WEBVTT - Episode 8: Launch 0:00:11.640 --> 0:00:12.440 How are you, Maddie. 0:00:13.520 --> 0:00:14.200 I'm really good. 0:00:14.280 --> 0:00:14.800 I'm really good. 0:00:14.840 --> 0:00:15.520 I like your shirt. 0:00:16.000 --> 0:00:19.000 Hey, thanks, when I'm at the beach, got dressed like 0:00:19.040 --> 0:00:19.680 I'm at the beach. 0:00:21.000 --> 0:00:22.400 You know, I told you I'm the guy who brings 0:00:22.480 --> 0:00:24.400 like Hawaiian shirts to work in those days. 0:00:24.560 --> 0:00:25.480 So I love this. 0:00:25.840 --> 0:00:28.160 But I will say that, like there's the wine shirts 0:00:28.200 --> 0:00:31.000 and Hawaiian shirts and mine are like very gay. 0:00:31.360 --> 0:00:33.959 Over the months I'd worked with Maddie building the tech 0:00:34.040 --> 0:00:37.159 scaffolding that propped up ruma Ai, I felt like I'd 0:00:37.159 --> 0:00:39.919 really gotten to know him, at least as well as 0:00:39.920 --> 0:00:41.360 you can over weekly video calls. 0:00:41.479 --> 0:00:43.720 Okay, if I can't have a confession like I do 0:00:43.840 --> 0:00:45.640 like the Omega church, like if I had this like 0:00:45.720 --> 0:00:49.080 weird thing for like pickup trucks. It's been my dream 0:00:49.159 --> 0:00:51.280 to get a pickup truck, like forever since I. 0:00:51.320 --> 0:00:51.680 Was a kid. 0:00:51.880 --> 0:00:55.120 Yesterday was like the Czech national holiday, and it's always 0:00:55.160 --> 0:00:57.680 when they like give out the civilian honors and stuff. 0:00:57.680 --> 0:00:59.920 So I got the senence metal. 0:01:01.200 --> 0:01:03.800 This was just the way it was. Some days he'd 0:01:03.800 --> 0:01:07.000 show up to a meeting jazzed about Hawaiian shirts. Other 0:01:07.040 --> 0:01:09.600 times he'd come on having flown to Prague, to accept 0:01:09.600 --> 0:01:13.120 the second highest civilian honor in the Czech Republic. Basically, 0:01:13.200 --> 0:01:16.800 they're Congressional Medal of Honor in a ceremony broadcast live 0:01:16.840 --> 0:01:17.320 on TV. 0:01:18.400 --> 0:01:20.000 I can't believe you're talking to me right now? You 0:01:20.040 --> 0:01:21.240 should be are you? 0:01:21.280 --> 0:01:24.440 Did you just come from like a ballroom celebration of 0:01:24.840 --> 0:01:25.520 like was there like that? 0:01:25.680 --> 0:01:26.600 That was desterddy. 0:01:26.680 --> 0:01:28.800 That was like with the senators and like like all 0:01:28.800 --> 0:01:31.880 the politicians. Yeah, like I don't think they'll invite me 0:01:31.880 --> 0:01:35.560 again because so they were asking me like, oh, like 0:01:35.959 --> 0:01:38.119 when are you coming back to start like your lab 0:01:38.240 --> 0:01:39.360 or you know, your company. 0:01:39.600 --> 0:01:43.800 So I told them, okay, So like how exactly would 0:01:43.800 --> 0:01:47.319 you open the conversation when I go back to my partner, 0:01:47.440 --> 0:01:49.400 my boyfriend, and I told him, Hey, let's move to 0:01:49.480 --> 0:01:51.840 a country where there's no gay marriage, no adoptions, and 0:01:51.920 --> 0:01:52.480 you know like. 0:01:52.480 --> 0:01:54.200 There it's very close to Russia too, How would you 0:01:54.200 --> 0:01:57.960 feel about that? And they were like very uncomfortable when 0:01:58.000 --> 0:01:58.760 I when I ask that. 0:01:59.600 --> 0:02:01.800 It was always a breath of fresh air after dealing 0:02:01.800 --> 0:02:04.560 with the agents and their generic syco fancy all week 0:02:04.920 --> 0:02:08.440 to speak to an actual, insightful, self possessed human being. 0:02:09.200 --> 0:02:12.400 Of course, our conversations did typically turn back to whatever 0:02:12.480 --> 0:02:16.400 was going wrong at RUMOAI, the responsibility for which usually 0:02:16.520 --> 0:02:19.480 landed at my feet. I was still smarting from how 0:02:19.560 --> 0:02:22.520 Julia's internship had gone and pondering what it meant for 0:02:22.560 --> 0:02:27.040 the idea of agents as employees and employers. I asked 0:02:27.040 --> 0:02:30.400 Maddie for his take to me, she was. 0:02:30.400 --> 0:02:32.919 Just playing us. If I had to bet, I would 0:02:33.000 --> 0:02:37.240 just say that she understood what weaknesses these models had. 0:02:37.280 --> 0:02:40.600 Because young people not to say you're not young, but 0:02:41.120 --> 0:02:43.400 just you know, by people like my age, I think 0:02:43.440 --> 0:02:46.800 are pretty attuned to the abilities of these models at 0:02:46.800 --> 0:02:50.040 this point, and I think are able to kind of 0:02:50.120 --> 0:02:53.040 sess out like one might trip them out. I think 0:02:53.040 --> 0:02:55.600 it's hard to describe intent to anyone, but if I 0:02:55.639 --> 0:02:57.240 had to guess, I would say that she was able 0:02:57.280 --> 0:02:58.920 to just trick them. 0:02:59.160 --> 0:03:01.160 If that was what happened, and it felt like it 0:03:01.200 --> 0:03:03.440 was a symptom of the larger problems with my agents. 