WEBVTT NOTE This file was generated by Descript 00:00:01.610 --> 00:00:05.820 Nabeel Hyatt (2): I feel like in a world where most of us are trying to 00:00:05.820 --> 00:00:11.549 process the entry of this new alien species, this new thing, this LLM and 00:00:11.549 --> 00:00:15.120 how it's going to work in our life, I think we've mostly processed it in 00:00:15.120 --> 00:00:17.060 very, very short go to market spurts. 00:00:17.739 --> 00:00:17.959 Right? 00:00:17.960 --> 00:00:22.300 the world is filled with incubators that are asking you, you know, in three months, 00:00:22.300 --> 00:00:25.220 make sure you launch the product and get some traction to show up for demo day. 00:00:25.500 --> 00:00:26.180 Or. 00:00:26.335 --> 00:00:27.195 It's a hackathon. 00:00:27.205 --> 00:00:29.575 Let's see what we can make in two days over the weekend. 00:00:29.715 --> 00:00:33.304 But My personal feeling was like, no, also like we should get 00:00:33.304 --> 00:00:35.315 weirder before it gets better. 00:00:35.364 --> 00:00:38.755 Fraser Kelton: I think that we have undoubtedly seen in the market that if 00:00:38.755 --> 00:00:41.055 you go try to get like your merry men of. 00:00:41.260 --> 00:00:46.009 Researchers together to go like gallivant around and start a company. 00:00:46.510 --> 00:00:48.140 I don't think that usually works out well. 00:00:48.240 --> 00:00:51.410 Nabeel Hyatt (2): and then inevitably, in the kind of rollercoaster of things, 00:00:51.410 --> 00:00:56.255 you're going to wake up one day, And, you know, there's some headline 00:00:56.255 --> 00:01:00.045 of somebody who's raised a billion dollars at, you know, whatever it 00:01:00.045 --> 00:01:02.025 is, it's 10x more than you've got. 00:01:02.225 --> 00:01:03.105 That's for sure. 00:01:05.065 --> 00:01:06.485 Fraser Kelton: Hey everybody, welcome back. 00:01:06.705 --> 00:01:07.575 This time for real. 00:01:08.565 --> 00:01:09.265 I'm Fraser. 00:01:09.295 --> 00:01:09.855 Nabeel Hyatt (2): I'm Nabeel. 00:01:09.855 --> 00:01:10.665 Welcome to Hallway Chat. 00:01:10.965 --> 00:01:12.085 Fraser Kelton: Welcome to Hallway Chat. 00:01:12.330 --> 00:01:14.050 You have a product for us, Nabeel. 00:01:14.070 --> 00:01:18.680 You came metaphorically running into the Slack channel where we talk 00:01:18.690 --> 00:01:21.890 about products and you dump something in, and it was the first time in a 00:01:21.890 --> 00:01:25.780 while where I've seen a lot of us jump on it almost instantaneously. 00:01:25.780 --> 00:01:28.870 Tell us, tell us what it is and let's jump in. 00:01:28.890 --> 00:01:30.950 Nabeel Hyatt (2): Well, yeah, I think we should talk about granola. 00:01:31.390 --> 00:01:32.100 this week. 00:01:32.160 --> 00:01:33.820 I've used this product a couple times. 00:01:33.860 --> 00:01:35.680 Uh, Chris Pedregal is the CEO. 00:01:35.720 --> 00:01:39.090 He's actually previously founded Socratic, which was a Spark portfolio 00:01:39.090 --> 00:01:42.110 company back in the day, doing a whole nother product category. 00:01:42.110 --> 00:01:46.340 We could talk about another time, which is, uh, Socratic was the early, take a 00:01:46.340 --> 00:01:51.329 photo of your homework and use AI to solve it, which there's a whole category of 20, 00:01:51.329 --> 00:01:53.459 000 of them in the app store right now. 00:01:53.489 --> 00:01:54.569 But that's not what we're talking about. 00:01:54.619 --> 00:01:59.769 He's doing a new company called Granola just launched this week, And, you know, it 00:01:59.779 --> 00:02:03.659 fits loosely into the AI note taking app. 00:02:04.439 --> 00:02:06.389 Everyone jumps on it in Slack because we all like to use 00:02:06.389 --> 00:02:07.719 products, but let's be real. 00:02:07.849 --> 00:02:11.059 We also do a lot of meetings and this is an AI note taking app for meetings. 00:02:11.219 --> 00:02:15.289 So it's also a little bit of like customer fit as well for the team. 00:02:15.290 --> 00:02:18.759 And, um, and some of us really already know and respect Chris. 00:02:19.179 --> 00:02:24.239 The product really simply is you jump into a Zoom with somebody, There's a little 00:02:24.239 --> 00:02:28.339 tag on the right hand side, the little G for granola that says join now and 00:02:28.339 --> 00:02:31.839 it pulls up a blank sheet of paper for taking notes the same way you would have 00:02:31.859 --> 00:02:33.559 brought up Apple Notes or something else. 00:02:34.089 --> 00:02:39.949 I think there are a lot of very interesting product decisions that Chris, 00:02:40.279 --> 00:02:45.514 he's just a product person, did that are counter to the, 30 other products that 00:02:45.514 --> 00:02:47.574 have tried some version of AI note taking. 00:02:47.574 --> 00:02:52.414 And I'm thinking about everything from stuff like Otter and Notta and Fireflies, 00:02:52.734 --> 00:02:57.444 uh, to whole screen recording, companies like Limitless, to a product that we 00:02:57.444 --> 00:03:01.114 talked about, I think, in maybe six or seven or eight weeks ago, called 00:03:01.114 --> 00:03:04.504 SuperPowered, which is the previous AI note taking product I had been using. 00:03:05.064 --> 00:03:06.174 And so I think we should just get into it. 00:03:06.194 --> 00:03:09.228 I think there's some interesting product decisions here with 00:03:09.228 --> 00:03:10.780 lessons for other AI products. 00:03:10.870 --> 00:03:14.030 Fraser Kelton: Before we jump in, let's set up the discussion by talking around 00:03:14.030 --> 00:03:16.860 the high level value proposition, because I think that's why we were 00:03:16.870 --> 00:03:21.440 all, uh, drawn to it initially, is when he shared the announcement, 00:03:21.440 --> 00:03:24.254 he said, if a bot is writing notes for you, it's thinking for you. 00:03:24.254 --> 00:03:27.308 And instead, Granola starts with your notes and makes them better. 00:03:27.708 --> 00:03:31.888 You know, we've talked a lot about whether or not there's an agent or an assistant 00:03:32.128 --> 00:03:35.148 that, is taking control of tasks for you. 00:03:35.533 --> 00:03:40.993 Uh, and we've also then spoken about co pilots, or paired programmer like 00:03:41.023 --> 00:03:45.473 experiences that are collaboratively building, your work product. 00:03:45.863 --> 00:03:48.333 And this is the first note taking app, at least that I've used, 00:03:48.353 --> 00:03:49.603 that's in that second camp. 00:03:49.653 --> 00:03:54.193 The other ones seem to basically want to take full control and put on autopilot 00:03:54.193 --> 00:03:57.708 the notes, whereas this one you have I don't know, for lack of a better 00:03:57.708 --> 00:04:01.618 word, editorial control over what it is that it's going to like key in on 00:04:01.628 --> 00:04:05.868 based on the notes that you're taking while the, the conversation unfolds. 00:04:06.328 --> 00:04:08.078 Nabeel Hyatt (2): The experience for those who haven't tried it yet 00:04:08.328 --> 00:04:12.228 is you're just writing into a blank notepad that way you normally would. 00:04:12.258 --> 00:04:16.828 And then at the end, it goes through a transcription and note taking step where 00:04:17.108 --> 00:04:20.808 it then does the thing that say Otter or Nada would do, which is to give you 00:04:20.808 --> 00:04:26.952 a structured, summary of your overall conversation, but it then integrates and 00:04:26.952 --> 00:04:28.905 is influenced by the notes that you took 00:04:28.935 --> 00:04:33.985 . So if you had, say, written down a basic outline as you were going through things 00:04:33.985 --> 00:04:37.875 or a couple of key insights that you really captured or a quote that you really 00:04:37.875 --> 00:04:42.385 loved, it'll fit into the hierarchy of the way it then does summary controls, 00:04:42.395 --> 00:04:44.385 kind of merging the two things together. 00:04:45.025 --> 00:04:48.470 In practice, it feels almost like You wrote down the things that 00:04:48.470 --> 00:04:52.880 super mattered, and you had an intern fill out all the rest. 00:04:52.970 --> 00:04:54.490 And, and I really like that. 00:04:54.650 --> 00:04:57.930 I think, you know, we've had lots of conversations about agents. 00:04:57.930 --> 00:05:00.590 We had a long conversation about Devin and Cognition and Coding Agents. 00:05:00.850 --> 00:05:04.840 These things that might try and go away for a little while and go do a thing. 00:05:05.380 --> 00:05:10.860 None of these things at this stage of AI are really perfect or great. 00:05:11.260 --> 00:05:14.350 I remember when Anthropic came in, they, they gave the analogy of, We're 00:05:14.350 --> 00:05:17.860 going to start with an AI that's good enough to be a kind of okay intern, 00:05:17.890 --> 00:05:21.480 and then slowly move your way up the stack in terms of capabilities. 00:05:21.870 --> 00:05:25.180 And so another way of kind of like inverting that and thinking about a UI 00:05:25.180 --> 00:05:30.040 cycle is, What are the types of tasks in the world, not that can be done 00:05:30.040 --> 00:05:33.150 entirely by somebody else, like that I would outsource to India to go do, 00:05:33.150 --> 00:05:36.290 which is like customer service, use cases and so forth, but what are the 00:05:36.290 --> 00:05:40.360 kinds of tasks that normally somebody is already looking over your shoulder? 