WEBVTT NOTE This file was generated by Descript 00:00:00.500 --> 00:00:02.179 Nabeel Hyatt: Do we know what this thing is gonna be called, Fraser? 00:00:02.750 --> 00:00:05.990 Fraser Kelton: No, but we decided that the world definitely needs another podcast. 00:00:06.070 --> 00:00:09.760 That is exactly what, exactly what's required. 00:00:09.810 --> 00:00:12.440 ChatGPT came up with some names. 00:00:12.440 --> 00:00:18.150 They were all pretty bad puns playing on podcasts, AI and startups, none of which 00:00:18.150 --> 00:00:19.790 had any taste . I forget what they were. 00:00:19.920 --> 00:00:21.870 Nabeel Hyatt: If you can pull them up at some point, maybe towards the 00:00:21.870 --> 00:00:23.420 end of the segment, it'll be great. 00:00:23.970 --> 00:00:25.700 So, I am Nabeel Hyatt. 00:00:26.700 --> 00:00:27.690 Fraser Kelton: I'm Fraser Kelton. 00:00:28.595 --> 00:00:33.685 Nabeel Hyatt: We are doing a thing that is at least spiritually connected to 00:00:33.795 --> 00:00:39.105 a podcast that I used to do with my old friend and partner, Bijan, when 00:00:39.105 --> 00:00:41.025 I basically first got into venture. 00:00:41.345 --> 00:00:45.045 We started a thing called Hallway Chat and the goal was not to be, 00:00:45.165 --> 00:00:48.905 I mean, very clearly, not to be the most well structured, highest 00:00:48.915 --> 00:00:51.115 production value place on the block. 00:00:51.345 --> 00:00:55.455 The goal was really just to lead with authenticity about them really crazy time. 00:00:55.455 --> 00:00:58.455 At the beginning of the mobile revolution, where it felt like everything 00:00:58.455 --> 00:00:59.715 was changing every single week. 00:01:00.255 --> 00:01:00.375 And. 00:01:00.375 --> 00:01:01.515 of course, now we have. 00:01:01.515 --> 00:01:02.685 a new time. 00:01:02.745 --> 00:01:03.285 in AI. 00:01:03.285 --> 00:01:03.495 That. 00:01:03.495 --> 00:01:04.785 feels an awful lot like that. 00:01:04.785 --> 00:01:06.675 And I've got a new partner in Frazier. 00:01:06.705 --> 00:01:09.521 Who's uh, now the new VC, who's a former. 00:01:09.881 --> 00:01:11.961 Founder and an operator , actually Fraser. 00:01:11.961 --> 00:01:13.761 Do you want to talk about yourself at all? 00:01:14.005 --> 00:01:14.785 Fraser Kelton: No, I do. 00:01:14.805 --> 00:01:17.019 Nobody wants to hear us talk about ourselves on this. 00:01:17.069 --> 00:01:17.754 I don't think. 00:01:17.868 --> 00:01:19.228 You say this to founders all the time? 00:01:19.228 --> 00:01:23.958 There's nothing worse than VCs bloviating about themselves and their ideas. 00:01:23.958 --> 00:01:26.148 And so we have to keep the second part of this. 00:01:26.148 --> 00:01:28.928 Otherwise we don't have a podcast, but we can save people from the first. 00:01:29.928 --> 00:01:32.398 Nabeel Hyatt: There's these things called, search engines, although we will 00:01:32.398 --> 00:01:35.888 discuss the potential disruption and those such things where you can look 00:01:35.888 --> 00:01:37.392 up our bios and figure out who we are 00:01:37.719 --> 00:01:41.016 And if tradition holds and every week, we're going to focus on our new 00:01:41.196 --> 00:01:44.976 AI product that we're actually using and trying to wrestle with and any of 00:01:44.976 --> 00:01:46.056 the insights that come out of that. 00:01:46.686 --> 00:01:50.436 Uh, just crying, frankly, stay curious as we try to navigate all this new space. 00:01:50.916 --> 00:01:51.456 And of course. 00:01:51.876 --> 00:01:54.846 Frazier we'll navigate the new venture landscape.. 00:01:54.846 --> 00:01:55.386 together. 00:01:55.806 --> 00:01:58.986 Uh, you can try and keep the one venture question per week. 00:01:59.687 --> 00:02:00.187 Fraser Kelton: Okay. 00:02:00.367 --> 00:02:00.607 Yeah. 00:02:00.607 --> 00:02:00.767 Yeah. 00:02:01.647 --> 00:02:03.277 try my best to limit it to one. 00:02:03.277 --> 00:02:04.367 This is a weird world. 00:02:05.117 --> 00:02:08.837 So you, you told me the other week that you use perplexity 00:02:08.847 --> 00:02:11.807 for basically everything that you used to use Google for. 00:02:12.687 --> 00:02:14.187 Nabeel Hyatt: Yeah, we did have this conversation. 00:02:15.047 --> 00:02:18.767 I, I don't use Google anymore really as a search engine. 00:02:19.127 --> 00:02:21.867 I, I find it's just not good enough. 00:02:22.217 --> 00:02:25.247 That doesn't mean I don't use Google because it turns out there's a lot of 00:02:25.267 --> 00:02:28.767 things Google's done over the years that are not about search engines per se. 00:02:29.287 --> 00:02:32.267 But pretty much all of my searches have gone to perplexity and then 00:02:32.307 --> 00:02:35.397 some percentage of what you would loosely categorize as search. 00:02:35.882 --> 00:02:37.372 Now goes to Claude or ChatGPT. 00:02:38.362 --> 00:02:43.332 And yeah, I, I think perplexity I think there's rumors about IVP 00:02:43.442 --> 00:02:46.842 having invested, that we're having the information this week, some 00:02:46.842 --> 00:02:48.442 large round and so on and so forth. 00:02:48.442 --> 00:02:52.362 And, on the surface it feels insane to to be investing and 00:02:52.362 --> 00:02:54.662 looking at something that's trying to compete directly with Google. 00:02:55.092 --> 00:02:59.146 But at the same time, how many amazing AI products have you actually used and stuck 00:02:59.146 --> 00:03:00.994 in the last nine months since ChatGPT? 00:03:01.514 --> 00:03:06.434 And, and the obvious one that people talk about is probably CodeComplete, right? 00:03:06.974 --> 00:03:07.634 But, but 00:03:07.684 --> 00:03:09.804 Fraser Kelton: massive numbers, user numbers this week. 00:03:09.805 --> 00:03:13.144 So I think it's it's GitHub copilot. 00:03:13.714 --> 00:03:14.964 It's chat GPT. 00:03:15.734 --> 00:03:21.064 There's a lot of interest early on around RAG like systems for chatting with your 00:03:21.074 --> 00:03:23.644 docs chatting with enterprise content. 00:03:23.739 --> 00:03:24.519 Nabeel Hyatt: talk to my PDFs. 00:03:24.586 --> 00:03:25.106 Fraser Kelton: I think so. 00:03:25.106 --> 00:03:26.866 I want my PDFs to talk back to me though. 00:03:26.916 --> 00:03:27.513 That's what I want. 00:03:27.566 --> 00:03:31.166 I want the synthesized voice of Sky to come back and tell me that. 00:03:31.396 --> 00:03:32.566 That I'm writing this memo wrong. 00:03:33.076 --> 00:03:37.146 But, talk a little bit about perplexity and the product experience and the 00:03:37.146 --> 00:03:42.106 user like value that you get from it versus chat GPT versus Google. 00:03:42.116 --> 00:03:45.306 Like it's interesting to hear that you go there all of the time for what 00:03:45.306 --> 00:03:47.186 used to be, quote unquote search. 00:03:48.186 --> 00:03:51.736 Nabeel Hyatt: I think it's a great example of sometimes little things matter. 00:03:52.401 --> 00:03:58.481 I think Arvind and that team have done a lot of subtle things that 00:03:58.501 --> 00:04:01.121 make that product really work that are worth chatting about. 00:04:01.131 --> 00:04:05.721 First of all, the, the like six months ago when agents were suddenly 00:04:05.721 --> 00:04:08.891 all the thing for a minute they launched a little, a little agent 00:04:08.921 --> 00:04:10.501 which takes your search engine query. 00:04:10.966 --> 00:04:15.166 and then tries to ask follow up clarifying questions in order to narrow 00:04:15.166 --> 00:04:16.936 down and make your search better. 00:04:17.146 --> 00:04:19.506 And it's a great example of does it work all the time? 00:04:19.536 --> 00:04:20.086 No. 00:04:20.126 --> 00:04:22.036 When it works, it's absolutely magical. 00:04:22.036 --> 00:04:27.366 And so, you're searching for something and then it actually, in dynamic 00:04:27.396 --> 00:04:32.046 real time, builds a little UI of, say, checkboxes or radio buttons that 00:04:32.046 --> 00:04:33.610 say, oh, did you mean this or this? 00:04:33.740 --> 00:04:37.714 Now, obviously, in the world of agents, everybody's figuring out exactly, it's 00:04:37.714 --> 00:04:41.694 almost like agents are like self driving cars you can demo them, 80 percent in 00:04:41.694 --> 00:04:44.384 a wonderful demo in two months, but actually getting them to be really 00:04:44.394 --> 00:04:48.084 performant and reliable is a, is a Herculean task, as I think all our builder 00:04:48.084 --> 00:04:49.904 friends are, are figuring out right now. 00:04:49.904 --> 00:04:53.264 But, but I think it actually works very well in Perplexity's case. 00:04:53.654 --> 00:04:57.194 And not least of which is a good example is Have you used Google's new 00:04:57.274 --> 00:04:59.164 generative engine that they launched? 00:04:59.264 --> 00:04:59.894 The conversational 00:04:59.944 --> 00:05:02.504 they're basically their competitor to Perplexity. 00:05:02.589 --> 00:05:02.819 Fraser Kelton: yeah. 00:05:03.129 --> 00:05:05.409 SGE Search Generative Experience. 00:05:05.429 --> 00:05:09.589 I will tell you, it was the first time that I really had 00:05:09.589 --> 00:05:10.799 to reckon with two things. 00:05:10.799 --> 00:05:14.339 Number one how painful it is when you have to ship your org 00:05:14.359 --> 00:05:15.599 chart in your business model. 00:05:15.859 --> 00:05:19.399 Because, yeah, you can now, you can now use it to do your 00:05:19.399 --> 00:05:20.659 first draft, they call it. 00:05:20.679 --> 00:05:23.749 So you go to the search engine that you've used for 25 years and you 00:05:23.749 --> 00:05:25.779 can write write me a poem about... 