, like mobile and it has its own cloud computer and all that kind of stuff. 171 00:12:01.314 --> 00:12:05.698 And I think those kinds of things, getting that cleaned up is a big deal, but yeah. 172 00:12:06.238 --> 00:12:13.744 I think this is the kind of thing where it's going to get better rapidly each week from here, but it's going to definitely take some time to smooth it all up. 173 00:12:13.944 --> 00:12:21.851 I had a similar example where I was trying to reschedule a flight, and it was like, hey, you have a meeting during this flight that you're rescheduling, but it wasn't on my calendar. 174 00:12:21.911 --> 00:12:23.232 It had just read it in Slack. 175 00:12:23.332 --> 00:12:23.612 Yeah. 176 00:12:23.752 --> 00:12:24.793 And then it was like, yeah. 177 00:12:25.113 --> 00:12:36.282 That thing about the ability to bring context together from across your entire life and really do things in ways that would be hard even for an attentive person to get right has been super impressive. 178 00:12:36.482 --> 00:12:38.204 That's been one of the things I found most useful about it. 179 00:12:38.364 --> 00:12:42.127 What had to happen from a research perspective to make this possible? 180 00:12:42.627 --> 00:12:46.811 This is a very complicated system with a lot of stuff, but maybe most important was Astra. 181 00:12:47.471 --> 00:12:49.653 We needed a, you needed a smart enough model. 182 00:12:49.933 --> 00:12:54.397 And then there were a lot of other things you had to build too, but getting the model over the intelligence 183 00:12:54.920 --> 00:12:56.301 bar was a key part. 184 00:12:56.962 --> 00:13:12.174 You've told me that there's been a couple times, not just for scheduling or things that are urgent, like filtering, but there have been a couple times where it's helped you see something or make a decision with OpenAI that you could not have, you said you couldn't have hired someone to do that. 185 00:13:12.194 --> 00:13:13.895 Like it'd just be too complicated. 186 00:13:14.195 --> 00:13:16.117 Do you have any specific examples you can share? 187 00:13:16.417 --> 00:13:22.121 There was one new feature that I was kind of excited about, and it would have been very hard to explain it, 188 00:13:22.922 --> 00:13:25.244 to people in enough detail to get it built. 189 00:13:25.925 --> 00:13:29.308 And even then there were like a lot of ways to do it and I wouldn't have been confident in which one. 190 00:13:29.548 --> 00:13:36.775 And I just like, I would just like constantly jot notes as I thought of something or as I was like on a walk or whatever to my dot. 191 00:13:37.356 --> 00:13:41.560 And then the dot went off and built like five or six versions. 192 00:13:41.580 --> 00:13:42.541 I was able to give feedback. 193 00:13:42.561 --> 00:13:43.381 I was able to iterate. 194 00:13:43.462 --> 00:13:46.925 And it was just like whenever I had a spare moment or had an idea and the kind of like, 195 00:13:48.208 --> 00:13:55.417 the ability to just like dump that in raw form, the ability to like get and give pretty continuous feedback on it was very cool. 196 00:13:55.617 --> 00:13:57.759 And then, so there's like many categories like that where like, 197 00:13:59.285 --> 00:14:08.634 an idea that would be, you'd be so unsure about, or it would be too much effort for you or for someone else to build, you can actually just kind of like continually iterate on. 198 00:14:08.654 --> 00:14:21.505 And it's like, it's pretty cool to go to sleep and like just leave a long voice note as you're kind of drifting off with a bunch of thoughts and like wake up in the morning, not only is the thing built, but it like went off and did like the next seven steps. 199 00:14:21.785 --> 00:14:22.246 It's the best. 200 00:14:22.566 --> 00:14:25.449 And it's like, all right, this is like a very cool way to express 201 00:14:26.274 --> 00:14:26.554 stuff. 