00:05:41.270 --> 00:05:44.120 to help with that your manager is already kind of heavily 00:05:44.120 --> 00:05:45.220 involved with and looking at. 00:05:45.450 --> 00:05:49.780 Those are probably really prime examples of situations where an AI 00:05:49.830 --> 00:05:51.770 can be the co pilot, the assistant. 00:05:52.300 --> 00:05:55.350 Uh, and it's funny, we've had co pilot around for a couple of years now 00:05:55.350 --> 00:05:58.340 and like everyone tried to move past it really quickly without, I think 00:05:58.350 --> 00:06:01.380 Grinnell is a very good example of Oh, there's a lot still to mine there. 00:06:01.620 --> 00:06:06.550 Um, somehow we jumped out of that when we did AI note taking and we went straight 00:06:06.560 --> 00:06:09.930 to, Nope, this whole thing is totally filled out instead of thinking of it. 00:06:10.540 --> 00:06:11.610 You know, it's pair programming. 00:06:11.800 --> 00:06:15.610 This is like pair programming applied to pair note taking, which is, I 00:06:15.620 --> 00:06:16.950 think, a really interesting metaphor. 00:06:17.000 --> 00:06:17.410 Fraser Kelton: Yep. 00:06:17.800 --> 00:06:18.620 I think so. 00:06:18.620 --> 00:06:20.700 I think the other thing that this highlights at a very high level 00:06:20.730 --> 00:06:24.050 is the importance of product taste and product decisions. 00:06:24.050 --> 00:06:28.090 Like the product is super immature and young still, so it has, you 00:06:28.090 --> 00:06:29.500 know, limitations and rough edges. 00:06:29.500 --> 00:06:31.930 And we can talk about those if it's interesting, but it is pretty clear 00:06:31.940 --> 00:06:33.280 those are going to be on the roadmap. 00:06:33.630 --> 00:06:37.960 But what they shipped is so opinionated versus what is in the 00:06:37.960 --> 00:06:39.660 market today for note taking apps. 00:06:40.005 --> 00:06:41.505 And, thoughtfully done. 00:06:41.535 --> 00:06:44.855 One of the themes that we've talked about for a handful of weeks has been 00:06:44.855 --> 00:06:48.585 the idea that these models aren't great and you need to fail gracefully 00:06:48.585 --> 00:06:52.685 across different use cases and find ways to do that to engender trust. 00:06:53.105 --> 00:06:55.985 I'll tell you, there was a meeting where I was using it. 00:06:55.985 --> 00:06:58.195 I carefully took a note that was actually pretty important. 00:06:58.235 --> 00:07:01.435 And then the auto summarize, uh, I don't know what you would call it, like the, 00:07:01.455 --> 00:07:06.235 the co pilot piece filled in the blanks and gave me a nice robust note at the end 00:07:06.235 --> 00:07:10.635 of it, but it had dropped that one piece that I had taken as critical information 00:07:11.345 --> 00:07:13.475 and I, I, I was like really angry, right? 00:07:13.475 --> 00:07:14.435 I'm like, how dare you? 00:07:14.435 --> 00:07:16.735 If I write it down don't, don't do that. 00:07:16.735 --> 00:07:16.815 Yeah. 00:07:16.815 --> 00:07:17.845 You're yelling at the intern. 00:07:17.915 --> 00:07:18.145 Yeah. 00:07:18.485 --> 00:07:18.795 Yeah. 00:07:21.190 --> 00:07:22.860 Um, constructive feedback. 00:07:22.860 --> 00:07:24.700 I was giving the constructive feedback. 00:07:25.090 --> 00:07:28.320 Um, but there's a toggle at the bottom where you can actually just go back to 00:07:28.330 --> 00:07:30.530 the entirety of just your notes, right? 00:07:30.600 --> 00:07:33.680 And so you can basically turn off any of the, like the intern 00:07:33.690 --> 00:07:36.830 scribbles around the margins and get back to where you want to be. 00:07:37.030 --> 00:07:41.210 And the other place where they do that, is the entire transcript is there. 00:07:41.220 --> 00:07:41.410 Nabeel Hyatt (2): Right. 00:07:41.430 --> 00:07:43.040 You can, you can drill down to the transcript, right? 00:07:43.040 --> 00:07:44.740 Fraser Kelton: Yeah, you can drill down to the transcript. 00:07:45.000 --> 00:07:46.070 Um, and then 00:07:46.150 --> 00:07:47.350 Nabeel Hyatt (2): I'm not sure if I agree with that one. 00:07:47.350 --> 00:07:48.980 I don't know if I like that choice. 00:07:49.020 --> 00:07:53.890 I can understand why they did it for exposure, but you know, there's something 00:07:53.900 --> 00:07:57.350 about staring at a transcript of all the uhs and ums and all the terrible 00:07:57.360 --> 00:08:01.800 things that feels, boasts like a little bit more of an invasion of privacy. 00:08:02.055 --> 00:08:04.995 a little bit more cringy, which is hilarious, of course, because you know 00:08:04.995 --> 00:08:09.425 it's recording, but staring at all of the words feels quite different to me than 00:08:09.735 --> 00:08:11.195 you're only allowed to get a summary. 00:08:11.195 --> 00:08:12.315 So I'll disagree with you there. 00:08:12.355 --> 00:08:13.855 I probably would have hidden the transcript. 00:08:13.875 --> 00:08:16.755 I'm sure, I'm sure Chris and the team have lots of good reasons why they did it. 00:08:17.025 --> 00:08:20.805 Fraser Kelton: Here's where I think it bubbles up in useful ways. 00:08:20.895 --> 00:08:25.695 The intern's notes are referenced or deep linked to the relevant 00:08:25.705 --> 00:08:26.725 part of the transcript. 00:08:26.735 --> 00:08:27.995 So, uh, yeah, that's right. 00:08:28.150 --> 00:08:28.400 Right. 00:08:28.420 --> 00:08:32.430 If it, if it shows you a bullet point that it's written, you can say, I don't 00:08:32.430 --> 00:08:34.010 know, what, what was the context of that? 00:08:34.010 --> 00:08:38.170 You click into it, a little overlay appears, and then you see the transcript, 00:08:38.220 --> 00:08:39.880 just the snippet, the relevant snippet. 00:08:39.920 --> 00:08:40.620 Nabeel Hyatt (2): I agree with you. 00:08:40.660 --> 00:08:42.920 Okay, so that, that part I agree with you, that, that's brilliant. 00:08:42.940 --> 00:08:44.020 Like the, the, yeah. 00:08:44.320 --> 00:08:49.310 You see a note that says, you know, we discussed, you know, AI eye 00:08:49.310 --> 00:08:52.790 contacts, and then I click a little button on it, the plus to it, and it 00:08:52.890 --> 00:08:56.400 actually shows me this short bit of transcript, a quote, basically the 00:08:56.400 --> 00:08:57.880 inline quote that referenced that. 00:08:57.930 --> 00:08:58.720 That's brilliant. 00:08:59.590 --> 00:09:01.325 You could do that without having to show me anything. 00:09:01.715 --> 00:09:04.465 All of the horrific things I said for an hour and a half as I 00:09:04.465 --> 00:09:06.185 was rambling off the top of it. 00:09:06.185 --> 00:09:06.715 For sure. 00:09:06.985 --> 00:09:08.795 Fraser Kelton: But I'll, I'll like, I agree. 00:09:08.795 --> 00:09:09.915 I agree with, with that. 00:09:09.965 --> 00:09:11.855 Um, here's a, here's one that will make you laugh. 00:09:11.865 --> 00:09:15.215 Cause I'm just looking at the notes from, uh, our call before we jumped on here. 00:09:15.215 --> 00:09:18.105 And there's a bullet point that just simply says Mike Kelly 00:09:18.155 --> 00:09:20.645 as a follow up and Mike Kelly. 00:09:20.905 --> 00:09:24.665 Um, you can imagine that when you click in that Mike Kelly is, uh, something 00:09:24.855 --> 00:09:27.145 very different than, than the name mike. 00:09:27.280 --> 00:09:27.690 Kelly. 00:09:27.720 --> 00:09:29.030 Nabeel Hyatt (2): I think it's Kevin Kelly. 00:09:29.060 --> 00:09:29.870 That's a mess up. 00:09:29.870 --> 00:09:31.210 We'll get to Kevin Kelly in a second. 00:09:31.450 --> 00:09:35.470 Oh, um, here's, you want to, you want a really awesome granola thing? 00:09:35.730 --> 00:09:37.250 So take that Mike Kelly. 00:09:37.400 --> 00:09:37.830 Yeah. 00:09:37.950 --> 00:09:39.700 Change it to Kevin. 00:09:39.760 --> 00:09:40.050 Yeah. 00:09:40.450 --> 00:09:46.810 And now if you click open the transcript quote, it has taken your correction and it 00:09:46.810 --> 00:09:49.250 has also corrected it in the transcript. 00:09:49.620 --> 00:09:50.170 Fraser Kelton: Okay. 00:09:50.620 --> 00:09:51.290 That's pretty good. 00:09:51.740 --> 00:09:52.800 Nabeel Hyatt (2): Isn't that beautiful? 00:09:52.860 --> 00:09:53.660 Fraser Kelton: That is beautiful. 00:09:54.230 --> 00:09:59.280 I think the other, theme that is really well done on this product is that 00:09:59.280 --> 00:10:04.730 they've reduced the friction across the, uh, the product in dramatic ways. 00:10:04.850 --> 00:10:08.090 And a couple of examples, it offs into your calendar. 00:10:08.510 --> 00:10:13.400 And so when you have a meeting coming up, it's the product that basically 00:10:13.400 --> 00:10:16.590 gives you the notification that a meeting is coming up and there's a one click, 00:10:17.000 --> 00:10:19.330 uh, way to launch into the video call, 00:10:21.555 --> 00:10:22.035 I don't know. 00:10:22.485 --> 00:10:23.365 Uninteresting? 00:10:23.415 --> 00:10:23.655 Sure. 00:10:23.695 --> 00:10:28.975 But what it does is it like also then drops you into the note taking experience, 00:10:29.025 --> 00:10:31.035 perfectly positioned on the screen. 00:10:31.035 --> 00:10:34.315 You have your video on one side, the note taking app on the other. 00:10:34.675 --> 00:10:37.395 The title is already filled in for the meeting. 00:10:37.395 --> 00:10:40.755 Cursor's already live where you can just start typing. 00:10:41.085 --> 00:10:45.115 Somebody has cared enough about the product experience 00:10:45.115 --> 00:10:47.215 to craft all of those details. 00:10:47.475 --> 00:10:48.115 That's one. 00:10:48.165 --> 00:10:51.735 And then the second one is, is what you pointed out, and that is, it 00:10:51.735 --> 00:10:53.855 takes over the system settings. 