00:05:25.944 --> 00:05:26.614 Tulips. 00:05:27.114 --> 00:05:31.004 And it will write the poem about tulips, and then you have your search results 00:05:31.014 --> 00:05:33.324 for, for poems about tulips at the bottom. 00:05:33.434 --> 00:05:34.234 Like it just makes 00:05:34.544 --> 00:05:36.684 Nabeel Hyatt: Oh, wait, just they still put in the Google searches, right? 00:05:36.684 --> 00:05:38.734 With the Google ads and things, it makes no sense, 00:05:39.304 --> 00:05:41.224 Fraser Kelton: it makes no sense, no sense at all. 00:05:41.254 --> 00:05:45.844 And the, the other thing that I realized was They're vulnerable and 00:05:45.844 --> 00:05:46.954 they realize that they're vulnerable. 00:05:46.954 --> 00:05:51.884 If they're willing to ship this type of janky experience into their core product 00:05:51.954 --> 00:05:56.224 they, they don't know what to do because they have a business model, which is, 00:05:56.234 --> 00:05:59.874 maybe the best business model of all, of all time, that's at real risk here. 00:06:00.094 --> 00:06:03.854 And their solution is to allow you to do your first draft within 00:06:03.854 --> 00:06:05.174 the search results below it. 00:06:05.414 --> 00:06:07.824 I think maybe the optimistic story on this is. 00:06:08.119 --> 00:06:11.169 They're shipping early, they're going to learn, they're going to iterate 00:06:11.179 --> 00:06:14.519 and they're going to, to figure it out because it's so economically 00:06:14.519 --> 00:06:15.739 important for them to do it. 00:06:16.139 --> 00:06:19.399 But I think that this is the first time in the past 11 months where I 00:06:19.399 --> 00:06:22.159 thought, Oh Oh, this is going to be a really interesting couple of years. 00:06:23.079 --> 00:06:25.619 I mean, and that's the optimistic story for, for Arvind, right? 00:06:25.669 --> 00:06:26.699 And perplexity. 00:06:27.694 --> 00:06:28.534 Nabeel Hyatt: Yeah, I think it is. 00:06:28.584 --> 00:06:32.834 In general with AI products, I've just tried to take the idea of slowing 00:06:32.834 --> 00:06:35.294 down all of my computer use, right? 00:06:35.304 --> 00:06:38.004 That's how I've thought about the next two years, which is just if I'm 00:06:38.004 --> 00:06:42.514 gonna go do a search, then one out of every three times, take the time. 00:06:42.929 --> 00:06:45.159 Do the search on three or four different engines. 00:06:45.179 --> 00:06:46.249 Try weird stuff. 00:06:46.309 --> 00:06:49.779 And yes, that slows down everything, but you have to test your defaults. 00:06:49.899 --> 00:06:52.879 If you're gonna, if you're gonna be on the edge, if you're gonna really feel 00:06:52.909 --> 00:06:55.599 how all these things are really working, and you'll take the time to do that, and 00:06:55.619 --> 00:06:57.069 have to fit into your day to day life. 00:06:57.309 --> 00:07:00.509 Otherwise, it's some random thing for a half an hour on a Thursday 00:07:00.509 --> 00:07:02.679 afternoon, you're like, I'm supposed to try this product now. 00:07:02.689 --> 00:07:03.469 What's my search? 00:07:03.489 --> 00:07:04.269 And that just doesn't 00:07:04.969 --> 00:07:05.399 Fraser Kelton: Right. 00:07:05.529 --> 00:07:09.049 Nabeel Hyatt: I, I, so I've tried, I've probably done dozens of, of... 00:07:09.644 --> 00:07:14.634 of searches on Bing, the new Google ish thing and Perplexity. 00:07:15.094 --> 00:07:17.204 I had a weird one where, I wear a beanie all the time. 00:07:17.464 --> 00:07:18.974 I saw Brad Pitt wearing a beanie. 00:07:19.344 --> 00:07:21.894 I was like, where does he get his, where does he get his beanie? 00:07:22.214 --> 00:07:24.874 And if you just think about the old world of Google, where 00:07:24.874 --> 00:07:25.684 would that content come from? 00:07:26.124 --> 00:07:28.514 That content would come from some content farm somewhere. 00:07:28.524 --> 00:07:30.214 Someone at BuzzFeed wrote a listicle 00:07:31.124 --> 00:07:31.584 Fraser Kelton: Mm hmm. 00:07:31.664 --> 00:07:31.824 Mm 00:07:31.904 --> 00:07:35.334 Nabeel Hyatt: about, about Brad Pitt's beanies or, or celebrity beanies, 00:07:35.574 --> 00:07:38.764 and then, and then Google scrapes that, and the economic bundle is 00:07:38.764 --> 00:07:42.784 obviously that I hopefully click on it sometimes, and then, and then 00:07:42.804 --> 00:07:44.574 BuzzFeed or whoever else gets a penny. 00:07:45.074 --> 00:07:49.894 This whole model , is going to completely break down because, one, Google, those 00:07:49.894 --> 00:07:55.644 searches were way worse on, on their website it was slow, it gave back only 00:07:55.644 --> 00:07:59.574 one answer, not all the answers, and Perplexity just did a very, very good 00:07:59.574 --> 00:08:03.629 job of I don't know what they're doing under the hood, but very, very good job 00:08:03.629 --> 00:08:09.099 of, I think, just doing rag across a deep corpus set of search results retrieving 00:08:09.109 --> 00:08:12.099 out the right things and then actually presenting them in a good format. 00:08:12.369 --> 00:08:15.269 Now, if I have to fast forward a couple of years, I don't think it'll even 00:08:15.279 --> 00:08:16.829 look like a search engine page, right? 00:08:17.129 --> 00:08:18.009 It should. 00:08:18.299 --> 00:08:22.049 Eventually look almost like that BuzzFeed page, just dynamically generated,? 00:08:22.109 --> 00:08:23.949 It will, it will be nicely laid out. 00:08:23.969 --> 00:08:25.429 The photos of the beanies will be on there. 00:08:25.429 --> 00:08:27.916 I'm using a really bad example, but these are the kinds of ridiculous 00:08:28.196 --> 00:08:30.220 things that people search for randomly on a Friday afternoon. 00:08:30.270 --> 00:08:30.600 Fraser Kelton: yeah, 00:08:31.000 --> 00:08:32.795 Nabeel Hyatt: And and I also like, one thing we talked about as 00:08:32.795 --> 00:08:35.453 well, I want to come back to is not all my searches go to Perplexity. 00:08:36.023 --> 00:08:39.333 And I think a partner, Yaz, brought up, and we were talking about the 00:08:39.333 --> 00:08:43.023 hallway, is there really going to be one search engine in the future? 00:08:43.033 --> 00:08:47.603 So do you differentiate what you go to chat GPT for right 00:08:47.603 --> 00:08:50.453 now, as an example, versus what you go to Google for right now? 00:08:51.224 --> 00:08:55.874 Fraser Kelton: yeah, when I want to learn something I go to ChatGPT, right? 00:08:55.874 --> 00:08:59.834 So a lot of our job is, is being curious and trying to understand things. 00:08:59.834 --> 00:09:04.824 I no longer go because Google makes you do both a lot of work to pull out the 00:09:04.824 --> 00:09:09.064 information as well as like people have spent, I don't know, two decades now 00:09:09.084 --> 00:09:10.994 trying to game the Google search results. 00:09:11.014 --> 00:09:14.054 So that has become, I don't know, this adversarial thing where you're trying 00:09:14.054 --> 00:09:15.644 to put in the work to learn how to. 00:09:15.834 --> 00:09:19.234 To get to the answers that you want, and there's a whole other 00:09:19.704 --> 00:09:23.904 side of the equation fighting to give you the junk on the SEO side. 00:09:24.304 --> 00:09:27.344 I used to have some hacks, everybody does, who's an early 00:09:27.344 --> 00:09:29.404 adopter of a pen site reddit. 00:09:29.404 --> 00:09:32.284 com, if you want to find out what the, the best thing is. 00:09:32.784 --> 00:09:35.574 And you can now ask ChatGPT and it gives it to you. 00:09:35.674 --> 00:09:39.784 Oh yeah probably the people that I have the most respect for as early adopters 00:09:39.784 --> 00:09:43.984 in terms of taste and product sense and like discerning input on these types 00:09:43.984 --> 00:09:46.937 of things have basically said the same thing as you about Perplexity , which, 00:09:47.086 --> 00:09:48.336 which really caught me off guard. 00:09:48.576 --> 00:09:51.974 For me, it's a creature of habit and then the power of defaults 00:09:51.974 --> 00:09:53.584 and distribution where I have 00:09:53.849 --> 00:09:54.269 Nabeel Hyatt: Of course. 00:09:54.354 --> 00:09:55.264 Fraser Kelton: actively fight that. 00:09:55.264 --> 00:09:56.884 And so I'm still finding myself going to. 00:09:57.404 --> 00:09:58.144 to Google. 00:09:58.404 --> 00:10:01.894 And then when it's an important enough search, this is, I mean, maybe the 00:10:01.894 --> 00:10:06.794 optimistic story, when it's an important enough search that Google lets me down, 00:10:06.814 --> 00:10:08.874 I find myself then going to perplexity. 00:10:09.044 --> 00:10:11.774 And how many times am I going to do that before I just go right to 00:10:11.774 --> 00:10:13.524 perplexity or something similar to that? 00:10:13.894 --> 00:10:17.454 To your question of when I go to Google, it's if I want to know how late the 00:10:17.454 --> 00:10:21.874 restaurant is open down the street, or I want to it's, it's all the stuff that 00:10:21.884 --> 00:10:23.654 you've mentioned earlier that are, are. 00:10:23.989 --> 00:10:28.689 Appendages to the original type of reasons that we went to Google, right? 00:10:28.689 --> 00:10:30.819 Is the one box type of stuff, 00:10:31.319 --> 00:10:31.939 Nabeel Hyatt: It is one box. 00:10:31.969 --> 00:10:37.869 I think Google as an internet search engine is terrible, it's 00:10:37.869 --> 00:10:41.289 not just because of AI and being able to ask questions of things. 