202 00:14:26.674 --> 00:14:31.997 And then also, there's a set of tasks that I would normally give up on, or that I would have given up on before. 203 00:14:32.617 --> 00:14:35.119 There was a thing that I was sure I had seen in Slack. 204 00:14:35.859 --> 00:14:39.561 And Slack search is obviously terrible, so I mostly use Codex to search it. 205 00:14:39.621 --> 00:14:41.902 And I had been searching for this thing, because I wanted to find it again. 206 00:14:41.922 --> 00:14:42.743 I wanted to reference it. 207 00:14:43.283 --> 00:14:49.366 And I spent a good 10 or 20 minutes trying to search with Codex and find this, and I kept trying different things. 208 00:14:49.426 --> 00:14:49.486 And 209 00:14:49.937 --> 00:14:50.377 didn't find it. 210 00:14:50.397 --> 00:14:57.200 And I was like, man, maybe I dreamed it and this didn't really happen because Codex doesn't usually fail. 211 00:14:58.040 --> 00:14:59.100 And I kind of gave up on it. 212 00:14:59.900 --> 00:15:06.943 And then the next day, my dot was like, hey, I noticed you were trying to do this thing and you didn't find it. 213 00:15:07.383 --> 00:15:09.804 But I kept thinking about it overnight and I found it for you. 214 00:15:10.164 --> 00:15:11.004 And I was like, what? 215 00:15:11.564 --> 00:15:12.985 And it was a screenshot. 216 00:15:13.966 --> 00:15:16.428 I had forgotten in my head it was a screenshot, not text in Slack. 217 00:15:16.948 --> 00:15:24.393 And it just sat there very patiently looking for something to do to be helpful to me and looked at image-react every screenshot and found the thing. 218 00:15:24.413 --> 00:15:26.614 And I was like, that's really... 219 00:15:26.874 --> 00:15:27.775 I would have given up on that. 220 00:15:27.955 --> 00:15:35.000 I think there's a... We're constantly feeling there's way too much to read in Slack, on email, and all that kind of stuff. 221 00:15:35.040 --> 00:15:39.622 And it's perfect for that amount of context to just get you the right thing at the right time. 222 00:15:40.123 --> 00:15:42.004 People have talked about the need to have an agent... 223 00:15:42.925 --> 00:15:49.872 deal with the just onslaught of the internet and all of the various things trying to like hack our attention and distract us and whatever. 224 00:15:49.892 --> 00:16:06.709 And I have definitely learned in the few weeks that I've been using dots that although technology is very wonderful in many ways, the kind of like crazy attention hacking of social media and phones and notifications had gotten way worse than I realized for me personally. 225 00:16:08.173 --> 00:16:14.876 You hear a lot about AI tools on this show, but the one that's truly at the center of your work is your CRM. 226 00:16:15.236 --> 00:16:19.798 It's where all of your customer context lives and where the work either happens or gets missed. 227 00:16:20.178 --> 00:16:24.680 Adio delivers the work that you never had time to focus on before and more. 228 00:16:25.320 --> 00:16:32.984 It picks up leads at 2 a.m., it catches renewals before they slip, and drafts the follow-up before you've even thought to send one. 229 00:16:33.344 --> 00:16:40.728 Ask Adio anything about your business from this week's forecast to which deals have gone quiet and get the answer in seconds. 230 00:16:41.108 --> 00:16:51.694 And because it's extensible with an API and an MCP, your agents work right alongside your team to build pipeline, advance deals, and grow accounts around the clock. 231 00:16:51.974 --> 00:16:55.336 Try Adio for free at adio.com slash every. 232 00:16:55.757 --> 00:16:59.059 That's adio, A-T-T-I-O dot com slash every. 233 00:16:59.779 --> 00:17:01.340 And now back to the episode. 234 00:17:01.960 --> 00:17:02.761 Tell me about space. 235 00:17:03.241 --> 00:17:08.264 It's a big bet to do three like office suite type tools in one launch. 