00:10:53.925 --> 00:10:57.525 Nabeel Hyatt (2): Look, I think metaphorically, if we think about this, 00:10:57.675 --> 00:11:01.715 how would I want a note taking AI to insert my note taking, you can think of 00:11:01.715 --> 00:11:03.815 these things as three different buckets. 00:11:04.055 --> 00:11:09.635 And the first one is things like Mem or Notion, which is you're taking notes 00:11:09.635 --> 00:11:14.450 exactly the way you always took them, And then we're going to layer a ChatGPT 00:11:14.460 --> 00:11:16.090 like experience on top of it, right? 00:11:16.110 --> 00:11:16.380 Right? 00:11:16.430 --> 00:11:18.630 I can right click on it and say, Summarize. 00:11:18.910 --> 00:11:21.600 I can right click on it and say, Expand. 00:11:21.600 --> 00:11:23.620 I can give you an outline and say, Flesh it out. 00:11:23.940 --> 00:11:27.855 That's kind of like, Yeah, it's a ChatGPT for my notes. 00:11:28.165 --> 00:11:32.585 Um, with notes, the way you previously used it being the primary affordance. 00:11:33.165 --> 00:11:36.025 The second is the inverse of that. 00:11:36.245 --> 00:11:37.935 Um, it's not a note taking app. 00:11:37.995 --> 00:11:40.985 It's a, the AI is the thing that takes the notes, and then you can 00:11:40.985 --> 00:11:42.115 do something with them differently. 00:11:42.115 --> 00:11:45.715 And that's the Otter, the Nada, and the Fireflies of the world. 00:11:46.180 --> 00:11:47.480 which drive me crazy. 00:11:47.760 --> 00:11:48.820 I've used all of them. 00:11:48.820 --> 00:11:50.040 I've tried all of them. 00:11:50.280 --> 00:11:55.940 They are okay for some specific use cases, but especially the ones that don't 00:11:55.940 --> 00:11:57.370 let you edit the prompts themselves. 00:11:57.370 --> 00:12:01.080 So it's some sales template thing, uh, that somebody at one of these 00:12:01.080 --> 00:12:03.950 companies decided is the only way you're supposed to get your summarized meeting. 00:12:04.180 --> 00:12:05.860 It's just, just not interesting. 00:12:06.010 --> 00:12:09.590 The one I do use constantly is called SuperPowered, which is quite useful. 00:12:09.865 --> 00:12:13.285 I record lots of, you know, meetings when I need to, and they have a 00:12:13.295 --> 00:12:16.055 custom prompt framework so I can have it output the outputs that I really 00:12:16.055 --> 00:12:17.185 want and summarize it the way I want. 00:12:17.215 --> 00:12:18.325 It's great, actually. 00:12:18.325 --> 00:12:19.025 I really like it. 00:12:19.385 --> 00:12:21.055 Um, that's why we talked about it before. 00:12:21.525 --> 00:12:24.765 But this is the, this is something different, which is just it's not that 00:12:24.765 --> 00:12:30.210 the AI is the first, input, and it's not that my old view of the world, the way I 00:12:30.210 --> 00:12:33.850 would think of Notion as the old input, it's this, again, it's this kind of co 00:12:33.860 --> 00:12:39.390 pilot view that I, I now, I immediately came away from Granola trying to think 00:12:39.400 --> 00:12:42.880 about lots of other consumer categories and enterprise categories and saying Oh, 00:12:43.060 --> 00:12:45.060 well, What would you do if it was that? 00:12:45.380 --> 00:12:50.540 And so I love it when a product is so inventive in its product surface area 00:12:50.540 --> 00:12:53.870 that it makes you kind of reevaluate, um, what, and I won't list all the 00:12:53.870 --> 00:12:56.100 other categories or anything like that because we'll have time to go 00:12:56.100 --> 00:12:56.960 find those founders or whatever, 00:12:59.020 --> 00:13:02.340 but it certainly made me think about what are the implications, 00:13:02.400 --> 00:13:03.310 um, for other things. 00:13:03.320 --> 00:13:04.620 And normally, I mean, It. 00:13:04.620 --> 00:13:11.060 The last thing I'll say is it just also reminded me how much we have not really 00:13:11.060 --> 00:13:12.680 been experimenting with interfaces. 00:13:12.740 --> 00:13:12.860 Fraser Kelton: Mm-Hmm. 00:13:13.160 --> 00:13:15.080 , Nabeel Hyatt (2): you know, we're, we're a little ways into 00:13:15.080 --> 00:13:20.730 chatt pt and there is just not enough people trying weird things. 00:13:20.760 --> 00:13:21.120 Right. 00:13:21.330 --> 00:13:24.450 Uh, with how AI is supposed to surface in our life. 00:13:24.890 --> 00:13:28.150 I was at a dinner last night with a bunch of people talking about agents 00:13:28.155 --> 00:13:31.930 and ai and, and, and, you know, UIs came up and somebody brought up. 00:13:32.400 --> 00:13:35.590 The mother of all demos, which if you don't know what it is, 00:13:35.600 --> 00:13:36.640 we'll drop it in the show notes. 00:13:37.180 --> 00:13:45.320 You know, it was a presentation in, I think it was like early 1969, late 1968, 00:13:45.400 --> 00:13:49.110 it was a presentation from Engelbart at the Augmented Research Center, 00:13:49.150 --> 00:13:56.425 ARC, where he basically laid out the next 40 years of computing interfaces. 00:13:56.465 --> 00:14:03.105 Like he had a device connected between Palo Alto at SRC up to, 00:14:03.105 --> 00:14:04.645 I think it was San Francisco. 00:14:04.955 --> 00:14:09.515 Um, Alan Kay is in there, like Bob Sproul's in the audience, like it's 00:14:09.515 --> 00:14:12.015 a, you know, hall of fame group. 00:14:12.065 --> 00:14:16.145 Thankfully, thankfully I think it was like, uh, Stuart Brand happened 00:14:16.145 --> 00:14:17.565 to be in the audience with a camera. 00:14:17.565 --> 00:14:18.595 And so he was recording it. 00:14:18.605 --> 00:14:19.355 It's like insane. 00:14:19.675 --> 00:14:23.045 And he's you know, it's let's show you what a mouse might be like. 00:14:23.175 --> 00:14:25.945 Let's show you these dual screens and how you might communicate, 00:14:26.035 --> 00:14:28.205 communicate over a computer. 00:14:28.715 --> 00:14:31.305 Like stuff, frankly, that's in there that we still haven't gotten to. 00:14:31.625 --> 00:14:40.145 And I feel like in a world where most of us are trying to process the entry of 00:14:40.145 --> 00:14:43.825 this new alien species, this new thing, this LLM and how it's going to work in our 00:14:43.825 --> 00:14:45.915 life, I think we've mostly processed it. 00:14:46.325 --> 00:14:49.385 in very, very short go to market spurts, right? 00:14:49.385 --> 00:14:53.995 We are now, you know, 50, 60, 70 years later, um, the world is filled 00:14:53.995 --> 00:14:57.785 with incubators that are asking you, you know, in three months, make sure 00:14:57.785 --> 00:15:01.305 you launch the product and get some traction to show up for demo day, or 00:15:01.475 --> 00:15:04.795 it's a hackathon, let's see what we can make in two days over the weekend. 00:15:05.175 --> 00:15:08.565 And, you know, it was kind of counter to the conversation we had at dinner, to be 00:15:08.565 --> 00:15:11.665 honest, because a lot of the conversation with dinner was like, wow, a lot of these 00:15:11.885 --> 00:15:16.055 agentic companies are producing demos that kind of like 80 percent work or 00:15:16.055 --> 00:15:19.315 70 percent work, but they don't really work enough to deploy into production. 00:15:19.665 --> 00:15:21.405 And that is also true. 00:15:21.995 --> 00:15:25.595 But my personal feeling was like, no, also we should get 00:15:25.595 --> 00:15:26.895 weirder before it gets there. 00:15:26.915 --> 00:15:33.025 I don't mean hackathon weird, I mean like, deeply ambitious alien weird. 00:15:33.195 --> 00:15:38.395 We need to explore the design space a lot more to kind of figure out, 00:15:38.885 --> 00:15:41.795 what is the GUI of this world? 00:15:42.615 --> 00:15:45.855 We're sitting, staring at MS DOS prompts, and no one's made Windows yet. 00:15:45.925 --> 00:15:49.755 And, and I'm not, and I I think, you know, I don't even know what 00:15:49.755 --> 00:15:53.545 the Xerox PARC or the MIT Media Lab should be for this era. 00:15:53.555 --> 00:15:57.165 It probably, it's not going to be, you know, Google R& D. 00:15:57.415 --> 00:16:00.245 All the incentives nowadays are very differently aligned. 00:16:00.295 --> 00:16:01.975 And so I don't think that's going to happen there. 00:16:02.195 --> 00:16:04.445 It's, it's probably not going to be the Media Lab. 00:16:04.820 --> 00:16:07.420 So I actually don't know where those experiments are going to come from. 00:16:07.420 --> 00:16:10.140 They probably have to come from founders, right, doing the work. 00:16:10.390 --> 00:16:11.070 Fraser Kelton: I would think so. 00:16:11.070 --> 00:16:13.200 But it's interesting to frame it like that. 00:16:13.500 --> 00:16:16.580 When you told me this initially, my response was, this 00:16:16.580 --> 00:16:18.370 happened in the labs, right? 00:16:19.120 --> 00:16:24.595 You come in and Aditya shows you a, uh, Tennis ball thing that's pixelated and 00:16:24.605 --> 00:16:26.445 you know, what's going on with this? 00:16:26.475 --> 00:16:30.065 But if you squint, you're like, okay, we're, we're a couple of cranks 00:16:30.385 --> 00:16:33.765 away from having, uh, Dolly, right? 00:16:33.885 --> 00:16:40.415 And there's so many weird, lovely things that are, percolating there. 00:16:40.415 --> 00:16:41.725 We just haven't seen it yet. 00:16:41.745 --> 00:16:45.855 We see it after it scales up and, and they've polished it or, or 00:16:45.855 --> 00:16:48.785 we don't, but you're saying at a, at a level even above that. 00:16:48.995 --> 00:16:50.085 Nabeel Hyatt (2): Yeah, I would push back at that. 00:16:50.185 --> 00:16:55.655 I, it was OpenAI, frankly, like when you joined OpenAI, right? 