00:10:41.289 --> 00:10:44.399 It's also just the degradation of content on the internet, which of course, 00:10:44.439 --> 00:10:47.069 when there's a bunch of AI generated content, will only degrade faster. 00:10:47.619 --> 00:10:51.409 I think I encapsulate that there's basically three searches, and maybe 00:10:51.409 --> 00:10:55.069 they end up being one product that rules them all eventually, but it's also 00:10:55.069 --> 00:10:58.159 possible sometimes there's bifurcation, and we see this happen in industries. 00:10:58.239 --> 00:11:00.109 And so maybe these end up being three different companies. 00:11:00.499 --> 00:11:03.529 The first one is, I want one box results. 00:11:03.659 --> 00:11:05.489 I know it's a one shot answer. 00:11:05.539 --> 00:11:06.759 It's probably eight words. 00:11:07.019 --> 00:11:07.999 You probably know it. 00:11:08.049 --> 00:11:09.049 I know you know it. 00:11:09.079 --> 00:11:10.079 Just, just tell me. 00:11:10.439 --> 00:11:14.069 And, and Google has done a very good job of largely ripping a bunch of Yelp and 00:11:14.069 --> 00:11:17.599 other people's content off turning it into beautiful little one box results. 00:11:17.899 --> 00:11:19.379 I still go to Google for that all the time. 00:11:19.499 --> 00:11:22.659 Second is inquiry, which you use the word learning. 00:11:22.769 --> 00:11:26.509 When I know that the first answer is probably wrong, or I will be confused 00:11:26.509 --> 00:11:30.889 by it, and I want to ask follow ups, then I go to GPT or Claude. 00:11:31.169 --> 00:11:35.549 And then the third is research, which is close to learn but different, 00:11:35.789 --> 00:11:39.089 where I just want an answer, but it's a big multivariate answer. 00:11:39.389 --> 00:11:42.699 That is, the research questions, which is obviously the, actually, 00:11:43.049 --> 00:11:47.119 Perplexity does an amazing job on, on the Reddit style questions. 00:11:47.259 --> 00:11:49.409 And, frankly, academic scholarly publications as well, 00:11:49.469 --> 00:11:50.339 it does a really good job. 00:11:51.339 --> 00:11:54.329 So for me research goes to Perplexity, OneBox goes to Google 00:11:54.669 --> 00:11:57.199 and, and inquiry goes to ChatGPT. 00:11:57.489 --> 00:11:58.049 But this is... 00:11:58.524 --> 00:12:01.624 This is early in the kind of muck of everybody messing around with this stuff, 00:12:02.004 --> 00:12:04.724 so I'm curious to see how it goes in a 00:12:04.934 --> 00:12:05.204 Fraser Kelton: Yeah. 00:12:05.204 --> 00:12:08.554 Somebody was telling me that the, the the pessimistic story for all of this 00:12:08.564 --> 00:12:14.024 is that Google and now OpenAI have so many resources and, and some semblance 00:12:14.024 --> 00:12:18.944 of either distribution or brand that they may not innovate on the right 00:12:18.944 --> 00:12:22.704 product experience, but they can just rip it off once somebody delivers 00:12:22.704 --> 00:12:25.394 something that's like clearly resonating. 00:12:25.724 --> 00:12:26.374 And we'll see. 00:12:26.544 --> 00:12:30.004 I think that that's what makes this time so fascinating to me is in 00:12:31.138 --> 00:12:35.048 2008, 2009, I think you and I would have had very similar views on how 00:12:35.408 --> 00:12:40.178 the mobile landscape was going to play out between Android and iOS. 00:12:40.398 --> 00:12:42.668 Maybe we'd get the relative share a little bit wrong. 00:12:43.008 --> 00:12:47.248 I think in this case, nobody knows anything and there's a lot of big Big 00:12:47.588 --> 00:12:50.848 markets and products that are up for grabs for the first time in a long time. 00:12:51.848 --> 00:12:56.048 Nabeel Hyatt: I mean, this is true, despite what Sam Lessin says, that I 00:12:56.058 --> 00:12:57.708 think there actually will be an AI market. 00:12:58.268 --> 00:12:59.598 Fraser Kelton: You told me you were worked up. 00:12:59.598 --> 00:13:02.898 I haven't, I haven't read anything that it has to say. 00:13:02.898 --> 00:13:07.298 There's some semblance of 37 signals content marketing going on here that 00:13:07.298 --> 00:13:10.008 I just can kind of like push a little bit into the corner, but what's up? 00:13:10.883 --> 00:13:13.173 Nabeel Hyatt: no, I like, he's, he's, okay, yes, of course, he's a little 00:13:13.173 --> 00:13:16.513 provocative, but I think these are long held beliefs not just hot takes. 00:13:16.783 --> 00:13:21.023 Sam has a podcast with three other people, which is actually, I think, the best new 00:13:21.033 --> 00:13:23.533 podcast of the last year focused on tech. 00:13:23.793 --> 00:13:28.023 Less is more, because specifically, it's four humans that actually like each 00:13:28.023 --> 00:13:29.243 other and actually know each other. 00:13:29.243 --> 00:13:31.483 And so there's real authenticity discussion as they talk about a bunch of 00:13:31.483 --> 00:13:32.803 things, which is not most shows, right? 00:13:33.313 --> 00:13:34.823 Most shows are canned content marketing. 00:13:34.923 --> 00:13:38.133 A anyway So look, there's a first part of what Sam's rant is and he has 00:13:38.133 --> 00:13:42.543 a rant on his show and then he just released like a 975 page, probably 00:13:42.543 --> 00:13:47.153 generated by AI, PowerPoint deck that kind of like regales how seed is 00:13:47.153 --> 00:13:49.973 dead, AI is just a weird pipe dream. 00:13:50.283 --> 00:13:53.283 Trying to save VC, it's all going to crap, and everyone should just 00:13:53.283 --> 00:13:54.743 start cottage software businesses. 00:13:55.013 --> 00:14:00.713 That's ostensibly the long story short of Sam's pitch about the evolving ecosystem. 00:14:00.983 --> 00:14:05.463 I, there's a part of it that, first, I agree with, which is that he has a 00:14:05.623 --> 00:14:11.893 framing that the kind of industrialization of venture capital phase is done. 00:14:12.333 --> 00:14:17.983 That, that at this point, the factory farmed B2B SaaS company is over. 00:14:18.333 --> 00:14:21.253 And that he sees is that causing a bunch of late stage. 00:14:22.836 --> 00:14:27.356 Crowd into seed spray and pray, they're not really working with those companies. 00:14:27.356 --> 00:14:29.596 Just yeah, you know that some junior partner has been 00:14:29.596 --> 00:14:30.566 told to put money to work. 00:14:30.736 --> 00:14:34.166 So the jamming money into random seed companies and it's messing everything up. 00:14:34.166 --> 00:14:36.781 And so if you're a seed or series a fund oh, wow, you got 00:14:36.781 --> 00:14:37.901 a real, you got a real problem.. 00:14:37.951 --> 00:14:39.511 So, I believe that. 00:14:39.611 --> 00:14:43.691 But the second bit where he gets to AI, I think, he's just anchored in his 00:14:43.691 --> 00:14:47.771 priors before AI, where it just felt like there was not enough innovation. 00:14:48.261 --> 00:14:52.111 So, his feeling, I could like, I hope I don't boil it down the wrong way, is 00:14:52.111 --> 00:14:57.711 that fundamentally AI is a sustaining innovation, not a disruptive innovation, 00:14:57.711 --> 00:15:02.531 so it'll change the world, but it'll just be that Google adds AI, and 00:15:02.541 --> 00:15:04.491 Facebook adds AI, and it'll be fine. 00:15:04.831 --> 00:15:08.381 If you're Adobe, you win anyway, you just add AI, and that kind of thing. 00:15:08.681 --> 00:15:12.771 And that there's no moat, because it's all, open source anyway so 00:15:12.771 --> 00:15:14.751 what's the moat you're going to have and there's no network effects. 00:15:15.011 --> 00:15:17.761 And that the tech is moving too fast for, for startups to keep up. 00:15:18.181 --> 00:15:22.661 I have a couple of like pretty basic flawed issues with this. 00:15:23.661 --> 00:15:27.741 Fraser Kelton: If the tech is moving so fast, isn't that the opportune time for 00:15:27.971 --> 00:15:28.751 Nabeel Hyatt: There you go! 00:15:29.316 --> 00:15:32.666 Fraser Kelton: Arvind is a 25 person team, I think, and he's out 00:15:32.666 --> 00:15:35.476 executing like everybody else who has. 00:15:35.701 --> 00:15:38.851 tried to push into this space and, and there's something 00:15:38.851 --> 00:15:40.461 remarkable about what he's doing. 00:15:40.471 --> 00:15:45.381 He, he's shipping both at the technical level inference that is like 2. 00:15:45.381 --> 00:15:48.601 5 times faster than, than other inference platforms that are 00:15:48.601 --> 00:15:50.141 specifically focused on that. 00:15:50.591 --> 00:15:53.651 And then his product decisions are like tasteful and thoughtful, and 00:15:53.651 --> 00:15:58.121 he is not having to fight against legacy business models and an entire 00:15:58.121 --> 00:16:02.316 org chart that has calcified around an an old way of doing things. 00:16:02.316 --> 00:16:04.206 So anyway, you, you teed me up, but 00:16:04.206 --> 00:16:05.116 Nabeel Hyatt: Yeah, no, exactly. 00:16:05.856 --> 00:16:08.706 I just fundamentally disagree that big tech is going to dominate. 00:16:08.706 --> 00:16:11.210 One of the things I love about what's happening in AI right now is that 00:16:11.210 --> 00:16:13.639 we talk to lots of investors that are quote unquote active in AI. 00:16:14.239 --> 00:16:17.149 And they are activating wildly different ideas about what's 00:16:17.149 --> 00:16:18.239 going to work than we are. 00:16:18.269 --> 00:16:20.039 And Sam's going to operate as part of the ecosystem. 00:16:20.269 --> 00:16:22.559 That is ultimately a good thing for founders, right? 00:16:22.599 --> 00:16:25.459 We're going to get a spread of different capital to try a 00:16:25.459 --> 00:16:26.609 bunch of different experiments. 