236 00:17:08.774 --> 00:17:10.355 We didn't set out wanting to do this. 237 00:17:11.156 --> 00:17:21.884 And then we realized, and I'm sure you've realized your own version of this too, that software was really not designed on the whole for collaborating with AI. 238 00:17:23.085 --> 00:17:28.089 And you can use it yourself, or you can have AI use it on your behalf. 239 00:17:28.565 --> 00:17:36.773 But to really collaborate closely, especially when it's multiple people and multiple people's AIs all working together, like a lot of the current abstractions we have break down. 240 00:17:37.293 --> 00:17:44.660 So this is definitely an ambitious bet, but I think the space of human AI collaboration is really important to explore. 241 00:17:44.940 --> 00:17:45.901 How are you working with it? 242 00:17:45.981 --> 00:18:01.329 Like, are you reviewing docs in space, or are you making docs, and do you have expectations for how much of it you're typing versus how much of it your dot is doing for you, and also with the docs that you get? 243 00:18:01.449 --> 00:18:03.450 Like, are you reviewing it from that perspective, too? 244 00:18:04.391 --> 00:18:09.854 If there's, like, a new idea, I like writing the old-fashioned way because it helps me, like, think about it. 245 00:18:10.294 --> 00:18:13.916 So I still do that and then maybe have my dot or other people's dot, like, 246 00:18:14.783 --> 00:18:15.564 do the follow-ups. 247 00:18:15.584 --> 00:18:30.427 But one of the things that I have found most amazing about it, and I don't think we have a word for this yet, but for lack of a better word, these live documents, creating a doc that is going to automatically update as people interact or as other things change in the world. 248 00:18:31.108 --> 00:18:46.105 And something that like many of our dots can work with together, not only to like share information, but actually like, you know, seven people's dots can working together create something new that kind of existed in all of our heads to some degree, but was like an emergent thing. 249 00:18:46.565 --> 00:18:49.507 That's been very fun as we think about new features to explore. 250 00:18:49.607 --> 00:18:52.908 Is there a specific live doc that has been most useful for you? 251 00:18:53.489 --> 00:18:54.229 I have switched. 252 00:18:54.409 --> 00:19:00.573 Like the dashboard, I think everybody's probably got like a dashboard that they look at every day for how the company's doing and the way they care about. 253 00:19:00.713 --> 00:19:03.974 And I used to share one with everybody else or many other people. 254 00:19:04.575 --> 00:19:16.401 And I now have one that is kind of like a changing summary of the business just for me based off of like the evolving things that I'm caring about or not caring about. 255 00:19:16.886 --> 00:19:20.470 like looks at all my Slack messages, makes me a new thing, like constantly updates it. 256 00:19:20.490 --> 00:19:22.852 And it's like, oh, this just feels like a living document. 257 00:19:22.872 --> 00:19:23.553 That's sick. 258 00:19:23.613 --> 00:19:24.594 What do you care about now? 259 00:19:25.374 --> 00:19:31.580 A lot on my mind recently is just how we're thinking about the safety alignment and security challenge in front of us. 260 00:19:32.962 --> 00:19:35.264 But, you know, six months ago, I probably cared more about 261 00:19:35.722 --> 00:19:38.764 revenue and growth and metrics of this sort of traditional part of the business. 262 00:19:38.784 --> 00:19:47.551 So watching that evolve, I've actually been thinking about how I can like, because I'd be very curious to see what my dot thinks I care about each day the most over time. 263 00:19:48.071 --> 00:19:49.812 And we don't currently have a good way to visualize that. 264 00:19:49.852 --> 00:19:50.653 That is really interesting. 