00:16:55.845 --> 00:17:00.255 But, but after, after the ChatGPT and before, right, when they're making Rubik's 00:17:00.255 --> 00:17:06.245 Cubes robots and things like that, I think once the ChatGPT gun goes off, A 00:17:06.245 --> 00:17:09.545 lot of it is, it's not that there isn't advanced research, and not that people 00:17:09.545 --> 00:17:13.535 aren't doing amazing, crazy, interesting things, but it's all in service of 00:17:13.535 --> 00:17:15.235 something that's relatively deterministic. 00:17:16.075 --> 00:17:20.455 And, and the problem with the corporate labs, the large ones like Google, is, 00:17:20.495 --> 00:17:24.325 we've had this conversation before I'm trying to get my paper published. 00:17:25.525 --> 00:17:28.965 And that's also a very, very different incentive than if you 00:17:28.975 --> 00:17:32.345 think about what's going on inside of Xerox PARC at that point in time. 00:17:32.795 --> 00:17:33.135 Fraser Kelton: Yeah. 00:17:33.315 --> 00:17:36.295 You don't think there's the same crazy shenanigans happening 00:17:36.295 --> 00:17:37.325 and we just haven't seen it? 00:17:37.555 --> 00:17:43.225 There's not the equivalent of Sora 3 that you can squint and 00:17:43.225 --> 00:17:46.255 see, but we're not going to see it in the world for five years? 00:17:46.555 --> 00:17:47.655 Nabeel Hyatt (2): It's a good challenge. 00:17:47.935 --> 00:17:53.385 I would say there's probably a little bit happening at the edges of Google R& D. 00:17:53.385 --> 00:17:53.425 Yeah. 00:17:53.810 --> 00:17:55.000 at least a year ago. 00:17:55.840 --> 00:18:00.830 I'm often giving Google a hard time for maybe even being the Xerox PARC of this 00:18:00.840 --> 00:18:04.450 era in that they like invented all this technology and then we'll see if they ever 00:18:04.470 --> 00:18:06.340 effectively bring any of it to market. 00:18:06.660 --> 00:18:11.220 But, look, in the world of consolidation, where everything's moving together at 00:18:11.220 --> 00:18:14.700 Google I kind of want them to consolidate so that they can hopefully execute 00:18:14.720 --> 00:18:18.380 really well so that we can actually see Google as a good citizen in this time. 00:18:18.550 --> 00:18:22.070 set of product surface areas and, and invent the future of AI with 00:18:22.070 --> 00:18:26.590 everybody else instead of just kind of right, um, making demos, but the 00:18:26.650 --> 00:18:30.270 downside of that is, and we've kind of had this experience, we interview 00:18:30.270 --> 00:18:33.240 people that are coming out of Google, what kind of research did you do there? 00:18:33.250 --> 00:18:39.650 Since you're joining XYZ new startup, it is thinning in its aperture. 00:18:39.955 --> 00:18:41.085 not widening, right? 00:18:41.085 --> 00:18:41.295 Right. 00:18:41.325 --> 00:18:43.025 It's narrowing in its aperture, right? 00:18:43.025 --> 00:18:47.065 It's not widening that Google has targets now that they're going after. 00:18:47.255 --> 00:18:49.735 And so they're going to do less of that work over time. 00:18:49.925 --> 00:18:51.905 And so this is me just kind of like raising a hand and being 00:18:51.905 --> 00:18:55.295 like, I understand granola is not like the most mind blowing, 00:18:55.325 --> 00:18:58.265 crazy, you know, it's not Sora. 00:18:58.715 --> 00:19:00.115 But that's kind of my point. 00:19:00.125 --> 00:19:03.385 I think there's still a crazy amount of innovation to happen even just 00:19:03.385 --> 00:19:08.175 at the UI and UX and product layer, not at the deep research algo layer. 00:19:09.135 --> 00:19:12.945 There's nothing crazy technical about inventing the GUI, the 00:19:12.945 --> 00:19:14.115 graphical user interface. 00:19:14.205 --> 00:19:16.965 The first time you ever saw a window you could move around with a mouse. 00:19:17.545 --> 00:19:19.685 There's actually nothing crazy technical about the mouse 00:19:19.705 --> 00:19:20.715 when it was first invented. 00:19:21.325 --> 00:19:26.280 It is thinking about the product experience and then developing technology 00:19:26.280 --> 00:19:29.320 that maps to the product experience that we think we want to have with this new 00:19:29.350 --> 00:19:30.710 technology that's entered our lives. 00:19:30.950 --> 00:19:33.840 I think we're going to do great at new fancy algorithms. 00:19:33.950 --> 00:19:34.260 Fraser Kelton: Yeah. 00:19:34.390 --> 00:19:36.740 Nabeel Hyatt (2): I'm less certain that we have the right 00:19:36.740 --> 00:19:39.150 pockets in technology right now. 00:19:39.885 --> 00:19:43.845 Innovate on new product experiences because everybody's chasing growth 00:19:44.135 --> 00:19:47.075 and we have been organized to think about customer development 00:19:47.215 --> 00:19:49.435 and MVPs and up into the right. 00:19:49.435 --> 00:19:49.665 Yep. 00:19:49.665 --> 00:19:52.115 So I can raise my billion dollar round tomorrow and all the rest of that stuff. 00:19:52.125 --> 00:19:52.355 Yep. 00:19:52.355 --> 00:19:52.575 Yep. 00:19:52.575 --> 00:19:52.725 Yep. 00:19:52.745 --> 00:19:54.145 And I'm not that I'm not against growth. 00:19:54.155 --> 00:19:57.225 Like eventually we're in the venture scale business, but it's, we're just early. 00:19:57.265 --> 00:20:01.365 And we like it's just, there's the room for weirdness is important. 00:20:01.515 --> 00:20:02.445 Fraser Kelton: I get all of that. 00:20:02.475 --> 00:20:04.735 I was just sitting with the founder that I work with. 00:20:06.050 --> 00:20:12.080 And he's exceptionally early in a market that's early and he's iterating, you know, 00:20:12.460 --> 00:20:13.990 uh, toward a product that's resonating. 00:20:14.010 --> 00:20:18.550 And my feedback to him was like, that sounds good, but, At this moment in time, 00:20:18.590 --> 00:20:22.390 I would make sure that you're taking large swings adjacent to where you're 00:20:22.390 --> 00:20:25.840 currently going to make sure that you're not marching up a local maxima, right? 00:20:25.900 --> 00:20:26.260 I agree. 00:20:26.450 --> 00:20:28.000 There's so much stuff going on. 00:20:28.090 --> 00:20:28.470 I don't know. 00:20:28.470 --> 00:20:32.350 I feel like we have to give permission to be creative at times. 00:20:32.900 --> 00:20:34.780 We as in the tech industry. 00:20:34.800 --> 00:20:35.040 Yeah. 00:20:35.260 --> 00:20:36.220 And we have to be patient. 00:20:36.430 --> 00:20:41.090 I think that's a big thing though is if you were to see a demo of 00:20:41.210 --> 00:20:44.480 that type of stuff now, think about the world's response. 00:20:44.480 --> 00:20:44.540 Yeah. 00:20:44.725 --> 00:20:47.485 Nabeel Hyatt (2): Yeah, well there's, there's a world where some of those people 00:20:47.485 --> 00:20:51.385 are trying to do some of those demos and maybe the world just says this is fake. 00:20:51.740 --> 00:20:57.590 Maybe the world says this isn't shippable and, you know, this founder is full of it. 00:20:58.330 --> 00:21:02.340 Fraser Kelton: Yeah listen, we've seen that exceptionally loudly on Twitter. 00:21:02.710 --> 00:21:06.890 You shipped a waitlist, you shipped a demo, you shipped there's a 00:21:06.950 --> 00:21:11.035 cacophony of voices screaming at you, and you have to be patient. 00:21:11.345 --> 00:21:12.915 Twitter's not, one, 00:21:12.915 --> 00:21:16.965 Nabeel Hyatt (2): Twitter's not reality, uh, two, all we, all you really ask when 00:21:16.965 --> 00:21:20.825 you're shipping a demo or talking about a concept is that you be transparent. 00:21:20.825 --> 00:21:21.535 It's a concept. 00:21:22.285 --> 00:21:25.465 There's nothing wrong with talking about a crazy idea. 00:21:25.905 --> 00:21:28.575 The problem with talking about a crazy idea is presenting 00:21:28.575 --> 00:21:30.165 it as shipping in 30 days. 00:21:30.385 --> 00:21:31.745 Just give me your a hundred dollars. 00:21:31.795 --> 00:21:32.315 Right. 00:21:32.425 --> 00:21:33.135 And it's ready. 00:21:33.175 --> 00:21:34.185 Right, right. 00:21:34.405 --> 00:21:35.265 That's the problem. 00:21:35.670 --> 00:21:39.570 There's nothing wrong with making claims and being highfalutin and, and 00:21:39.580 --> 00:21:41.900 being, you know, you can be Walt Disney. 00:21:41.910 --> 00:21:44.620 Just make sure you everybody understands that you're 00:21:44.620 --> 00:21:46.480 dreaming at this moment, right? 00:21:47.220 --> 00:21:48.080 Um, yep. 00:21:48.180 --> 00:21:51.010 I do think it is really hard to be patient though, Fraser I agree. 00:21:51.140 --> 00:21:55.340 It's very hard to keep iterating on a thing to really get to new things, 00:21:55.360 --> 00:22:01.210 especially When all of your competitors are probably chasing a little bit of 00:22:01.210 --> 00:22:06.110 traction and are probably raising a bunch of money And so you know, you're trying 00:22:06.110 --> 00:22:09.960 to iterate on refining the product and getting it perfect and you're rethinking 00:22:09.960 --> 00:22:14.080 You're like throwing out the UX for the third time to really make it work 00:22:14.210 --> 00:22:17.120 And then you wake up one morning and you open up TechCrunch and somebody 00:22:17.120 --> 00:22:19.620 else is doing something orthogonal It's raised just raised a hundred and fifty 00:22:19.620 --> 00:22:23.560 million dollars or whatever the number is like, you know, this is we as a venture 00:22:23.560 --> 00:22:28.190 community are part and parcel of in trying to solve our own problems on a monthly 00:22:28.190 --> 00:22:29.670 basis and get our, you know, whoever. 