00:16:26.819 --> 00:16:28.569 We'll let it play out over time. 00:16:28.789 --> 00:16:31.559 But I do think, yeah, there are two flaws in Sam's argument. 00:16:31.609 --> 00:16:35.109 So one is like he overestimates the big tech's ability to execute, right? 00:16:35.139 --> 00:16:36.359 Most large companies. 00:16:36.794 --> 00:16:38.184 A struggle to adopt new tech. 00:16:38.284 --> 00:16:41.814 And startups can find ways to sell against Google and Microsoft. 00:16:41.814 --> 00:16:44.594 And frankly, one of my little examples, there was a world where 00:16:44.594 --> 00:16:46.724 Microsoft dominated all of GovTech. 00:16:47.084 --> 00:16:49.034 Big, entrenched sales. 00:16:49.354 --> 00:16:55.134 And Google, when it was still a relatively early company, actually used cloud and 00:16:55.134 --> 00:16:58.654 pushed into that, into that dominance in a way that I think, frankly, beforehand 00:16:58.654 --> 00:16:59.694 you would have said would never work. 00:17:00.354 --> 00:17:03.024 And there's lots and lots and lots of examples of startups 00:17:03.034 --> 00:17:07.294 using The disadvantage of a large company to their benefit, right? 00:17:07.584 --> 00:17:12.224 The second is that I think Sam, as a lot of that entire generation of 00:17:12.224 --> 00:17:16.344 founders, grew up in a world where network effects were the biggest moat. 00:17:17.004 --> 00:17:20.894 And so I think there's an anchoring idea that isn't that the moat? 00:17:21.429 --> 00:17:26.249 And, and so, if you grew up in that world, you look at Facebook, and look, 00:17:26.439 --> 00:17:30.389 network effects are an incredible, incredible barrier to entry, but 00:17:30.799 --> 00:17:33.693 at the same time, it's not the only way startups build durability. 00:17:33.773 --> 00:17:36.135 There were many startups with no network effects. 00:17:36.402 --> 00:17:39.432 Frankly, Google Search versus Microsoft that we were talking about 00:17:39.432 --> 00:17:41.382 earlier is an exact example of this. 00:17:41.432 --> 00:17:45.562 Vertical integrations like Microsoft, working from the OSS layer up to the 00:17:45.567 --> 00:17:49.376 office layer and launching Microsoft Office . Apple . Cisco . Cruise 00:17:49.476 --> 00:17:52.385 . There's lots of other ways other than network effects for a startup to 00:17:52.385 --> 00:17:53.945 build competitive barriers to entry. 00:17:54.135 --> 00:17:58.235 I think he's also over indexing on large companies that are innovative in AI. 00:17:58.605 --> 00:18:01.745 And forgetting that AI is going to transform every industry 00:18:02.095 --> 00:18:03.355 beyond big tech, right? 00:18:03.575 --> 00:18:06.915 Manufacturing and healthcare, and it's going to be a 20 year run 00:18:06.915 --> 00:18:08.965 just like it was for the internet. 00:18:09.310 --> 00:18:11.690 It's going to go hit every little part of the economy. 00:18:11.880 --> 00:18:14.680 Lots of those parts of the economy are going to deeply struggle 00:18:15.000 --> 00:18:16.680 to come up with a solution. 00:18:16.920 --> 00:18:19.130 And there, there's no incumbent advantage. 00:18:19.150 --> 00:18:21.640 Yes, some big company could build a solution for that area, 00:18:21.720 --> 00:18:22.920 but also so could a startup. 00:18:23.060 --> 00:18:25.990 And so it's all the normal things that happen with startups, right? 00:18:26.590 --> 00:18:30.420 Fraser Kelton: Yeah, yeah, listen, I'm a Satya Stan and, and that lovely line 00:18:30.420 --> 00:18:34.970 that he made Google Dance was, like, so fun, but come on it wasn't Bing and 00:18:34.980 --> 00:18:39.820 Sydney that made Google Dance, I had, it's just a really bad, crummy product, and 00:18:39.820 --> 00:18:44.100 everybody knows it's a bad, crummy product It is open AI and, and what, what was 00:18:44.100 --> 00:18:46.380 shipped there that is making Google dance. 00:18:46.380 --> 00:18:51.290 And we just talked earlier about how like their entire business model is going to 00:18:51.295 --> 00:18:55.880 have to be reconsidered for a world where you can write a poem about tulips and 00:18:55.880 --> 00:18:57.800 then have search results put below it. 00:18:57.800 --> 00:19:00.500 Like they, they have real fundamental problems that 00:19:00.500 --> 00:19:01.580 they're gonna have to navigate. 00:19:01.585 --> 00:19:04.005 Nabeel Hyatt: I think it's very easy to underestimate how hard 00:19:04.005 --> 00:19:06.835 it is for incumbents to change and trench workflows, right? 00:19:06.875 --> 00:19:10.515 It's, it's easier for startups to start from scratch. 00:19:10.955 --> 00:19:14.705 And there's an inherent conflict, and you called it out, very first thing, between 00:19:14.935 --> 00:19:17.985 these two ideas, that, that this is a... 00:19:18.500 --> 00:19:23.220 Sustaining innovation, so companies are just going to add it, and that 00:19:23.220 --> 00:19:24.800 it's moving unbelievably fast. 00:19:25.290 --> 00:19:27.970 Because it's not just moving unbelievably fast, and this is a thesis to how 00:19:27.970 --> 00:19:30.230 we, how we are navigating the world right now with founders, which is, 00:19:30.510 --> 00:19:34.450 it's not just moving very fast at the foundational level layer, the model 00:19:34.450 --> 00:19:37.690 layer, where there's new capabilities coming out constantly, it's also moving 00:19:37.690 --> 00:19:40.870 really fast at the user expectation layer, right, at that top layer. 00:19:41.255 --> 00:19:44.839 And so the, the, that, the last revolution in AI 15 years ago, we 00:19:44.839 --> 00:19:47.863 actually, as a firm, we didn't go that deep into it . We had Cruise 00:19:47.863 --> 00:19:49.383 and a couple of other investments. 00:19:49.633 --> 00:19:53.883 But part of the problem in that age of small models is that folks with the 00:19:53.883 --> 00:19:56.183 data and the distribution win, right? 00:19:56.223 --> 00:19:58.883 If you have all the data and you're Google and you got a distribution, 00:19:58.933 --> 00:19:59.743 you're just going to win. 00:20:00.503 --> 00:20:04.393 What feels different here is, one, the data, everybody has access to 00:20:04.423 --> 00:20:06.877 the data, with a few big exceptions . So there's a data advantage. 00:20:06.997 --> 00:20:09.477 And then the third D , if I want to keep this little D thing going, 00:20:09.477 --> 00:20:10.987 is is the design is changing, right? 00:20:10.997 --> 00:20:16.447 Like the, the kind of concept of a product like Descript is that you used AI to 00:20:16.447 --> 00:20:19.957 completely change the workflow on the front end of how somebody actually makes 00:20:19.957 --> 00:20:23.667 a podcast or edit a video in a way that was never true for the last 20 years. 00:20:24.037 --> 00:20:25.607 When the workflow and design changes. 00:20:26.042 --> 00:20:30.262 Then the incumbents, they can't completely change their workflows 00:20:30.282 --> 00:20:33.742 without alienating their current base, so they have incumbency disadvantage. 00:20:34.152 --> 00:20:38.492 And you see this with Adobe, they didn't, their, their launch of a copycat 00:20:38.492 --> 00:20:41.572 product to Descript was Adobe Podcast. 00:20:41.642 --> 00:20:45.402 Now, they didn't change a previous product, they had to start from scratch. 00:20:45.662 --> 00:20:49.682 They built a new UI, they launched a copy, it's pretty good, it's just 00:20:49.712 --> 00:20:51.122 getting started, we'll see what happens. 00:20:51.122 --> 00:20:54.912 They obviously have money and brand to go push it out into the market. 00:20:54.912 --> 00:20:57.902 They're going to try bundling, all the things that large companies do. 00:20:58.172 --> 00:21:01.222 But that's no different than how large companies have tried to compete with 00:21:01.222 --> 00:21:03.592 startups for the last 30 years and 00:21:03.592 --> 00:21:04.072 startups have been 00:21:04.132 --> 00:21:05.372 Fraser Kelton: 30 plus years, sure. 00:21:05.852 --> 00:21:11.812 Yep, the I think if you look at the past 12 months, you have seen incumbents 00:21:11.822 --> 00:21:13.992 move very quickly into this space. 00:21:14.042 --> 00:21:15.352 The whole world has, right? 00:21:15.382 --> 00:21:21.441 And I think part of it is, we have a short term overestimating the impact, and then 00:21:21.441 --> 00:21:24.348 long term we underestimate, and there's going to be a very profound change. 00:21:24.698 --> 00:21:28.898 But if you look at what the incumbents are doing, they're layering in Cheap 00:21:28.908 --> 00:21:31.228 ways to leverage this technology, right? 00:21:31.278 --> 00:21:34.028 Hey, listen, we can now do some sort of like really amazing 00:21:34.038 --> 00:21:36.298 summarization around unstructured data. 00:21:36.308 --> 00:21:40.238 And our app is unstructured data because it's recording sales calls. 00:21:40.518 --> 00:21:43.808 And that is incrementally more useful for that product. 00:21:44.168 --> 00:21:47.398 And it, it's pretty awesome that they moved so quickly to introduce these 00:21:47.398 --> 00:21:48.868 types of features of functionality. 00:21:49.248 --> 00:21:53.188 But I think that It's an absence of imagination to think that there's 00:21:53.188 --> 00:21:57.838 not going to be entirely new user experiences that are delivered off of 00:21:57.838 --> 00:22:01.378 this technology, especially because the technology is strictly only 00:22:01.378 --> 00:22:03.078 going to get better from here on out. 00:22:03.458 --> 00:22:06.388 And, you said this, I'm stealing your line, is I don't even think we've seen 00:22:06.388 --> 00:22:09.554 from the artists and creatives how to use this technology appropriately 00:22:09.554 --> 00:22:11.084 yet, and I think that that's fair. 00:22:11.094 --> 00:22:13.344 Nabeel Hyatt: If you think the problem is a puzzle, then large 00:22:13.344 --> 00:22:14.784 companies have a huge advantage. 