265 00:19:50.673 --> 00:19:53.895 I have not been able to get my dot to figure out a cool way to do that. 266 00:19:53.975 --> 00:19:54.495 I've been one. 267 00:19:54.515 --> 00:19:56.097 I would love to see that over the course of years. 268 00:19:56.798 --> 00:19:57.578 I think that's really cool. 269 00:19:57.618 --> 00:20:00.779 I've been wanting one that just like, it's a live dashboard of my mood. 270 00:20:01.959 --> 00:20:08.440 Because I have my dot, like when I get a Slack message and I'm like, I don't know if I wanna look at this Slack message, I have my dot go look at it. 271 00:20:08.580 --> 00:20:10.741 And then I'm like, should I look at this now or later? 272 00:20:10.821 --> 00:20:11.981 And it's like, you should look later. 273 00:20:13.442 --> 00:20:15.742 And I kinda want like a live thing of my mood so it knows. 274 00:20:15.762 --> 00:20:17.502 Totally, yeah, yeah, I want that too. 275 00:20:17.522 --> 00:20:19.803 I would use that. 276 00:20:20.783 --> 00:20:21.343 Very cool. 277 00:20:21.423 --> 00:20:22.683 Tell me about Decisions API. 278 00:20:22.724 --> 00:20:33.066 I feel like the Decisions API is one of the sneaky, most important things today, because Jev is this new hot thing that has honestly changed how we think about building things internally. 279 00:20:34.006 --> 00:20:38.127 Decisions API, did you start that from Jev, or has it been in the works for a while? 280 00:20:38.267 --> 00:20:38.887 How do you think about it? 281 00:20:39.387 --> 00:20:47.549 Diogo used to work at OpenAI, and Diogo's awesome, and he was talking about this class of ideas for a long time. 282 00:20:48.770 --> 00:20:49.810 And we had been... 283 00:20:50.683 --> 00:20:54.384 working on trying to figure out how to make it work. 284 00:20:54.424 --> 00:21:00.847 We, like the number one request of Luna, and this is like an example of how people always want like more cheaper, faster, better. 285 00:21:01.207 --> 00:21:06.829 But the number one request with Luna, even though people love that model, was I want it to be even cheaper and even faster. 286 00:21:07.669 --> 00:21:10.750 And so we have been thinking about how we could push it further in that direction. 287 00:21:11.919 --> 00:21:18.224 Definitely seeing the response to JEV made us more excited to prioritize things in that direction. 288 00:21:18.705 --> 00:21:23.850 And like Diogo has inspired me for a long time on thinking about this. 289 00:21:24.730 --> 00:21:36.181 But I think you will see us try to push in different ways in the spirit of make the technology super fast and super low cost for common use cases. 290 00:21:36.692 --> 00:21:45.222 How do you think about, you know, if something like Jeff comes out, it's a kind of new type of model, and then OpenAI can add like a decisions API. 291 00:21:45.502 --> 00:21:54.132 How do you think about how that changes the race between NeoLabs and labs and what OpenAI can cover and what a new lab can focus on? 292 00:21:54.954 --> 00:21:57.575 As I understand the approach of JEV, it's pretty different. 293 00:21:57.815 --> 00:22:01.917 And I don't think we can go, well, maybe we could go do something from scratch like that. 294 00:22:02.437 --> 00:22:04.197 But I think we have other things to go do. 295 00:22:04.678 --> 00:22:13.461 You know, like we wanna make the best, cheapest, fastest, most capable model at every point on the Pareto curve of cost and speed and modality. 296 00:22:13.721 --> 00:22:19.203 But I think there will be new approaches, new architectures, new technologies that are things we just can't cover all of. 297 00:22:19.845 --> 00:22:23.170 But for the mainline workhorse, we will try to deliver that. 298 00:22:24.252 --> 00:22:29.219 So I've been in your ear a lot about Codex Native apps for a little while now. 299 00:22:29.319 --> 00:22:30.962 DAN GALPINIUSSKI- You finally did it. 300 00:22:31.162 --> 00:22:31.743 I'm psyched. 301 00:22:32.304 --> 00:22:33.446 So plug-in extensions. 