00:22:30.485 --> 00:22:32.225 principle at some firm is trying to get a promotion. 00:22:32.225 --> 00:22:36.395 So he's trying to jump on the hot thing that week is is certainly makes 00:22:36.395 --> 00:22:39.175 it harder to build really good product and make really good decisions. 00:22:39.175 --> 00:22:39.935 No question, right? 00:22:40.895 --> 00:22:44.465 Fraser Kelton: I also think that personal incentives are incredibly 00:22:44.465 --> 00:22:47.435 important in everything like full stop. 00:22:47.605 --> 00:22:51.955 And I think the structure of performance reviews and yearly perf 00:22:52.005 --> 00:22:55.555 packages that people pull together, I think it'd be almost impossible 00:22:55.555 --> 00:22:59.655 in that framework to find the structure that rewards that level of. 00:23:00.330 --> 00:23:02.260 Exploration, right? 00:23:02.260 --> 00:23:06.160 Like it's, it's probably, probably doable, but it would take even at 00:23:06.160 --> 00:23:11.050 the org level, some real careful consideration and sculpting. 00:23:11.290 --> 00:23:11.590 Nabeel Hyatt (2): Yeah. 00:23:11.590 --> 00:23:14.690 The, the, the old adage from Charlie Munger, show me the incentive 00:23:14.690 --> 00:23:15.680 and I'll show you the outcome. 00:23:15.850 --> 00:23:16.260 Fraser Kelton: Oh, for sure. 00:23:16.260 --> 00:23:17.390 A hundred percent, right? 00:23:17.420 --> 00:23:18.290 Yeah, yeah, yeah. 00:23:18.290 --> 00:23:22.190 You know, talking about, um, waking up and seeing the headline 00:23:22.190 --> 00:23:25.360 that, that your competitors raised hundreds of millions of dollars. 00:23:25.750 --> 00:23:31.760 That is a topic that has been on my mind for a while now, um, and 00:23:31.760 --> 00:23:32.960 I'm sure it's been on yours, too. 00:23:33.140 --> 00:23:36.360 Nabeel Hyatt (2): Yeah, it's impossible to not be operating in the AI market 00:23:36.430 --> 00:23:39.470 and trying to be either an employee in a startup, a CEO of a startup, an 00:23:39.470 --> 00:23:41.630 investor in a startup, and not wake up. 00:23:41.780 --> 00:23:44.970 On both sides of this equation, at some point in your life, you're going 00:23:44.970 --> 00:23:48.810 to wake up, and you're going to feel great about having raised a bunch more 00:23:48.810 --> 00:23:51.040 money than somebody else, and you're trying to figure out how to use it, 00:23:51.070 --> 00:23:54.720 and we certainly have those companies, and then inevitably, in the kind of 00:23:54.720 --> 00:23:59.680 rollercoaster of things, you're going to wake up one day, And there's some 00:24:00.090 --> 00:24:04.800 headline of somebody who's raised a billion dollars at, at 10 billion or 100 00:24:04.800 --> 00:24:06.960 billion at, at, you know, whatever it is. 00:24:06.980 --> 00:24:09.750 It's 10x more than you've got, that's for sure. 00:24:10.140 --> 00:24:12.650 And then they call you, Fraser, I know you've had this 00:24:12.650 --> 00:24:13.630 happen, I've had this happen. 00:24:13.690 --> 00:24:17.520 What do you say to somebody who calls and says, man, they just raised that 00:24:17.520 --> 00:24:21.465 much money off in the, Is they're the same size as us or even smaller, 00:24:21.465 --> 00:24:22.915 or like, how did that happen? 00:24:23.225 --> 00:24:26.785 What do you say to a founder and how do you think a team, not even just a 00:24:26.785 --> 00:24:30.195 founder, how do you think a team should internalize when there's money flung 00:24:30.195 --> 00:24:33.715 around in that kind of like haphazard way that that may or may not feel 00:24:33.715 --> 00:24:34.925 like it just took you out that day? 00:24:35.140 --> 00:24:40.020 Fraser Kelton: I think any good piece of discussion around startups, uh, there's 00:24:40.030 --> 00:24:44.400 way too much nuance for us to have a tight, solid piece of wisdom here. 00:24:44.400 --> 00:24:45.220 Nabeel Hyatt (2): It's not three headlines. 00:24:45.240 --> 00:24:46.830 It's not the three steps to the epiphany. 00:24:48.600 --> 00:24:51.070 Fraser Kelton: It, uh, because listen, we went through an era where 00:24:51.070 --> 00:24:53.925 software was like, Bank thought that they could overcapitalize every 00:24:53.925 --> 00:24:56.475 company and crown the market winner. 00:24:56.535 --> 00:24:59.355 And I think that we've seen that that experiment didn't work out. 00:24:59.745 --> 00:25:02.865 Nabeel Hyatt (2): You know like all things, it didn't work in 20 places 00:25:02.865 --> 00:25:03.885 except for the place where it worked. 00:25:03.890 --> 00:25:05.415 Like it really worked for DoorDash. 00:25:05.445 --> 00:25:05.805 Right. 00:25:05.985 --> 00:25:07.725 Like that it absolutely worked. 00:25:07.785 --> 00:25:09.135 And then there was like, right. 00:25:09.135 --> 00:25:11.805 You know, a MA and probably maybe one other place, but the, yeah. 00:25:11.835 --> 00:25:14.565 Capital creates the winner was a graveyard otherwise. 00:25:14.565 --> 00:25:14.835 Right, 00:25:14.840 --> 00:25:15.180 Fraser Kelton: right. 00:25:15.755 --> 00:25:16.045 Right. 00:25:16.065 --> 00:25:19.415 It just incinerated a bunch of capital and an awful lot of startups. 00:25:19.465 --> 00:25:19.665 Yeah. 00:25:19.765 --> 00:25:24.495 And I feel that in my bones, like one thing that I think is, is like the 00:25:24.525 --> 00:25:29.315 cardinal sin of a founder, and I felt it when I was a founder, was overcapitalizing 00:25:29.325 --> 00:25:33.515 and celebrating the amount that you raised or your, your valuation. 00:25:33.925 --> 00:25:37.785 Because I think that, I don't know, it's probably not causation, but it's 00:25:37.785 --> 00:25:42.755 pretty darn close, um, to that, right, is, is, it feels, I can't think of 00:25:42.755 --> 00:25:44.935 one example where that's the case. 00:25:45.215 --> 00:25:49.295 You have on one end of the, the, the ledger, uh, a situation where over 00:25:49.295 --> 00:25:55.215 capitalization kills companies, the idea that capital can crown the king 00:25:55.285 --> 00:26:01.015 also really didn't work out, and yet I feel like in many markets right now, 00:26:01.670 --> 00:26:06.780 Uh, you will lose your edge if you're not at least in the same ballpark. 00:26:07.110 --> 00:26:11.250 If you're a company that's raised 20 million and you have competitors that 00:26:11.250 --> 00:26:18.150 have raised 150, 250, 400 million, if you're trying to compete for the same 00:26:18.570 --> 00:26:23.100 large enterprise customer, if you're trying to compete for the same researcher 00:26:23.100 --> 00:26:28.430 who's been at DeepMind for five, six years, I do think that you're going to be 00:26:28.430 --> 00:26:30.470 perceived very differently in the market. 00:26:30.850 --> 00:26:32.530 Um, with that type of a split. 00:26:32.590 --> 00:26:36.740 And then I think that over time, uh, it's going to have a real effect, a 00:26:36.740 --> 00:26:38.520 negative effect on, on the company. 00:26:38.770 --> 00:26:40.070 Nabeel Hyatt (2): I think that's somewhat true. 00:26:40.090 --> 00:26:42.860 I think it's worth digging in and giving a little bit of nuance back to 00:26:42.860 --> 00:26:44.400 a team that's trying to process this. 00:26:44.460 --> 00:26:47.620 Like, when you're thinking about larger competitors, I just think it's important 00:26:47.630 --> 00:26:50.560 to keep your customer in mind, and it all roots back to your customer. 00:26:50.580 --> 00:26:52.585 Your customer is going to make your success, not you. 00:26:53.145 --> 00:26:54.735 Not a VC and not your bank account. 00:26:55.125 --> 00:26:58.305 And so sometimes the money matters because your customer matters. 00:26:59.065 --> 00:27:02.185 Sometimes the money matters because it matters to your customer. 00:27:02.225 --> 00:27:02.255 Fraser Kelton: Not, 00:27:03.415 --> 00:27:05.355 Nabeel Hyatt (2): it's often that not because. 00:27:06.635 --> 00:27:11.635 It is a situation where it made the next engineering hire happen or not. 00:27:12.015 --> 00:27:15.725 I started a company a long time ago called Conduit Labs. 00:27:15.905 --> 00:27:19.205 About nine months into its life, it pivoted into Facebook games as the 00:27:19.205 --> 00:27:22.275 Facebook app platform opened up, and that led to the acquisition by Zynga. 00:27:22.585 --> 00:27:25.035 But early on, we were building virtual worlds in this kind 00:27:25.035 --> 00:27:26.475 of post second life world. 00:27:26.935 --> 00:27:34.055 And I just remember the product we were working on, MTV funded and 00:27:34.065 --> 00:27:36.475 built a competitor to that product. 00:27:36.975 --> 00:27:39.315 And I was freaking out. 00:27:39.645 --> 00:27:43.345 And I remember talking to Alex Rogopoulos, who's the CEO of Harmonix, guys who made 00:27:43.345 --> 00:27:45.165 guitar here in rock band at the time. 00:27:45.665 --> 00:27:49.585 And I just remember the advice he was giving me, which is just right 00:27:49.585 --> 00:27:51.265 now you are all default debt. 00:27:51.345 --> 00:27:53.265 I don't really care how much money comes in. 00:27:53.545 --> 00:27:58.305 The default process is that whatever they just come up with is just not good enough. 00:27:59.085 --> 00:28:05.335 And so don't stare at them with their money and be covetous stare 00:28:05.335 --> 00:28:08.315 at their product and be covetous if they've come up with the right thing. 