00:22:15.084 --> 00:22:17.514 Puzzles can get solved by throwing money at the problem. 00:22:18.204 --> 00:22:20.754 But if the problem is a mystery, Founders have a chance. 00:22:20.754 --> 00:22:23.154 to win because agility, wins and you need judgment. 00:22:23.704 --> 00:22:24.154 Something that. 00:22:24.274 --> 00:22:25.714 Large companies have problem with. 00:22:26.279 --> 00:22:26.589 Fraser Kelton: Yep. 00:22:26.689 --> 00:22:27.659 I very much agree. 00:22:27.659 --> 00:22:31.829 Like the way that I think about the world is there's this continuum of 00:22:31.939 --> 00:22:37.969 project to product and on the project end is doing things right and on the 00:22:37.979 --> 00:22:41.019 product end is, Doing the right things. 00:22:41.509 --> 00:22:46.639 And I feel like the jigsaw metaphor is like slide it all the way over to 00:22:46.639 --> 00:22:49.249 project where you're doing things right. 00:22:49.425 --> 00:22:52.216 Nabeel Hyatt: Right, so this is a puzzles versus mystery situation, and 00:22:52.216 --> 00:22:55.786 there will be parts of the market, absolutely, which feel like puzzles. 00:22:56.176 --> 00:22:59.576 Let me just add a chatbot client, let me get the retrieval properly, it'll 00:22:59.576 --> 00:23:01.326 be added to my product, and we're done. 00:23:01.816 --> 00:23:05.406 And then the mystery is places where this technology will lead 00:23:05.726 --> 00:23:09.886 to new types of interfaces, and new types of experiences, that... 00:23:10.516 --> 00:23:15.016 Don't sit anywhere comfortably where an incumbent has previously been. 00:23:15.266 --> 00:23:16.536 That's where startups have the advantage. 00:23:17.382 --> 00:23:17.802 Fraser Kelton: And you're right. 00:23:17.812 --> 00:23:21.765 If you just need to have a 12 month known roadmap 00:23:21.765 --> 00:23:26.013 For shipping AI, then let's coordinate all your cross functional stakeholders 00:23:26.013 --> 00:23:29.023 and make sure that they're bought in and, and, you check the boxes. 00:23:29.693 --> 00:23:36.353 And if it's figuring out the, the right thing, I think we're so early in, in 00:23:36.383 --> 00:23:40.873 understanding the types of experiences that this technology can provide for us. 00:23:40.973 --> 00:23:43.453 And, and we're, we're just getting started. 00:23:43.453 --> 00:23:46.686 Like I think as you said, again, like the creatives haven't shown 00:23:46.686 --> 00:23:48.406 us what's possible just yet 00:23:48.956 --> 00:23:49.616 Nabeel Hyatt: Yeah, that's right. 00:23:49.716 --> 00:23:50.016 That's right. 00:23:50.336 --> 00:23:52.196 I'm, I'm going to throw to you, actually, on the spot. 00:23:52.196 --> 00:23:54.526 I know you're not prepared at all for this, but ... We just went through 00:23:54.526 --> 00:23:57.446 a process where we were talking to a bunch of limited partners. 00:23:57.496 --> 00:24:02.026 So for those who don't know venture capital, we actually have to raise 00:24:02.026 --> 00:24:03.536 money just like founders raise money. 00:24:03.666 --> 00:24:07.676 And largely from non profits, endowments pension funds, places like that. 00:24:07.686 --> 00:24:10.666 So when we make money, that's where the money goes to 00:24:10.786 --> 00:24:12.246 those kinds of organizations. 00:24:12.706 --> 00:24:15.846 Was there anything about that process of talking to some LPs over 00:24:15.846 --> 00:24:19.326 the course of the AGM in this last week that caught you by surprise? 00:24:19.826 --> 00:24:20.939 Fraser Kelton: It's surprising how much... 00:24:20.939 --> 00:24:25.948 Felt very similar to being a founder raising capital from VCs and then 00:24:25.948 --> 00:24:27.988 how some of it felt so different. 00:24:28.088 --> 00:24:32.018 You and I sat through a whole host of different meetings and there 00:24:32.018 --> 00:24:36.808 was meetings where the vibe just was great from the hello, right? 00:24:36.933 --> 00:24:37.823 Nabeel Hyatt: Yeah, yeah. 00:24:37.848 --> 00:24:39.158 Fraser Kelton: in and it feels. 00:24:39.528 --> 00:24:40.518 It just feels great. 00:24:40.768 --> 00:24:42.458 The entire experience feels great. 00:24:42.558 --> 00:24:45.158 And then there's some where you sit down and you're like, Well, the vibe 00:24:45.158 --> 00:24:48.248 here is not quite the same as the vibe that just was in the other meeting. 00:24:48.258 --> 00:24:50.328 I'm not sure how this is going to turn out. 00:24:50.848 --> 00:24:54.398 The other thing that is very similar is... 00:24:54.983 --> 00:24:59.583 You just get into a groove, but you have to have some feeling of spontaneity in 00:24:59.583 --> 00:25:04.233 the discussions, and when, when I was pitching VCs, you'd get to the point 00:25:04.233 --> 00:25:08.543 with the deck where you would have a pause and you knew, you knew nine times 00:25:08.543 --> 00:25:11.523 out of ten they're going to lean in with this question, and you didn't. 00:25:11.708 --> 00:25:14.548 You didn't, you didn't pitch it, so that you could get ahead of the question. 00:25:14.548 --> 00:25:17.338 You pitched it so that they would ask the question because you knew the answer 00:25:17.338 --> 00:25:19.378 to give, but it had to feel spontaneous. 00:25:19.748 --> 00:25:21.268 Nabeel Hyatt: There's a theater to all of it, right? 00:25:21.268 --> 00:25:26.248 There's a theater to pitching, and I do think for an engineering mindset 00:25:26.268 --> 00:25:28.928 when they're going out to raise money, a lot of times the answer is, let me 00:25:28.928 --> 00:25:30.858 answer all the questions in the deck. 00:25:31.363 --> 00:25:33.503 And let me get it in order, and somebody asked that question 00:25:33.503 --> 00:25:34.563 there, I got a slide for that. 00:25:34.733 --> 00:25:38.193 I used to do a thing, it's hard now in Zoom land, but I used to do a 00:25:38.193 --> 00:25:41.313 thing when I was pitching, where I would take one question that I knew 00:25:41.313 --> 00:25:45.913 was just an absolutely plump T up, and and I would remove the slide, 00:25:46.303 --> 00:25:49.983 and then have them ask the question, and I'd be like, that's interesting. 00:25:50.083 --> 00:25:51.203 Do you got a whiteboard marker? 00:25:51.593 --> 00:25:55.223 And, and the act of you just, you get up and you walk over to the 00:25:55.223 --> 00:25:59.473 whiteboard and you're sketching and talking in real time and man it 00:25:59.473 --> 00:26:01.063 just, it pulls the room together. 00:26:01.083 --> 00:26:03.903 Fraser Kelton: The place where it was surprising and different, and it 00:26:04.043 --> 00:26:07.873 continues to be surprising for me is that we can spend our day talking to 00:26:07.903 --> 00:26:12.443 founders who are trying to create a future that may or may not take shape 00:26:12.453 --> 00:26:14.543 in two years, three years, four years. 00:26:14.573 --> 00:26:18.063 I mean, with Cruise, it was eight years of Kyle grinding, right? 00:26:18.123 --> 00:26:18.143 Yeah. 00:26:18.143 --> 00:26:18.473 Yeah. 00:26:18.743 --> 00:26:22.413 And then we step into a world where we have to talk about portfolio construction 00:26:22.443 --> 00:26:27.833 and it's just very strange to think that the entire world works because 00:26:27.833 --> 00:26:29.213 of portfolio construction, right? 00:26:29.213 --> 00:26:32.733 And the risk that you take on so that these, these outliers can actually 00:26:32.823 --> 00:26:34.363 have oxygen and come to life. 00:26:35.363 --> 00:26:38.943 Nabeel Hyatt: Well, I mean, I do think that for the folks who, who came into 00:26:38.943 --> 00:26:42.953 venture in the last four or five years, they're learning a really key lesson 00:26:42.953 --> 00:26:45.923 right now, and they have learned a really key lesson over the last year , there's 00:26:45.923 --> 00:26:48.383 a reason why it, it actually works. 00:26:48.433 --> 00:26:52.478 I'm a person who like, it's like what Winston Churchill said about democracy., 00:26:52.498 --> 00:26:54.098 Venture is a very flawed model. 00:26:54.128 --> 00:26:57.818 It's just that it's the best model that we have for sustaining innovation. 00:26:57.848 --> 00:27:01.608 I mean, the other alternatives, if you look at things like corporate R& D 00:27:01.928 --> 00:27:04.078 and how many wonderful amazing things come out of corporate R& D, you're 00:27:04.078 --> 00:27:05.528 like, that's not the way to do it. 00:27:05.688 --> 00:27:06.578 Okay, so what else do we do? 00:27:06.658 --> 00:27:08.988 Do you want the EU to figure out what we should build next? 00:27:08.988 --> 00:27:10.548 That's not the right way to do it. 00:27:10.748 --> 00:27:16.768 Like almost all the other models for how We get an ecosystem to sit on the right 00:27:16.818 --> 00:27:19.938 edge of risk and innovate and create what's going to be next in the world. 00:27:20.188 --> 00:27:24.078 I actually am a very big booster for this particular model. 00:27:24.108 --> 00:27:27.238 Like it, it takes builders and it takes VCs and so on and so forth. 00:27:27.518 --> 00:27:32.478 I also hate the industrialization of venture capital because it, it feels 00:27:32.543 --> 00:27:33.223 Fraser Kelton: just different. 