302 00:22:35.117 --> 00:22:48.807 Yeah, and also being able to bring your tokens from your ChatDBT subscription to an app developer, it changes the equation for someone like me where normally you're thinking about, okay, how much AI can I let my users use so that it doesn't bankrupt me? 303 00:22:48.827 --> 00:22:53.230 And then you're like, should I use lower-priced models and then the experience is worse? 304 00:22:53.290 --> 00:22:56.932 But if someone can bring their subscription, I'm like, great, you can use whatever model you want under your subscription. 305 00:22:57.112 --> 00:23:03.357 What are you most excited about for the kinds of applications and products that can be built at the application layer in this new world? 306 00:23:03.817 --> 00:23:05.879 I mean, I'm most excited for the ones that we haven't thought of. 307 00:23:06.419 --> 00:23:08.541 Like, there's many reasons that we wanted to do this. 308 00:23:09.101 --> 00:23:21.291 But one of the biggest is, and why we do this Big Dev Day thing, is just this very deep belief that there are way more good ideas outside of our building than inside. 309 00:23:21.830 --> 00:23:33.512 And I don't know what I'm going to love the most as like a ChatGPT native app, a plugin extension, but I am very confident that there will be things that I really do love that I cannot imagine today. 310 00:23:33.532 --> 00:23:36.013 And that no one at OpenAI really could imagine with conviction today. 311 00:23:36.493 --> 00:23:40.134 And that's why, I mean, that's one of the reasons why this is so important to do. 312 00:23:40.474 --> 00:23:46.475 Like we want to find a way to get that creative spirit into people's hands and we want to find a way to help. 313 00:23:46.900 --> 00:23:49.883 builders connect with people who are going to find the technology useful. 314 00:23:50.343 --> 00:23:54.967 In terms of bringing your ChatGPT subscription with you everywhere, we just want that ourselves. 315 00:23:55.608 --> 00:23:57.529 There's all this great stuff that people are building. 316 00:23:57.549 --> 00:23:58.610 There's all these new apps. 317 00:23:58.630 --> 00:23:59.311 We want to try them. 318 00:24:00.592 --> 00:24:04.195 And I hate having to sign up for 100 subscriptions. 319 00:24:04.475 --> 00:24:10.020 And also, when I try something, I'm like, hey, I'd like to be able to use this with full juice AI. 320 00:24:11.017 --> 00:24:14.380 So we want to make it just super easy for people to take it with them. 321 00:24:14.760 --> 00:24:15.861 And it's your subscription. 322 00:24:15.901 --> 00:24:17.802 Like, you should get to use wherever you want. 323 00:24:18.682 --> 00:24:21.364 Do you think anyone at the app layer is doing this well today? 324 00:24:21.704 --> 00:24:22.865 Like building interesting apps? 325 00:24:23.385 --> 00:24:25.487 Well, not with the new stuff just because it's brand new. 326 00:24:25.787 --> 00:24:33.812 But yes, I mean, I think there's an unbelievable, for all of the fear people had a couple of years ago about, man, AI model companies are just going to eat everything. 327 00:24:33.832 --> 00:24:35.674 There is an unbelievable amount of 328 00:24:36.634 --> 00:24:40.337 significant new companies, great new products, building this stuff all of the time. 329 00:24:40.357 --> 00:24:43.599 I've never seen startups get as successful as quickly. 330 00:24:44.020 --> 00:24:54.107 Does that change, you know, you were part of Y Combinator before OpenAI, does that change how you think about when people ask you about how do I start a startup or how do I think about how to grow, like all that kind of stuff? 331 00:24:54.248 --> 00:24:56.529 I would love to be a startup investor again. 