00:28:08.935 --> 00:28:13.525 And, uh, you know, the similar thing happened, Ambient Devices, a 00:28:13.555 --> 00:28:16.455 previous company I started that was a spin out from the MIT Media Lab. 00:28:16.855 --> 00:28:17.305 I remember LG. 00:28:18.575 --> 00:28:21.935 After we'd had extensive conversations about licensing agreements, a couple six 00:28:21.935 --> 00:28:26.385 months later, they started, they launched their Andean Computing Initiative at LG. 00:28:26.615 --> 00:28:28.905 And we were like, oh man, big guy. 00:28:29.235 --> 00:28:31.395 You know, it just turned out the whole category wasn't a category. 00:28:31.395 --> 00:28:32.345 It wasn't a massive category. 00:28:32.925 --> 00:28:34.285 So it's fine. 00:28:34.900 --> 00:28:40.060 You know, I think Cruise is another great example where, man, you 00:28:40.060 --> 00:28:43.300 know, that was a, a capital war. 00:28:43.580 --> 00:28:49.400 You had Google, uh, with Waymo budgets, after Cruise raised, 00:28:49.430 --> 00:28:53.470 not long later, Aurora raised, Chris Armisen, um, became the CEO. 00:28:53.770 --> 00:28:56.980 You know, he had, he headed the self driving, uh, car team at Alphabet 00:28:56.980 --> 00:29:01.535 for seven years and then it turned out, by the way, that I don't 00:29:01.535 --> 00:29:03.065 know, Cruz was the first to market. 00:29:03.065 --> 00:29:04.945 We'll see what GM does with it now. 00:29:05.325 --> 00:29:07.315 Waymo is another viable player. 00:29:07.625 --> 00:29:11.935 Aurora, we'll see what happens, but certainly even with the war chest that 00:29:11.935 --> 00:29:16.445 they built, didn't immediately make them winners in the market, that's for sure. 00:29:16.935 --> 00:29:21.065 So, I don't know, I think Capital, I think of it three ways. 00:29:21.705 --> 00:29:27.795 One is, if it's day zero and you're an overfunded pre product market company, 00:29:28.545 --> 00:29:32.905 capital is, is oxygen, and you could absolutely die by oxygenization. 00:29:34.785 --> 00:29:35.185 Fraser Kelton: Right. 00:29:35.205 --> 00:29:35.725 Yes. 00:29:36.115 --> 00:29:36.405 Good. 00:29:36.775 --> 00:29:39.405 Nabeel Hyatt (2): And I remember when Thalmic Labs raised a hundred million 00:29:39.405 --> 00:29:42.985 dollar round, uh, years ago, they're an early frontier tech startup we backed. 00:29:42.985 --> 00:29:47.395 You know, I remember showing at that point in time, the list of other hundred 00:29:47.395 --> 00:29:51.425 million dollar pre launch companies That it existed, like literally every 00:29:51.425 --> 00:29:58.625 single one that it existed, and it was a unbelievable quibby like graveyard 00:29:58.965 --> 00:30:01.325 of companies that did not work out. 00:30:01.775 --> 00:30:04.965 There's a bunch of reasons we need to go into now about why I believe that is true, 00:30:05.255 --> 00:30:08.245 but it's not categorically true, right? 00:30:08.305 --> 00:30:09.435 But it's pretty rare. 00:30:09.820 --> 00:30:13.740 And hard to raise that kind of capital and still keep a disciplined, fast 00:30:13.740 --> 00:30:15.710 moving, execution oriented startup. 00:30:15.710 --> 00:30:16.380 A hundred percent. 00:30:16.590 --> 00:30:18.080 Yeah, and keep up with everybody else. 00:30:18.090 --> 00:30:21.660 So sometimes a bunch of money into somebody else can be a gift. 00:30:22.210 --> 00:30:24.840 The second bit, which you were getting to, I'd say on a large, on 00:30:24.840 --> 00:30:27.930 a large raise, or somebody who's doing it is, is who is your customer? 00:30:28.270 --> 00:30:33.135 If your customer is a consumer I don't know, a hundred million dollars, five 00:30:33.135 --> 00:30:35.685 hundred million dollars in the bank, like it doesn't change your retention curves. 00:30:35.715 --> 00:30:36.445 Nobody cares. 00:30:36.495 --> 00:30:36.965 At all. 00:30:37.045 --> 00:30:37.255 That's right. 00:30:37.255 --> 00:30:39.125 Your customers barely pay attention. 00:30:39.145 --> 00:30:40.065 They don't really know. 00:30:40.285 --> 00:30:41.655 I wouldn't worry too much about it. 00:30:41.985 --> 00:30:44.695 If anything, again, it might handicap other people. 00:30:45.035 --> 00:30:53.635 If your customer is I don't know, some old money water company in France 00:30:53.815 --> 00:30:59.255 that wants to buy the safest thing in the planet, Then you should worry. 00:30:59.315 --> 00:30:59.655 Fraser Kelton: Yeah. 00:30:59.775 --> 00:31:00.045 Nabeel Hyatt (2): Right? 00:31:00.075 --> 00:31:04.975 If your, if your customer is very safety oriented, late adopter type, 00:31:05.275 --> 00:31:08.800 who just wants safety, then them looking at a startup that has a 00:31:08.800 --> 00:31:12.110 10 times higher bank account than you, it will affect your business. 00:31:12.310 --> 00:31:14.670 And you better find a way to fight the capital game. 00:31:14.860 --> 00:31:19.040 Or at least, as you said earlier, at least be within an order of magnitude. 00:31:19.310 --> 00:31:19.660 Right? 00:31:19.690 --> 00:31:20.200 Right. 00:31:20.610 --> 00:31:21.640 Yeah, same universe. 00:31:21.650 --> 00:31:22.970 In the same universe. 00:31:22.970 --> 00:31:24.570 You don't have to play a capital war. 00:31:24.590 --> 00:31:24.880 That's it. 00:31:24.880 --> 00:31:28.440 But you better have the same number of zeros as the other person. 00:31:28.720 --> 00:31:31.530 Or else, again, it's, it's, but again, that's not really 00:31:31.530 --> 00:31:32.500 about the money in the bank. 00:31:32.720 --> 00:31:36.830 That is about the amount of capital you have is. 00:31:37.775 --> 00:31:42.715 a feature of your company when you're trying to sell to a large organization. 00:31:42.725 --> 00:31:43.065 Fraser Kelton: For sure. 00:31:43.095 --> 00:31:45.335 Nabeel Hyatt (2): It's a part of your product, strangely enough. 00:31:45.345 --> 00:31:45.785 Fraser Kelton: For sure. 00:31:45.875 --> 00:31:47.365 Nabeel Hyatt (2): But that's very, that's very rare. 00:31:47.365 --> 00:31:52.205 Most of the times, most AI companies, are not trying to sell to late 00:31:52.205 --> 00:31:55.395 adopters kind of by necessity. 00:31:56.015 --> 00:32:00.185 If you're selling to a fear induced late adopter, first of all, thank you. 00:32:00.205 --> 00:32:02.875 Elephant hunting, whale hunting startup, it's hard. 00:32:02.875 --> 00:32:03.185 Right. 00:32:03.225 --> 00:32:03.745 Impossible. 00:32:03.775 --> 00:32:06.335 And it's going to be slow and it's going to be painful. 00:32:06.555 --> 00:32:09.085 You should probably see if you have early adopters you can sell to first. 00:32:09.225 --> 00:32:10.095 That's the nature of things. 00:32:10.105 --> 00:32:10.445 Right. 00:32:10.505 --> 00:32:13.815 But yeah, and then there will be the few situations just like there's no playbook, 00:32:13.975 --> 00:32:16.695 there's no three headlines to get you through, but there will be some situations 00:32:16.695 --> 00:32:18.950 where You know, you are in a capital war. 00:32:19.250 --> 00:32:20.130 That's just the bottom line. 00:32:21.330 --> 00:32:23.490 Fraser Kelton: The other thing that we've heard from, from very 00:32:23.490 --> 00:32:25.610 capable founders who tell us. 00:32:26.090 --> 00:32:31.180 that they know better, is that the types of talent that can make a 00:32:31.180 --> 00:32:37.020 company in this moment, and I'm talking about applied researchers or product 00:32:37.020 --> 00:32:44.760 focused researchers, are valuing total comp in the moment rather than 00:32:45.110 --> 00:32:47.240 the opportunity for appreciation. 00:32:47.980 --> 00:32:49.945 And we've met a couple of really great people. 00:32:50.405 --> 00:32:54.735 Really strong founders that I respect who then say, listen, I know, I know, I know, 00:32:55.335 --> 00:32:59.015 but to hire the three types of people that I need to hire over the next 18 00:32:59.015 --> 00:33:04.125 months, uh, to make the company, they're going to evaluate it on that vector. 00:33:04.155 --> 00:33:08.765 And so it's important that I'm again in the universe of valuation. 00:33:08.935 --> 00:33:13.595 Nabeel Hyatt (2): Yeah, there was a tweet a couple of weeks ago that 00:33:13.595 --> 00:33:21.280 was like, If you're interviewing at OpenAI and you're interviewing at a 00:33:21.280 --> 00:33:25.870 startup, then the startup should just stop interviewing you immediately 00:33:25.870 --> 00:33:26.780 or something along those lines. 00:33:27.130 --> 00:33:29.610 Fraser Kelton: It was, if you're a startup and you're interviewing 00:33:29.610 --> 00:33:32.580 somebody who's also interviewing at big tech, you should just stop. 00:33:33.535 --> 00:33:33.975 Nabeel Hyatt (2): That's right. 00:33:34.005 --> 00:33:36.655 Fraser Kelton: They're like a tourist who's going on a little side journey 00:33:36.655 --> 00:33:40.485 to understand, uh, how interesting it might be in a startup, but 00:33:40.485 --> 00:33:41.545 they're never going to come to you. 00:33:41.595 --> 00:33:44.075 Nabeel Hyatt (2): I will push back slightly on that, and I did in that 00:33:44.075 --> 00:33:47.535 conversation, which is that there are some people who actually don't 00:33:47.545 --> 00:33:53.945 really know the difference between joining OpenAI or joining Google and 00:33:53.945 --> 00:33:56.885 joining a startup, like there are some people that are just earlier in their 00:33:56.885 --> 00:34:00.995 educational journey about what that choice really means and the pluses and 00:34:00.995 --> 00:34:04.615 minuses of the choices of being a big tech or being in a startup environment. 