00:27:33.973 --> 00:27:34.503 Yep. 00:27:34.558 --> 00:27:38.238 Nabeel Hyatt: it's not just different, , I think it ruins parts of that model. 00:27:38.258 --> 00:27:41.728 It messes with parts of that model that I think are becoming more obvious now that 00:27:41.728 --> 00:27:44.738 we're out of the, B2B sass go go times, 00:27:45.273 --> 00:27:45.603 Fraser Kelton: Yeah. 00:27:45.603 --> 00:27:47.207 SASSification of everything. 00:27:48.057 --> 00:27:51.446 On a different topic, what, AI product have you slowed down your 00:27:51.446 --> 00:27:53.176 life to explore this past week? 00:27:54.176 --> 00:27:54.776 It's not even 00:27:54.776 --> 00:27:55.546 Nabeel Hyatt: slowed down. 00:27:55.916 --> 00:27:58.366 I have been, so I listened to a lot of podcasts. 00:27:58.956 --> 00:28:00.586 I basically have. 00:28:00.752 --> 00:28:04.922 The habit of anytime I have any downtime, unless it's specifically like 00:28:04.932 --> 00:28:08.422 thinking time I I'm doing the dishes. 00:28:08.422 --> 00:28:10.382 I'm going outside to take the dog for a walk. 00:28:10.382 --> 00:28:12.372 I have a kind of running list of podcasts going. 00:28:12.602 --> 00:28:17.412 I've been using Marco Armitage Overcast for many, years. 00:28:18.082 --> 00:28:20.726 One of my best go to examples of everything doesn't 00:28:20.726 --> 00:28:21.746 have to be venture scale. 00:28:21.956 --> 00:28:24.776 Like sometimes the most amazing product is a one person effort and it's a 00:28:24.776 --> 00:28:27.236 beautiful thing and I love that product and there's a lot of love in it. 00:28:28.186 --> 00:28:32.126 I certainly use it over say I don't know, like Spotify is trying to 00:28:32.126 --> 00:28:33.986 mix in my podcast with my music. 00:28:34.286 --> 00:28:36.626 Stupidness or Apple Podcasts. 00:28:36.631 --> 00:28:39.721 For the first time, and I don't know how long it's been, 5, 6, 7, 8 years I, don't 00:28:39.721 --> 00:28:41.521 use Overcast anymore, so I switched. 00:28:42.381 --> 00:28:44.851 To a product called Snipt, S N I P T. 00:28:45.851 --> 00:28:51.441 It is what an amazing difference it makes . Obviously, it's not 00:28:51.451 --> 00:28:53.211 changing how you listen to podcasts. 00:28:53.411 --> 00:28:54.851 It is still a podcast player. 00:28:55.021 --> 00:29:00.141 It's probably lower on features than than a product like Overcast, 00:29:00.141 --> 00:29:01.371 which has had many years. 00:29:01.676 --> 00:29:04.296 But that's what you want with technological innovation, right? 00:29:04.296 --> 00:29:08.906 Is that a startup can start not without having even hit table stakes for the 00:29:08.906 --> 00:29:12.456 market, but there's some feature that's so good that it drives your behavior. 00:29:12.666 --> 00:29:15.426 The thing in particular that I always want is you're listening to 00:29:15.426 --> 00:29:18.596 a podcast, I'm in the car, somebody says something you want to remember. 00:29:19.261 --> 00:29:24.261 And the feature I've always wanted is like, can you just please just clip? 00:29:24.331 --> 00:29:28.131 I want my DVR TiVo feature for my podcast. 00:29:28.141 --> 00:29:31.791 Like just clip the last 30 seconds and do something smart with it so that 00:29:31.811 --> 00:29:34.951 I don't lose these ideas as they're coming, as I'm coming across them. 00:29:35.791 --> 00:29:38.331 I think this company has been around for a little while now, but without 00:29:38.371 --> 00:29:39.791 all the AI features that they had. 00:29:39.791 --> 00:29:43.011 So I think it used to be a, I can triple click my earbuds. 00:29:43.781 --> 00:29:46.621 And it will grab the last little 30 seconds or something like that. 00:29:46.621 --> 00:29:48.301 I think it was the previous version of this product. 00:29:48.321 --> 00:29:50.981 I did, I think I tried this a year ago or more. 00:29:51.101 --> 00:29:51.751 It's fine. 00:29:52.101 --> 00:29:53.111 It's just not good enough. 00:29:53.631 --> 00:29:57.431 Now what it's doing is a bunch of incredibly wonderful things. 00:29:57.731 --> 00:29:59.831 First of all, as I'm scrolling through. 00:30:00.091 --> 00:30:00.621 podcast. 00:30:01.031 --> 00:30:05.121 It's using AI to do summarization of the whole podcast. 00:30:05.431 --> 00:30:09.091 So instead of it being the thing that you or I would write about this podcast, 00:30:09.101 --> 00:30:12.021 which may or may not be accurate, it's actually looking at the transcript 00:30:12.021 --> 00:30:15.091 and giving me a real summarization of the topics that are going on in there. 00:30:15.141 --> 00:30:15.841 Very interesting. 00:30:16.175 --> 00:30:19.178 It then breaks up the podcast dynamically into chapter clips. 00:30:19.318 --> 00:30:22.068 So if I don't like a topic that we're talking about right now, like if I 00:30:22.068 --> 00:30:22.918 don't want to hear about Snipped. 00:30:23.918 --> 00:30:26.418 I can double click on my ear pods and it will just go to 00:30:26.418 --> 00:30:28.638 the next chapter dynamically. 00:30:28.678 --> 00:30:29.358 That's awesome. 00:30:29.408 --> 00:30:30.218 That is awesome. 00:30:31.108 --> 00:30:34.248 Which is just like, it just changes the experience, because we all know... 00:30:34.683 --> 00:30:37.373 Frankly, including ours, all these podcasts go a little long. 00:30:38.373 --> 00:30:42.083 And part of it is, that is the joy of the podcast, is that it feels like 00:30:42.083 --> 00:30:44.563 you're in the room with somebody, you're listening to them go through a topic, 00:30:44.773 --> 00:30:47.193 you can get depth, but sometimes you just don't want to hear about something. 00:30:47.573 --> 00:30:50.683 And so the DoubleClick Next Chapter, even if they haven't set up, and it's 00:30:50.683 --> 00:30:55.583 not, it's chapter rankings on very finite, like two, three minute scales. 00:30:55.653 --> 00:30:58.473 It's not like 20 minute scales, which is what some podcasts do. 00:30:58.483 --> 00:30:59.673 And I just love that. 00:31:00.083 --> 00:31:04.748 And then the last feature, AI, Native feature, which is really wonderful, 00:31:04.748 --> 00:31:09.338 is now when you triple click, it is smart about how much it clips, so 00:31:09.338 --> 00:31:13.118 it clips only the topic area, which could be in the future as well. 00:31:13.618 --> 00:31:16.708 It clips the topic area that is related to the thing that you just triple 00:31:17.088 --> 00:31:21.388 clicked on, so somewhere, between 15 seconds and like a minute and a half 00:31:21.818 --> 00:31:25.178 is what it's grabbing to try and grab the concept, which is just brilliant. 00:31:25.628 --> 00:31:28.758 And then it transcribes it into text. 00:31:29.518 --> 00:31:33.578 So you have a transcription, and then I have that now plugged into... 00:31:33.913 --> 00:31:37.070 It has an API call that then plugs this into ReadWise , which 00:31:37.070 --> 00:31:41.870 is a kind of like note taking for podcasts and Kindle reading app. 00:31:42.260 --> 00:31:46.680 And then I have that, through an API, plugged into my note taking app. 00:31:47.180 --> 00:31:51.540 So that basically, every single time, I triple click. 00:31:51.540 --> 00:31:53.950 And I'm not going to talk about my note taking app, because I love my note 00:31:53.960 --> 00:31:56.800 taking app, but we gotta save that, you can't blow it all in one week. 00:31:57.800 --> 00:31:59.290 But I, triple click. 00:31:59.615 --> 00:32:02.795 And basically it now becomes a pipeline where that becomes a 00:32:02.835 --> 00:32:05.375 permanent note for reference later on. 00:32:05.805 --> 00:32:06.685 What the heck 00:32:06.686 --> 00:32:07.865 Fraser Kelton: do you do with that later? 00:32:08.355 --> 00:32:10.475 That's for the note taking app conversation. 00:32:10.475 --> 00:32:13.455 But I'm going back to our conversation last week that 99 00:32:13.455 --> 00:32:16.355 percent of our life is mundane. 00:32:16.590 --> 00:32:20.277 I used to, tag things on Delicious religiously, and I still find myself going 00:32:20.277 --> 00:32:22.137 back to that archive every now and then. 00:32:22.137 --> 00:32:27.117 And so there is some value there, but I haven't tagged anything in well over 00:32:27.117 --> 00:32:31.587 a decade, so I don't know what that says about the stickiness and long term 00:32:31.587 --> 00:32:33.107 value that I actually derive from it. 00:32:33.527 --> 00:32:33.887 Nabeel Hyatt: Anyway. 00:32:34.067 --> 00:32:37.837 Well, I like, look, I like when people say things that are insightful. 00:32:37.857 --> 00:32:42.145 Like a good example is I used to reference of puzzles versus mysteries. 00:32:42.495 --> 00:32:44.805 And that has been a guiding thing for me for a little while. 00:32:44.995 --> 00:32:46.835 I came from listening to a podcast years ago, right? 00:32:46.835 --> 00:32:49.295 It's when you, it's when you hear something insightful and you want to 00:32:49.295 --> 00:32:52.595 make sure you don't lose it so you can internalize it a little bit more. 00:32:52.970 --> 00:32:56.680 What I really want now is to feed into a spaced referential learning product. 00:32:56.680 --> 00:32:58.680 But maybe that's just a little bit too nerdy for most people, 00:32:59.650 --> 00:33:00.610 but the pod, but snip is 00:33:00.610 --> 00:33:00.860 Fraser Kelton: great. 00:33:01.030 --> 00:33:05.980 So yeah I'm a fellow Overcast user, and I think that Marco's just got great taste. 00:33:05.980 --> 00:33:07.300 It's an opinionated product. 