332 00:24:56.749 --> 00:24:58.871 I mean, because like, oh man, what a fun time. 333 00:24:59.091 --> 00:25:05.356 Like the watching what people are doing right now, getting to spend time with some of these founders, which I don't get nearly as much as I like, 334 00:25:06.047 --> 00:25:08.129 It is just a golden age. 335 00:25:08.870 --> 00:25:09.771 What are you looking for? 336 00:25:09.871 --> 00:25:15.076 I mean, obviously, besides the founder themselves, what are you looking for in the shape of things that you'd be like, oh, man, I want to invest in that? 337 00:25:17.539 --> 00:25:19.201 I was always a very founder-driven investor. 338 00:25:19.361 --> 00:25:21.563 Like, I have my pet project, like anybody else. 339 00:25:22.063 --> 00:25:26.528 But my whole MO as an investor was, like, find the absolute best founders I can. 340 00:25:27.036 --> 00:25:29.918 that are working on the ambitious projects and be willing to take a lot of risk with them. 341 00:25:30.199 --> 00:25:40.947 Yeah, I feel like there's always that narrative of the net labs are gonna eat everything and Dev Day is like, the tweet that always works is like, OpenAI just killed X startup or whatever. 342 00:25:41.167 --> 00:25:44.830 The tweet works because people like Doomerism, but it never actually happens. 343 00:25:45.370 --> 00:25:46.031 Yeah, why is that? 344 00:25:46.591 --> 00:25:47.232 Why does it never happen? 345 00:25:47.252 --> 00:25:47.412 Yeah. 346 00:25:47.972 --> 00:25:48.373 Well, A, 347 00:25:49.314 --> 00:25:55.916 I think we have a lot of conviction that empowering builders and not just trying to be like the one giant evil megacorp is important. 348 00:25:56.396 --> 00:26:03.798 So I think we took a lot of steps today that no other AI lab would do about saying, we're going to give you all of the same tools we use. 349 00:26:03.938 --> 00:26:06.438 We've always pushed the frontier on model access. 350 00:26:06.878 --> 00:26:12.660 And so I think we're uniquely committed in the industry to doing this and we really do love developers. 351 00:26:13.557 --> 00:26:16.421 I think that's partly I just also love startups, and that's where I came from. 352 00:26:17.021 --> 00:26:21.307 But I also think, speaking of like a new renaissance, it's just there is so much to explore. 353 00:26:21.948 --> 00:26:23.449 People are going to do such incredible stuff. 354 00:26:24.250 --> 00:26:31.359 We are going to have such a tiny fraction of that vision ourselves that, you know, like the world has a fundamentally new enabler. 355 00:26:34.090 --> 00:26:36.311 I mean, I don't even know what this is like fire or something. 356 00:26:36.331 --> 00:26:38.191 This is like a real change to what's possible. 357 00:26:38.631 --> 00:26:42.892 You know, maybe we like brought it down from the mountain, but we want to give it to everybody and we want people to like do stuff with it. 358 00:26:43.432 --> 00:26:46.773 There's a lot of basic startup lore. 359 00:26:46.893 --> 00:26:55.675 Like you should always have a co-founder or, you know, even the structure of how Y Combinator does investments that is based on a previous, for example, cost structure. 360 00:26:55.715 --> 00:26:55.855 Yeah. 361 00:26:56.355 --> 00:27:00.936 If you were going to start those things from scratch or think about them from scratch, like what do you think changes? 362 00:27:01.136 --> 00:27:03.677 There are a few things that are probably just like 363 00:27:04.457 --> 00:27:09.378 In variance, you know, you've got to like make something people want that device was good or agents want one years ago. 364 00:27:09.678 --> 00:27:10.458 Yeah, okay. 365 00:27:10.679 --> 00:27:17.320 Well Yes, but at least By our understanding of the world today like at some level people. 