00:34:04.965 --> 00:34:07.755 And so sometimes it's worth explaining to people if you 00:34:07.755 --> 00:34:10.495 haven't explained to people the. 00:34:10.905 --> 00:34:16.665 Appreciation of stock and the amount of change and responsibility and 00:34:16.665 --> 00:34:21.905 excitement and agency that comes with being in a startup versus working in a 00:34:21.905 --> 00:34:26.705 lab at Google, then shame on you as a founder, and there are plenty of people, 00:34:26.705 --> 00:34:28.005 especially if you haven't grown up. 00:34:28.330 --> 00:34:31.620 in Silicon Valley, there's still plenty of people on the planet, even those 00:34:31.620 --> 00:34:34.770 who've been working for many years, who don't really understand startup culture. 00:34:34.770 --> 00:34:36.890 They understand technology culture, but they don't really 00:34:36.890 --> 00:34:37.830 understand startup culture. 00:34:37.940 --> 00:34:41.410 So I think you have to explain to them what it means to join a startup, and what 00:34:41.410 --> 00:34:44.120 that means from a compensation standpoint, and also that, frankly, what it means from 00:34:44.120 --> 00:34:45.950 a working and responsibility standpoint. 00:34:46.390 --> 00:34:51.050 And then, yeah, if at this point, you're haggling over those kinds of 00:34:51.050 --> 00:34:56.160 choices between OpenAI or a startup they should probably just first pick 00:34:56.375 --> 00:35:00.025 big company or startup right before they even get to the name of the thing 00:35:00.085 --> 00:35:03.275 right because they're just wildly different paths like it's You know, 00:35:03.295 --> 00:35:05.335 it is like different careers in a way. 00:35:05.525 --> 00:35:07.075 Fraser Kelton: Listen, I get all of that. 00:35:07.195 --> 00:35:09.325 The market's telling us something different right now. 00:35:09.375 --> 00:35:12.005 The market might be irrational in the moment. 00:35:12.005 --> 00:35:12.535 We've seen that. 00:35:12.535 --> 00:35:16.205 But these are, these are like seasoned, capable founders who are 00:35:16.205 --> 00:35:19.745 telling us that they need to play that game to make the couple of hires 00:35:19.745 --> 00:35:20.785 that are going to make the company. 00:35:20.795 --> 00:35:21.965 Nabeel Hyatt (2): No, I just disagree. 00:35:22.005 --> 00:35:23.195 I like, I just disagree. 00:35:23.365 --> 00:35:27.145 Like you, that person, because the problem is that person is coin operated. 00:35:27.405 --> 00:35:30.145 That person wants whatever the highest comp package for that year is. 00:35:30.525 --> 00:35:33.325 And what that means is that they are going to join you and you're going 00:35:33.325 --> 00:35:36.235 to go through all this integration and they're going to get work done. 00:35:36.265 --> 00:35:39.195 And then six months later, somebody's going to give them a signing bonus 00:35:39.205 --> 00:35:40.475 and they're going to be gone anyway. 00:35:40.585 --> 00:35:42.845 Everything in my bones tells me you're right. 00:35:44.265 --> 00:35:44.555 Fraser Kelton: Yeah. 00:35:44.785 --> 00:35:45.145 Right? 00:35:45.335 --> 00:35:46.725 Everything in my bones tells me you're right. 00:35:46.725 --> 00:35:46.965 Yeah. 00:35:47.085 --> 00:35:50.425 No, I think that the challenge that I'm wrestling with is I've seen firsthand 00:35:50.425 --> 00:35:55.040 how one person One person can breathe life into something in this moment. 00:35:55.070 --> 00:35:58.270 That is, that is, uh, ethereal. 00:35:58.280 --> 00:36:01.160 Like it's, it's not like you're hiring a 10 X engineer who's going 00:36:01.160 --> 00:36:05.180 to come in and you're fighting for that type of incremental, like 00:36:05.190 --> 00:36:07.100 efficiency on your engineering team. 00:36:07.100 --> 00:36:10.370 And you're going to ship your like backend data pipelines a little bit faster. 00:36:10.390 --> 00:36:10.650 Yeah. 00:36:10.720 --> 00:36:13.060 I think we're sitting in a moment where there, there are, 00:36:13.100 --> 00:36:15.140 Nabeel Hyatt (2): it's like signs of life is what your point is. 00:36:15.160 --> 00:36:15.590 Like it's 00:36:15.780 --> 00:36:17.270 Fraser Kelton: signs of life from. 00:36:17.270 --> 00:36:17.495 Yeah. 00:36:17.875 --> 00:36:22.795 Uh, a very scarce number of people who can control the, the, the magic, um, 00:36:22.815 --> 00:36:25.435 and, and, and grant it to your company. 00:36:25.775 --> 00:36:30.445 If, if you are really going to push the frontier of AI research in a 00:36:30.445 --> 00:36:32.995 domain and in an applied setting. 00:36:33.595 --> 00:36:37.095 With that research, I don't know, like I've seen Alec Radford, 00:36:37.295 --> 00:36:39.305 like this guy is amazing. 00:36:39.615 --> 00:36:42.645 You probably should, you know, put yourself in a position where you can get 00:36:42.645 --> 00:36:45.265 the N of 4 in the world into your company. 00:36:45.545 --> 00:36:48.625 I think the challenge then is if that's true, people misconstrue that and do 00:36:48.625 --> 00:36:52.085 that same playbook for the backend engineer where it really is immaterial. 00:36:52.385 --> 00:36:54.685 Nabeel Hyatt (2): The trend I have felt in the last three or four 00:36:54.685 --> 00:36:58.525 or five months by folks who've been in AI for a long time is. 00:36:59.180 --> 00:37:04.020 I only want one of those deep researchers at most and, and 00:37:04.020 --> 00:37:05.870 me as the founder or the CTO. 00:37:06.365 --> 00:37:07.285 is that one. 00:37:07.835 --> 00:37:13.395 And I thought I needed 10, and the truth is I, I, now, now, I'm just gonna train 00:37:13.395 --> 00:37:16.565 some really good ML people internally, and I will get 99 percent of the way 00:37:16.565 --> 00:37:20.745 there on every single thing I want to do, uh, and frankly, they'll be more 00:37:20.745 --> 00:37:24.965 product oriented, more consumer oriented you know, uh, uh, uh, two years ago 00:37:24.965 --> 00:37:29.655 and a year ago, the kind of prevailing thought process was, I want a really 00:37:29.665 --> 00:37:35.750 hardcore world class research team on top of a world class engine product team. 00:37:35.990 --> 00:37:38.940 And I need to build both of those and keep those dualities together. 00:37:39.280 --> 00:37:42.940 And I, and we've both seen that the trend for the best performing 00:37:42.940 --> 00:37:48.700 companies now has been incredibly slim to almost no research team. 00:37:49.250 --> 00:37:52.310 Maybe, again, one to two people, often the founders. 00:37:52.370 --> 00:37:52.740 Right. 00:37:52.890 --> 00:37:56.505 And so, You know, again, unless you're Anthropic. 00:37:56.565 --> 00:38:00.835 Anthropic has a reason to have a massive research team to be in, you know, a 00:38:00.835 --> 00:38:02.465 state of the art large language model. 00:38:02.845 --> 00:38:05.385 Fraser Kelton: Yeah, this is, this is where so much nuance matters on 00:38:05.385 --> 00:38:07.155 the decision because I know, man. 00:38:07.165 --> 00:38:07.765 Yeah, I know. 00:38:07.875 --> 00:38:10.915 I think I agree with you for when you're starting out. 00:38:10.965 --> 00:38:15.060 I think that we have undoubtedly seen in the market that if you go try to 00:38:15.060 --> 00:38:19.350 get like your merry men of researchers together to go like gallivant 00:38:19.350 --> 00:38:23.910 around and start a company, I don't think that usually works out well. 00:38:23.950 --> 00:38:29.660 You have a researcher, usually the founder, founding CTO, who's, 00:38:29.670 --> 00:38:33.620 who's pushing forward with a lot of product motivation behind it. 00:38:34.360 --> 00:38:39.665 There is a time though, when that company is No longer three people 00:38:39.665 --> 00:38:42.525 and just founding, they have a product and market, or they have 00:38:42.565 --> 00:38:46.755 the scaffolding of an organization around them, and they've started to 00:38:46.755 --> 00:38:50.115 scale a team, you can basically, I don't know, is like the equivalent 00:38:50.115 --> 00:38:55.835 of a two pizza team again, you can replicate that playbook multiple times. 00:38:56.250 --> 00:38:59.140 Depending on where you are in the arc of your company, I would take multiple people 00:38:59.140 --> 00:39:03.090 like that and build a team around them and, and let them, let them run rampant. 00:39:03.150 --> 00:39:03.360 Yeah. 00:39:03.370 --> 00:39:06.160 Like it, it is, I think, I think there's a moment in time where that can happen. 00:39:06.880 --> 00:39:10.530 So let's, let's bring it back is, uh, a founder comes, they've raised, 00:39:10.580 --> 00:39:12.650 let's keep it at tens of millions. 00:39:12.940 --> 00:39:17.820 They say, Hey, listen, this company just raised a billion dollars. 00:39:18.855 --> 00:39:20.945 They just launched the company. 00:39:20.965 --> 00:39:22.005 They haven't even launched the product. 00:39:22.005 --> 00:39:22.205 Right. 00:39:22.205 --> 00:39:24.135 They've introduced the company with the billion dollar round. 00:39:24.495 --> 00:39:28.955 And actually all the people on the team are not the founder ethos. 00:39:28.975 --> 00:39:32.305 They're, you know, gray haired, grizzled industry veterans. 00:39:32.865 --> 00:39:33.985 100 percent of that case. 00:39:33.985 --> 00:39:35.895 I'm like, you should just ignore it. 00:39:35.925 --> 00:39:37.925 Like you don't even pay any mind share. 