00:33:07.300 --> 00:33:10.350 There's a bunch of different flourishes that have really 00:33:10.350 --> 00:33:11.710 resonated with me over the years. 00:33:11.920 --> 00:33:16.920 What do you miss in, in Snipped that Overcast gave you? 00:33:17.445 --> 00:33:20.495 Nabeel Hyatt: I'm opening up Overcast in front of me right now to look through. 00:33:20.635 --> 00:33:27.185 So he does a really good job of, having custom clipping at the 00:33:27.185 --> 00:33:28.425 beginning and the end of podcasts. 00:33:28.995 --> 00:33:30.875 So some of these podcasts start with. 00:33:31.475 --> 00:33:34.465 A minute and a half of ads in the beginning, or a lot of times 00:33:34.465 --> 00:33:36.985 I find like the last minute of the outro is always useless. 00:33:37.535 --> 00:33:40.875 Like they just go bloviate for, I lose a minute and a half of my life every 00:33:40.875 --> 00:33:42.195 time I listen to a podcast at the end. 00:33:42.555 --> 00:33:47.135 And I actually really like the simple, clean feature of like how many seconds 00:33:47.165 --> 00:33:50.545 of the outro do you want to cut from the end of every single podcast, which when 00:33:50.545 --> 00:33:53.305 you're listening ongoing is helpful. 00:33:53.595 --> 00:33:55.675 I also think his shortened silences. 00:33:56.675 --> 00:33:58.135 is a very nice feature. 00:33:59.135 --> 00:34:02.965 I think those two are probably the simple ones that kind of like bug me. 00:34:02.965 --> 00:34:04.275 There's probably other stuff. 00:34:05.275 --> 00:34:05.995 Fraser Kelton: Yeah, isn't it funny? 00:34:06.025 --> 00:34:08.695 The I forget what he calls it, but the shortened silences 00:34:08.695 --> 00:34:12.155 feature is so lovingly done. 00:34:12.175 --> 00:34:16.045 And when he's spoken about how he built that in the past, you just realize 00:34:16.045 --> 00:34:17.835 here's somebody who loves his craft. 00:34:18.835 --> 00:34:23.375 It is so great that when I listen on other players and I try their speed 00:34:23.375 --> 00:34:26.935 up equivalent, it's just jarring. 00:34:26.935 --> 00:34:27.645 Nabeel Hyatt: It's not as good. 00:34:27.645 --> 00:34:28.105 It's just jarring. 00:34:28.105 --> 00:34:29.475 It's not listenable. 00:34:29.725 --> 00:34:30.275 Fraser Kelton: That's right. 00:34:30.385 --> 00:34:30.845 That's right. 00:34:31.275 --> 00:34:31.635 Interesting. 00:34:31.645 --> 00:34:32.315 Oh, the last one is 00:34:32.315 --> 00:34:32.885 Nabeel Hyatt: Snipped. 00:34:33.885 --> 00:34:38.185 Very simple one also is Snipped is very bad at downloading episodes ahead of time. 00:34:39.185 --> 00:34:42.275 So it's fine for streaming, but like I get on a plane and the 00:34:42.275 --> 00:34:44.755 only thing it downloads is things are already sitting in the queue. 00:34:45.125 --> 00:34:46.665 And I want it to just, it's audio, man. 00:34:46.675 --> 00:34:47.575 It doesn't take up that much space. 00:34:47.960 --> 00:34:50.800 Just, like, download hundreds of podcasts so I can listen to what I want. 00:34:50.830 --> 00:34:51.890 It's still bad at that. 00:34:52.890 --> 00:34:52.990 Yeah. 00:34:53.200 --> 00:34:57.850 But the whole point is, like, is it gonna, is it, in a year, is Snipp gonna 00:34:57.850 --> 00:34:59.180 be able to download more podcasts? 00:34:59.280 --> 00:35:00.260 Like, yeah, probably. 00:35:00.350 --> 00:35:05.170 But they, they created a new, for me at least, they created a reason 00:35:05.170 --> 00:35:09.450 we're switching, and now I've got a new default, and defaults matter. 00:35:10.360 --> 00:35:10.760 Yep, 00:35:11.030 --> 00:35:11.410 Fraser Kelton: yep. 00:35:12.030 --> 00:35:16.120 Not to get too far ahead of things, but you can imagine that if it knows 00:35:16.690 --> 00:35:21.070 the unstructured segments of podcasts that you actually like, and the ones 00:35:21.070 --> 00:35:26.410 that you skip, There can be an entire discovery piece around, around segments 00:35:26.410 --> 00:35:28.480 rather than episodes or podcasts itself. 00:35:28.490 --> 00:35:34.470 Like here is some esoteric podcast talking in, depth around a note 00:35:34.500 --> 00:35:37.160 taking app that's going to resonate with you for these reasons. 00:35:37.180 --> 00:35:37.570 Cause 00:35:38.090 --> 00:35:39.040 Nabeel Hyatt: That's, your catnip. 00:35:39.530 --> 00:35:40.130 Oh yeah, exactly. 00:35:40.176 --> 00:35:43.136 That feels like, especially in a world where it feels like everybody 00:35:43.406 --> 00:35:45.296 woke up a year after COVID. 00:35:46.186 --> 00:35:49.836 With a great zoom setup and some mics and decided to start a podcast. 00:35:50.036 --> 00:35:51.686 Like we're going to have so much audio. 00:35:51.736 --> 00:35:52.736 It's hilarious. 00:35:52.746 --> 00:35:55.656 Like I know we're contributing to the problem, but everybody 00:35:56.136 --> 00:35:58.876 with a podcast does create a new problem, which is content discovery. 00:35:59.646 --> 00:36:02.636 And, I don't think for most podcasts, the answer is. 00:36:02.856 --> 00:36:06.216 New things to subscribe to, to listen to every single episode, but 00:36:06.386 --> 00:36:10.436 if Snippet works, you can also see this cycle where you can listen to 00:36:10.486 --> 00:36:12.286 snippets that are more valuable. 00:36:12.446 --> 00:36:15.756 It was, I'm sure their original value proposition, I know it was 00:36:15.796 --> 00:36:16.916 because I saw it a while ago. 00:36:17.296 --> 00:36:18.896 I don't think it works. 00:36:19.391 --> 00:36:21.911 You don't have enough data and you don't have enough nuance and 00:36:21.911 --> 00:36:24.361 unstructured organization before LLMs. 00:36:24.611 --> 00:36:27.671 And now with LLMs, I actually think there's a possibility where a kind 00:36:27.671 --> 00:36:31.201 of like okay idea for a product might actually be a real thing. 00:36:31.381 --> 00:36:31.691 We'll see. 00:36:32.691 --> 00:36:33.321 Yeah, I 00:36:33.351 --> 00:36:35.191 Fraser Kelton: can imagine that you're not even just getting recommended, 00:36:35.261 --> 00:36:38.611 but you're listening to a snipped podcast that is nothing but segments 00:36:38.611 --> 00:36:43.291 that they think you're going to like on, a topic across a plethora 00:36:43.291 --> 00:36:45.481 of different podcasts all stitched 00:36:45.481 --> 00:36:45.941 Nabeel Hyatt: together. 00:36:46.386 --> 00:36:48.556 Yeah, they have, they're already, I think they already are in 00:36:48.576 --> 00:36:49.696 progress on something similar. 00:36:50.466 --> 00:36:50.876 I see. 00:36:50.976 --> 00:36:52.366 Well, listen, ship it, guys. 00:36:52.416 --> 00:36:53.836 Jeremy was remarking the other night. 00:36:53.846 --> 00:36:54.566 I, I do. 00:36:55.116 --> 00:36:58.610 Basically record as much of my life as I can right now. 00:36:58.820 --> 00:37:02.282 Partially with the idea that it will be maybe valuable later on. 00:37:02.762 --> 00:37:05.562 I don't know that I'm saying anything insightful, but maybe something 00:37:05.622 --> 00:37:07.162 somebody else will say is insightful. 00:37:07.162 --> 00:37:09.356 Or maybe this product can do something with that audio. 00:37:09.576 --> 00:37:11.844 There's got to be a model somewhere that's going to do something with it. 00:37:12.594 --> 00:37:14.904 Fraser Kelton: I think on all of that stuff, we're going to realize that... 00:37:15.128 --> 00:37:19.008 Nothing exciting or interesting or novel happens in 99 percent of our 00:37:19.008 --> 00:37:24.968 lives, and so you'll have terabytes of content that will remind you of the dull 00:37:25.968 --> 00:37:26.128 Nabeel Hyatt: (Laughter). 00:37:26.673 --> 00:37:30.603 Fraser Kelton: repetitiveness of life, by, punctuated by very brief 00:37:30.603 --> 00:37:33.503 moments of intense highs and lows. 00:37:33.938 --> 00:37:38.038 Nabeel Hyatt: So Fraser, are you prognosticating, that our AI 00:37:38.038 --> 00:37:42.028 summarizing our lives perfectly, will drive a global nihilism understanding 00:37:42.068 --> 00:37:43.658 of the futility of our daily lives? 00:37:44.658 --> 00:37:46.518 Fraser Kelton: People don't even care to that degree, right? 00:37:46.528 --> 00:37:48.918 There will be a small pocket of people who have to confront the 00:37:48.918 --> 00:37:50.688 idea that there's nothing going 00:37:51.028 --> 00:37:52.148 Nabeel Hyatt: just not that important. 00:37:52.588 --> 00:37:52.988 Fraser Kelton: Yeah. 00:37:53.058 --> 00:37:56.568 And everybody else will be watching Sunday night football like they always had been. 00:37:56.774 --> 00:37:58.964 Nabeel Hyatt: Emmett Shear, has this thing about I think he talked about 00:37:58.964 --> 00:38:01.084 it at dinner the other night where he's thinking it's yeah, it turns out. 00:38:01.434 --> 00:38:02.484 That the reason Justin. 00:38:02.754 --> 00:38:07.454 tv doesn't work is that nobody has an interesting life on a day to day basis. 00:38:07.454 --> 00:38:10.964 And the reason that live reality television doesn't work all the time 00:38:10.964 --> 00:38:15.704 is because even a crazy celebrity still mostly spends their day doing 00:38:15.714 --> 00:38:17.904 mundane and stupid, stupid things. 00:38:17.964 --> 00:38:18.634 And so, 00:38:19.634 --> 00:38:20.214 Fraser Kelton: Yeah. 00:38:20.974 --> 00:38:25.233 And then you get moments like the other night when a founder is dealing with 00:38:25.233 --> 00:38:28.333 immigration hell and you're trying to figure out how to get this guy who's 00:38:28.403 --> 00:38:32.743 just flown into SFO and been held for 23 hours into the country because 00:38:32.743 --> 00:38:35.803 he's legally supposed to be there, but he didn't have the right paperwork. 