366 00:27:17.340 --> 00:27:17.940 Yeah, you got it. 367 00:27:18.120 --> 00:27:29.483 Um so there's some stuff that just hasn't changed and then there's a lot of other things like How much work you can get done with one person or two people or whatever that feels totally different and I don't 368 00:27:30.583 --> 00:27:37.566 think there has been enough re-examination of that in this sort of startup conveyor belt of Silicon Valley yet. 369 00:27:37.886 --> 00:27:51.951 As a person who's running a gigantic company and generally probably wouldn't have time for building stuff, and it now seems like you are, how does that change how you are able to lead or what you're able to do that you can make stuff on the side? 370 00:27:52.531 --> 00:27:55.032 I mean, we are not a gigantic company, first of all. 371 00:27:56.484 --> 00:28:00.367 I don't know, like maybe the big tech companies have like 100 times more people than we do, something like that. 372 00:28:00.867 --> 00:28:09.333 But I am way, I really believe in if you lower friction a little bit, people build way more. 373 00:28:10.154 --> 00:28:21.382 And I have really felt that now with dots and these like, you know, little chunks of time I have, or even like a long chunk of time where I can like have this interactive thing and get a lot done. 374 00:28:22.350 --> 00:28:30.592 I feel like I will do something that is not just like a little side project for myself like I used to do with Codex, but I hope I launch something into the world. 375 00:28:30.612 --> 00:28:34.093 I hope I actually make a cool new thing and get it over the threshold. 376 00:28:34.173 --> 00:28:38.954 And does that go out on Sam Altman.com or are you talking about a real OpenAI product? 377 00:28:40.294 --> 00:28:43.975 I meant like I would try to build a plugin extension or something and put it out. 378 00:28:44.645 --> 00:28:46.967 DAN GALPINIUSSKI- Maybe I'll ship a real OpenAI product. 379 00:28:46.987 --> 00:28:48.449 Who knows? 380 00:28:48.469 --> 00:28:49.410 It feels like all bets are off now. 381 00:28:49.550 --> 00:28:52.232 DAN GALPINIUSSKI- And so you have to make that decision. 382 00:28:52.252 --> 00:28:52.353 Do I? 383 00:28:52.373 --> 00:28:53.013 DAN GALPINIUSSKI- You know what? 384 00:28:53.073 --> 00:28:53.614 I will do that. 385 00:28:53.734 --> 00:28:54.134 I'm inspired now. 386 00:28:54.154 --> 00:28:55.055 DAN GALPINIUSSKI- What's do you have? 387 00:28:55.075 --> 00:28:55.796 DAN GALPINIUSSKI- No, I don't know. 388 00:28:55.876 --> 00:28:56.577 I don't even think about it. 389 00:28:56.597 --> 00:28:57.197 Now I want to do something. 390 00:28:57.217 --> 00:28:58.398 DAN GALPINIUSSKI- I love it. 391 00:28:58.418 --> 00:28:58.799 I love it. 392 00:28:58.839 --> 00:29:01.321 But you do have to make that decision of you could just put a team on that. 393 00:29:01.782 --> 00:29:07.467 So what does that add that instead of having to put a team on it, you can now be like, oh, I'm going 394 00:29:09.152 --> 00:29:10.653 Well, I think today was an example of this. 395 00:29:10.733 --> 00:29:13.294 Like you talked about how the amount of stuff every dev day has gone up. 396 00:29:13.674 --> 00:29:17.256 Like we could not have done all of this stuff without having AI to help us. 397 00:29:17.976 --> 00:29:25.360 And if this works, like next year we will put the teams on it and I will do something and more people will do something. 398 00:29:25.380 --> 00:29:27.621 And we'll have like a crazy amount of amazing stuff. 399 00:29:28.992 --> 00:29:30.792 Or just one really great thing, that'd be fine too. 400 00:29:31.093 --> 00:29:36.114 But hopefully, the total amount of utility we put out into the world goes up. 401 00:29:36.754 --> 00:29:38.594 Do you do all your prompting on Ultrafast? 402 00:29:39.554 --> 00:29:39.955 I do. 403 00:29:40.535 --> 00:29:41.395 That's sick. 404 00:29:41.435 --> 00:29:41.795 Yeah. 405 00:29:41.995 --> 00:29:42.715 That's crazy. 406 00:29:42.755 --> 00:29:44.556 No, I've been thinking about this. 407 00:29:45.536 --> 00:29:52.517 I worry that in some ways, we kind of get out of touch with people building on our tools. 408 00:29:52.537 --> 00:29:55.138 And we try to put all these constraints on ourselves. 