00:39:37.925 --> 00:39:40.525 They've just killed the company and they don't even appreciate it. 00:39:40.845 --> 00:39:44.525 Um, I think the thing to be worried about is the company or at least 00:39:45.350 --> 00:39:49.790 inquisitive about is the company that's in your market in a market 00:39:49.800 --> 00:39:52.500 where your customers care about things. 00:39:52.830 --> 00:39:56.050 And, or there's a small number of people who can wield this magic 00:39:56.050 --> 00:40:01.130 today, um, and care about outsized comp rather than appreciation and 00:40:01.130 --> 00:40:03.420 have raised many multiples of you. 00:40:03.870 --> 00:40:06.630 Because I think that your, uh, you said product earlier. 00:40:06.630 --> 00:40:09.760 I think that's your, I think that's a reflection of your brand in the market. 00:40:09.760 --> 00:40:11.940 And I, I think that people. 00:40:12.425 --> 00:40:15.215 A lot of markets buy on brand. 00:40:15.355 --> 00:40:16.045 Nabeel Hyatt (2): Yeah, that's right. 00:40:16.295 --> 00:40:17.635 I think they do buy on brand. 00:40:17.645 --> 00:40:19.395 I don't think consumers buy on brand. 00:40:19.455 --> 00:40:19.825 No. 00:40:19.905 --> 00:40:22.995 I think even mid market to small business, if you're, don't buy, 00:40:23.135 --> 00:40:27.080 the people who buy on brand and safety are middle to late adopter. 00:40:27.130 --> 00:40:27.500 Yeah. 00:40:27.580 --> 00:40:31.210 Large, large elephant hunting enterprise customers, which is actually a 00:40:31.210 --> 00:40:34.800 vanishingly small number of startups actually have that customer base. 00:40:34.830 --> 00:40:35.320 Fraser Kelton: That's right. 00:40:36.350 --> 00:40:38.720 Nabeel Hyatt (2): And for everybody else, like we're 00:40:38.880 --> 00:40:39.990 glossing over a bunch of things. 00:40:39.990 --> 00:40:45.120 Of course, money matters, but there are a million different things that 00:40:45.120 --> 00:40:46.570 are going to kill you as a startup. 00:40:46.630 --> 00:40:50.390 And frankly, Most startups are going to die by suicide, not homicide. 00:40:50.390 --> 00:40:51.297 Fraser Kelton: And 00:40:51.297 --> 00:40:55.580 Nabeel Hyatt (2): so the staring out your window at everything else is just 00:40:55.590 --> 00:41:00.060 distracting you from the fact that it is get your culture right, get your product 00:41:00.060 --> 00:41:03.720 shipping right, make product market fit right, take some weirder choices on, 00:41:04.080 --> 00:41:10.220 on UI until you find the right things, push your engineering and your, Research 00:41:10.220 --> 00:41:14.550 teams to see around the next corner, like all of those things are going to matter 00:41:14.550 --> 00:41:16.520 a lot more than capital, to be honest. 00:41:16.560 --> 00:41:19.070 And the interesting thing, of course, is that those sets of 00:41:19.070 --> 00:41:20.870 things do tend to beget capital. 00:41:21.320 --> 00:41:25.010 And I wish we had some charts or graphs of the number of times that over the 00:41:25.010 --> 00:41:29.540 years where you know, some of this is just once you've lived through it, like 00:41:29.550 --> 00:41:34.160 five or six or seven or eight times as a founder, And then as a, as a vc, you just, 00:41:35.040 --> 00:41:38.100 it's that like old Chinese proverb of the guy with the horse where you're like, you 00:41:38.100 --> 00:41:39.720 know, the thing that looks like a gift. 00:41:39.750 --> 00:41:42.510 You're not sure if it's actually a gift and we'll just turn 00:41:42.510 --> 00:41:43.950 another page and I don't know. 00:41:44.010 --> 00:41:48.700 And a number of times that the, the most well-funded player has won. 00:41:48.760 --> 00:41:49.120 Yeah. 00:41:49.150 --> 00:41:50.650 I don't even know if it's correlated. 00:41:50.710 --> 00:41:51.070 No. 00:41:51.160 --> 00:41:54.550 Uh, at least in the early days at some point, like IPO and beyond, like 00:41:54.550 --> 00:41:55.810 it's a different world, but Sure. 00:41:55.810 --> 00:41:57.280 But in these early. 00:41:57.870 --> 00:41:59.890 You know, startup malleable. 00:41:59.900 --> 00:42:01.180 We'll see who wins. 00:42:01.500 --> 00:42:04.900 Um, person who did well this week, person who won the month, 00:42:04.950 --> 00:42:06.160 person who won the quarter. 00:42:06.440 --> 00:42:07.190 It's a long game. 00:42:07.370 --> 00:42:10.970 Fraser Kelton: I think you can overstuff a duck and end up with foie gras. 00:42:11.040 --> 00:42:13.380 I think you can, you do that with a startup and you kill it. 00:42:13.640 --> 00:42:16.290 The way that you framed it to a founder once that I think has really 00:42:16.290 --> 00:42:19.670 resonated with me is, if you know what you want to do with the capital, 00:42:20.190 --> 00:42:22.130 All of that capital, go and do it. 00:42:22.890 --> 00:42:25.290 Uh, and if you don't know what you're going to do with the capital, uh, 00:42:25.290 --> 00:42:29.190 it is, just think of it as going to be a negative for your company. 00:42:29.740 --> 00:42:30.500 Nabeel Hyatt (2): Yeah, that's right. 00:42:30.550 --> 00:42:31.560 Is it fuel? 00:42:31.920 --> 00:42:34.030 Uh, and, and I mean fuel in the real way. 00:42:34.030 --> 00:42:36.084 You know what to do with it, not just more runway. 00:42:36.924 --> 00:42:39.513 If so, it's possible it'll help you get ahead. 00:42:39.803 --> 00:42:40.143 Right. 00:42:40.273 --> 00:42:40.693 That's right. 00:42:41.873 --> 00:42:43.603 Let's be done for this week, man. 00:42:43.633 --> 00:42:45.303 It's great chatting with you, again. 00:42:45.473 --> 00:42:49.813 As always, I'll leave with one, Little thing, which is, at a dinner 00:42:49.813 --> 00:42:56.173 last night, my friend Michael Galpert said is, is WebSims, the GeoCities. 00:42:56.528 --> 00:42:57.728 Of the LLM era. 00:42:57.768 --> 00:43:00.108 And so if you haven't tried Web Sims, we're not going to talk 00:43:00.118 --> 00:43:03.458 about it here, but I would, I would suggest going and playing with it. 00:43:03.538 --> 00:43:05.648 It is a crazy fun little demo. 00:43:05.648 --> 00:43:08.308 And as long as we're talking about new UIs and crazy things, I 00:43:08.308 --> 00:43:09.058 think you should go take a look. 00:43:09.108 --> 00:43:09.608 Okay. 00:43:09.688 --> 00:43:10.288 I got to do it. 00:43:10.298 --> 00:43:12.958 Didn't you have a product you were going to roll out with me or talk about as well? 00:43:13.808 --> 00:43:15.518 Fraser Kelton: I'm just going to drop it and then we'll hang up. 00:43:15.828 --> 00:43:19.528 I tried this new, amazing product called Superhuman. 00:43:20.223 --> 00:43:20.613 Nabeel Hyatt (2): Finally 00:43:21.092 --> 00:43:21.922 Fraser Kelton: I love Gmail. 00:43:22.032 --> 00:43:24.832 Yeah, I have lived in Gmail for now 20 years. 00:43:24.922 --> 00:43:29.962 I don't want to adopt a new workflow when I download a new email client. 00:43:30.312 --> 00:43:31.612 I just want to do email. 00:43:32.232 --> 00:43:37.342 And Superhuman has made the decision that you can just do email in a beautifully 00:43:37.342 --> 00:43:39.732 designed UI if you want to do it. 00:43:40.282 --> 00:43:42.522 They've introduced a handful of AI features. 00:43:42.552 --> 00:43:46.972 And I think the thing that I take away from that is that they're feathered in. 00:43:47.042 --> 00:43:48.502 That's the way that I would describe it. 00:43:48.502 --> 00:43:49.892 It's not ham fisted. 00:43:50.012 --> 00:43:53.372 Um, it's not trying to turn it into a marketing event where it's like, 00:43:53.372 --> 00:43:58.032 Hey, here's this new AI feature that we're going to cram down you. 00:43:58.172 --> 00:44:01.482 The summarization of threads is really nice. 00:44:01.592 --> 00:44:04.882 Um, and then you can actually like click and it expands to show you 00:44:04.912 --> 00:44:09.392 a, a different like depth of the summary, which is really nice. 00:44:09.442 --> 00:44:09.752 Right. 00:44:09.912 --> 00:44:13.822 They have instant replies that are nice. 00:44:14.747 --> 00:44:16.987 And then again, like thoughtfully done where they're there if you 00:44:16.987 --> 00:44:18.417 want it and not if you don't. 00:44:19.717 --> 00:44:22.967 Nabeel Hyatt (2): I will say though, I have so indoctrinated 00:44:22.967 --> 00:44:26.105 into the superhuman workflows, but it's interesting to ask, What's the 00:44:26.105 --> 00:44:29.455 Granola co pilot version of SuperHuman? 00:44:29.995 --> 00:44:36.815 And for me, I would want to log in in the morning and have the most 00:44:36.835 --> 00:44:44.065 common emails I would reply to have drafts of the replies already there. 00:44:44.775 --> 00:44:46.545 That would be the Granola way of doing this. 00:44:46.995 --> 00:44:48.095 And I think that would be awesome. 00:44:48.095 --> 00:44:50.425 And I could delete them, I could say, No, I didn't want to say that. 00:44:50.735 --> 00:44:56.845 But right now, it's a very notion view of how to influence AI, right? 00:44:56.845 --> 00:45:00.265 It is a little command key that I can type in, finish this 00:45:00.265 --> 00:45:01.725 email, or so on and so forth. 00:45:02.075 --> 00:45:05.509 And I'd love something a little bit more copilot y but assertive. 00:45:05.509 --> 00:45:09.380 But before that works , I think there's probably a hurdle on the 00:45:09.380 --> 00:45:12.193 writing style getting better as well, but it's a thought to end on. 00:45:13.181 --> 00:45:13.365 Cool. 00:45:13.695 --> 00:45:14.315 See you later, Fraser. 00:45:14.495 --> 00:45:15.479 Fraser Kelton: See ya.