00:38:36.353 --> 00:38:38.773 Nabeel Hyatt: I'm surprised that tech industry hasn't found some way 00:38:38.953 --> 00:38:42.363 to break through with both parties in Washington on immigration. 00:38:42.413 --> 00:38:45.193 You should not have a founder of a venture backed startup. 00:38:45.723 --> 00:38:50.753 Having to flee the country quickly because of visa problems when they're 00:38:50.753 --> 00:38:52.663 just trying to build a company in America. 00:38:52.663 --> 00:38:52.923 Fraser Kelton: Yeah. 00:38:52.983 --> 00:38:54.833 And, I mean, the visa's already been approved. 00:38:54.833 --> 00:38:56.813 He just presented the wrong piece of paper. 00:38:56.913 --> 00:38:58.283 It's just very strange. 00:38:58.443 --> 00:39:03.563 I mean, the idea that we train the best in the world and then kick 00:39:03.563 --> 00:39:07.593 them out some number of months after doesn't make any sense either, right? 00:39:07.593 --> 00:39:07.713 Nabeel Hyatt: Yeah. 00:39:07.713 --> 00:39:12.228 That's another thing . This may be counter to everybody hating on 00:39:12.228 --> 00:39:13.268 all the Ivy's and everything else. 00:39:14.268 --> 00:39:17.418 Higher education is a magnet that pulls. 00:39:17.823 --> 00:39:21.463 Immigrants to us to start with, because frankly, there's already 00:39:21.463 --> 00:39:24.653 a Google office in Mumbai, right? 00:39:24.763 --> 00:39:27.573 And so, they're going to come here for college, and then it's 00:39:27.583 --> 00:39:31.583 our job as an ecosystem to make sure they stay, that the best and 00:39:31.583 --> 00:39:32.593 brightest who come here for college. 00:39:32.663 --> 00:39:36.533 I actually, I have my issues with higher education and the way it's structured, 00:39:36.763 --> 00:39:41.283 but it is one of the industries of which we massively dominate. 00:39:41.303 --> 00:39:43.353 And so trying to disrupt it and kill it. 00:39:43.888 --> 00:39:46.468 Just hurts the whole ecosystem of startups. 00:39:47.008 --> 00:39:50.378 So I fix it But the we should just tear it all down and colleges don't 00:39:50.378 --> 00:39:52.818 matter is ignoring what colleges mean 00:39:52.868 --> 00:39:55.878 Fraser Kelton: the idea that you don't want to take the best electrical 00:39:55.878 --> 00:40:00.108 engineer in the world and put them into a six year PhD in some esoteric 00:40:00.168 --> 00:40:04.338 area where they're going to go so deep that there might just be peer 00:40:04.338 --> 00:40:06.228 basic research that comes out of it. 00:40:06.288 --> 00:40:10.453 Or it might be some, some other benefit that gets kicked off. 00:40:10.453 --> 00:40:13.623 Like we should just be investing and supporting that 00:40:13.623 --> 00:40:14.973 as much as we possibly can. 00:40:15.473 --> 00:40:17.914 Nabeel Hyatt: Yeah, look, Community colleges are actually very, very 00:40:17.914 --> 00:40:20.294 beneficial to local communities. 00:40:20.294 --> 00:40:22.824 But right above that, you got a bunch of schools that aren't doing crap for 00:40:22.824 --> 00:40:25.164 anybody except for, creating student debt. 00:40:25.194 --> 00:40:25.544 Fine. 00:40:26.024 --> 00:40:31.484 But for the top third to 40 percent it is the beginning of the process of 00:40:31.484 --> 00:40:34.634 the engine of what makes America work. 00:40:34.694 --> 00:40:35.864 Are there problems after that? 00:40:35.884 --> 00:40:39.944 Is the idea of yet yet another person going 12 years down PhD 00:40:39.944 --> 00:40:41.744 physics research only to... 00:40:42.019 --> 00:40:45.949 Find out that they're working in some field so deep that they're likely to 00:40:45.949 --> 00:40:47.289 never hit a breakthrough at this point. 00:40:47.299 --> 00:40:49.669 And they're throwing their whole life away as one of the brightest people in America. 00:40:49.719 --> 00:40:51.009 I can make that rant, too. 00:40:51.069 --> 00:40:55.639 We under indexed how much cross disciplinary learning there 00:40:55.639 --> 00:40:58.749 is when you get to actually coming through breakthroughs. 00:40:58.979 --> 00:41:01.599 . But I think if you remove it, you really do a lot of damage 00:41:01.599 --> 00:41:02.669 to the whole system, so. 00:41:03.019 --> 00:41:04.849 Fraser Kelton: But even that I'm just looking up Dario 00:41:04.849 --> 00:41:07.099 from Anthropic's LinkedIn. 00:41:07.449 --> 00:41:09.609 He has a PhD in Biophysics, right? 00:41:09.614 --> 00:41:12.184 So he went deep in some sort of, some sort of weird topic. 00:41:12.184 --> 00:41:13.994 I don't even know what biophysics would be. 00:41:14.094 --> 00:41:16.044 And now he's, he's running anthropic. 00:41:16.094 --> 00:41:19.894 I think that there's certain types of minds that, that get attracted to 00:41:19.894 --> 00:41:25.294 these deep, programs that also then learn how to learn and, and grind and 00:41:25.304 --> 00:41:28.624 all sorts of different skills that are quite applicable and valuable 00:41:28.634 --> 00:41:31.704 for creating pretty profound things. 00:41:32.409 --> 00:41:32.689 Nabeel Hyatt: Yeah. 00:41:33.079 --> 00:41:33.749 I think that's fair. 00:41:33.949 --> 00:41:37.989 . I think people have a hard time with the non deterministic nature of it all too. 00:41:38.039 --> 00:41:42.869 This idea of who do we let in and what are they going to study and it's we 00:41:42.869 --> 00:41:47.439 as a society seem to want everything to be rapidly deterministic, right? 00:41:47.689 --> 00:41:49.229 I come in to get the biophysics. 00:41:49.239 --> 00:41:51.199 How many biophysics jobs are there in the world? 00:41:51.519 --> 00:41:53.159 Do we need more people in biophysics? 00:41:53.159 --> 00:41:57.659 It's just a very linear thinking when, the nature of an economy, the 00:41:57.659 --> 00:42:00.189 nature of our, frankly, our human lives, the nature, certainly the 00:42:00.189 --> 00:42:01.399 nature of innovation and startups. 00:42:01.694 --> 00:42:05.384 Is that a little bit of randomness and a little bit of chaos is pretty important. 00:42:05.514 --> 00:42:09.304 And that comes to college admissions, that comes to job admissions, that comes 00:42:09.304 --> 00:42:11.954 to what you're going to work on 10 years from now in your life, all of that. 00:42:12.414 --> 00:42:15.294 But we live in an era of data science, and, and numbers. 00:42:15.294 --> 00:42:18.314 So the things that seem to belie that, that determinist 00:42:18.664 --> 00:42:20.074 nature people have trouble with, 00:42:20.124 --> 00:42:21.214 Fraser Kelton: So here are the names. 00:42:21.294 --> 00:42:28.974 Random Seeds, Latent Chat, Unsupervised Learning, Weekly Prompts, do you think? 00:42:29.974 --> 00:42:31.424 Nabeel Hyatt: So, okay, wait, wait, let's start over again. 00:42:31.924 --> 00:42:35.774 The potential names that, was this, did we use ChatGPT or Claude for 00:42:35.789 --> 00:42:36.949 Fraser Kelton: We use Claude and, 00:42:37.014 --> 00:42:37.614 Nabeel Hyatt: this podcast. 00:42:37.899 --> 00:42:38.669 and ChatGPT. 00:42:38.709 --> 00:42:41.129 Okay, so using both Claude and ChatGPT. 00:42:41.639 --> 00:42:45.349 The answers, and this is the, just so we're clear, this is a cold, this 00:42:45.349 --> 00:42:49.269 is like the top 10 percent of the answers that GPT and Claude gave us 00:42:49.319 --> 00:42:51.089 for names for this podcast, right? 00:42:51.579 --> 00:42:52.979 Run it down again, Fraser? 00:42:53.849 --> 00:42:55.049 Fraser Kelton: Random seeds. 00:42:55.369 --> 00:42:56.449 Latent chat. 00:42:56.899 --> 00:42:59.009 I'm even editing, so I won't edit it. 00:42:59.179 --> 00:43:00.239 Tensor talk. 00:43:00.479 --> 00:43:01.429 Tensor talk. 00:43:02.429 --> 00:43:03.939 Unsupervised learnings. 00:43:04.309 --> 00:43:07.139 And, and the other one that I removed at the start was gen talk, because 00:43:07.139 --> 00:43:10.599 anytime anybody says gen AI, I cringe and die a little bit inside. 00:43:11.099 --> 00:43:12.619 I'm not sure that any of these are great. 00:43:13.554 --> 00:43:16.674 Nabeel Hyatt: They do have a certain dad humor quality to them that I appreciate 00:43:17.389 --> 00:43:17.839 Fraser Kelton: Yeah. 00:43:17.899 --> 00:43:20.679 Isn't that interesting is it just reinforces how hard 00:43:21.139 --> 00:43:23.029 humor is to capture with 00:43:23.054 --> 00:43:25.094 Nabeel Hyatt: well, it's obviously there's just too much dad humor 00:43:25.099 --> 00:43:26.714 in the training data . That's the 00:43:26.764 --> 00:43:27.354 Fraser Kelton: well, that's what I mean. 00:43:27.354 --> 00:43:31.774 Like the funny people are funny because that's a scarce resource. 00:43:31.774 --> 00:43:33.954 It's not like everybody's walking around being hilarious. 00:43:34.669 --> 00:43:35.089 Nabeel Hyatt: That's right. 00:43:35.239 --> 00:43:39.049 I have no strong visceral response to this at all. 00:43:39.209 --> 00:43:39.659 You're right. 00:43:39.659 --> 00:43:41.429 Unsupervised learning is the least cringe. 00:43:41.459 --> 00:43:42.479 I can at least say that. 00:43:42.984 --> 00:43:45.774 But if you are one of the handful of folks that likely listen and you have 00:43:45.774 --> 00:43:48.354 a suggestion as well, let us know. 00:43:48.854 --> 00:43:50.214 And that's, let's call it for today. 00:43:51.004 --> 00:43:51.654 Fraser Kelton: Let's call it. 00:43:51.904 --> 00:43:52.364 Thanks. 00:43:53.314 --> 00:43:53.664 Nabeel Hyatt: Awesome. 00:43:54.124 --> 00:43:54.664 Talk to you later.