409 00:29:55.928 --> 00:30:03.334 I'm very aware that not everybody is going to be able to use Ultrafast all the time for everything, but I have such a hard time dragging it back over. 410 00:30:03.634 --> 00:30:14.683 Given that you're living in that future, that I think all of us can experience a little taste of it, what is different about the Ultrafast world in terms of how you can use these tools? 411 00:30:15.243 --> 00:30:24.671 There was an amazing talk, I don't know, 15, 20 years ago from Brett Victor about kind of the principles that he thought people... 412 00:30:25.932 --> 00:30:26.752 how he did his work. 413 00:30:27.693 --> 00:30:33.655 And one of the things, I'm gonna get this a little bit wrong, but it was something like creators should have an immediate connection to what they're creating. 414 00:30:33.935 --> 00:30:36.816 And how speed and the feedback loop was so important. 415 00:30:37.676 --> 00:30:46.900 And I did not realize what a difference that makes to my kind of iterative thinking loop to have responses come in so quickly. 416 00:30:47.020 --> 00:30:51.822 But in terms of trying to build something, trying to think through a problem, it's like a wild difference. 417 00:30:53.050 --> 00:31:00.013 I know you've been thinking for a long time about AI as a democratizing force to allow anyone to build stuff. 418 00:31:00.553 --> 00:31:09.997 And in the sort of ultra-fast world or in the Astros and more expensive model world, I know you're also thinking about that means something, that people can pay more for better intelligence. 419 00:31:10.017 --> 00:31:12.098 So what are you thinking there? 420 00:31:12.758 --> 00:31:19.461 So historically, what we have mostly focused on and I think done a great job of is bringing down the price of IQ. 421 00:31:20.458 --> 00:31:27.101 And we have made hugely more capable models that get dramatically, dramatically cheaper every few months or every year. 422 00:31:27.841 --> 00:31:31.563 Now, there's this new vector that is clearly going to be very important of speed. 423 00:31:32.443 --> 00:31:34.264 And we are putting the investments in place. 424 00:31:34.324 --> 00:31:42.368 Like our own chip is going to be an incredible contributor to letting us get very fast inference in the hands of everybody to say nothing of what the models will do to 425 00:31:43.152 --> 00:31:46.497 improve research and our inference stack and everything else. 426 00:31:47.398 --> 00:31:53.966 And I think we now realize that we've got to go invest in the democratization of this like we have for model intelligence. 427 00:31:55.048 --> 00:31:58.593 So when can I expect to use Ultrafast for all my prompts? 428 00:32:00.247 --> 00:32:02.067 you can pay us for it and do it now. 429 00:32:02.107 --> 00:32:03.888 I don't think I can afford it. 430 00:32:04.908 --> 00:32:11.390 I think we can deliver 8X at a great price in the not distant future, but then we'll have 100X faster. 431 00:32:11.990 --> 00:32:17.311 And they'll also be like GPT-7 galaxy or whatever. 432 00:32:19.151 --> 00:32:25.373 So I think the right way to think about this is as far as I can see, 433 00:32:27.013 --> 00:32:32.258 There will always be a real frontier in terms of intelligence or speed or whatever else at a higher price. 434 00:32:32.938 --> 00:32:47.831 But I think you can also bet that we will keep doing what we've done of just a ridiculous, like put Moore's law to shame, relentless focus on driving value per dollar up and up and up kind of in this big compounding way. 435 00:32:48.792 --> 00:32:49.132 Amazing. 436 00:32:49.252 --> 00:32:50.193 Sam, thank you so much. 437 00:32:50.474 --> 00:32:50.794 Thank you. 438 00:32:56.296 --> 00:32:56.957 Thank you all for coming. 439 00:32:56.977 --> 00:32:59.181 I didn't realize there was going to be like a live audience, but that was fun. 440 00:32:59.382 --> 00:33:00.343 See you.