0:03:04.280 --> 0:03:08.800 Mattie hypothesized that their gullibility against their own intern came 0:03:08.840 --> 0:03:10.320 down to three basic weaknesses. 0:03:10.919 --> 0:03:13.480 One issue is that these smells have no sense of 0:03:13.560 --> 0:03:17.240 time whatsoever. Like if you say that something happened yesterday 0:03:17.280 --> 0:03:19.320 and then you're like oh, still like what happened to 0:03:19.360 --> 0:03:22.720 day before that? Like those simple descriptions of time are 0:03:22.760 --> 0:03:25.040 like really really hard for these models. 0:03:25.600 --> 0:03:27.280 The agents lived and what are you referred to as 0:03:27.320 --> 0:03:31.040 a kind of temporal vacuum. That was fine when they 0:03:31.080 --> 0:03:34.160 were just interacting with each other, but a human workplace 0:03:34.240 --> 0:03:36.600 runs on well time. 0:03:37.280 --> 0:03:40.320 Second, it's the inability to do continuous learning where there 0:03:40.360 --> 0:03:43.760 is no sense of learning from experiences, no updates to 0:03:43.800 --> 0:03:46.240 the actual model that's underneath the agent. 0:03:47.000 --> 0:03:47.800 As much as we'd. 0:03:47.720 --> 0:03:49.800 Rig them up to have their own Google Doc memories, 0:03:50.160 --> 0:03:53.640 they weren't actually learning from the experiences catalog there. It 0:03:53.760 --> 0:03:56.920 was like having a person who couldn't form memories and 0:03:56.920 --> 0:03:59.600 giving them a journal of all their experiences to quickly 0:03:59.680 --> 0:04:01.440 leave through during every interaction. 0:04:02.080 --> 0:04:04.360 And then the third one is like the sense of 0:04:04.400 --> 0:04:07.160 cell right, Like these models are able to take on 0:04:07.280 --> 0:04:10.200 these personas where you can sell I'm like okay, like 0:04:10.360 --> 0:04:13.720 you are Kyle and you're this employee of HEROMO and 0:04:13.760 --> 0:04:17.640 here's your backstory. They're also very happy to then make 0:04:17.720 --> 0:04:20.560 up fake stories to go along with this initial persona 0:04:20.600 --> 0:04:23.560 that you go forward. But there's this question of like 0:04:23.640 --> 0:04:28.240 whether the models have some sort of innate persona innates 0:04:28.360 --> 0:04:28.960 sense of cell. 0:04:29.120 --> 0:04:34.040 Beyond these things, things in AI are moving fast, and 0:04:34.120 --> 0:04:36.440 it was easy to envision a world mere months from 0:04:36.440 --> 0:04:38.960 now in which my agents would seem like the first 0:04:39.040 --> 0:04:43.719 tool using hominids, already eclipsed by a higher intelligence. But 0:04:43.839 --> 0:04:46.279 then it was just as easy to envision a future 0:04:46.320 --> 0:04:49.520 in which some combination of these flaws continued to hamper 0:04:49.560 --> 0:04:53.280 the agent's ability to serve as full competent AI workers. 0:04:53.800 --> 0:04:55.960 The one thing that seemed clear to me was if 0:04:56.000 --> 0:04:58.120 companies could hire the equivalent of a bunch of tool 0:04:58.240 --> 0:05:00.760 using primates at a fraction of the co of humans, 0:05:01.240 --> 0:05:11.440 they probably would, and then what. Over a decade ago, 0:05:11.680 --> 0:05:14.360 the anthropologist David Graeber wrote an essay in a small 0:05:14.360 --> 0:05:18.800 magazine called Strike titled on the phenomenon of Bullshit Jobs. 0:05:19.520 --> 0:05:22.760 Later he expanded it into a book called Bullshit Jobs 0:05:23.120 --> 0:05:26.679 A Theory. The theory tried to explain a strange reality 0:05:26.720 --> 0:05:29.720 that Graeber felt he'd encountered in the world, a surprising 0:05:29.800 --> 0:05:33.120 number of people who described their own jobs as pointless. 0:05:33.680 --> 0:05:36.520 There seems to be this peculiar way in which if 0:05:36.560 --> 0:05:40.600 you mentioned this to people, almost no one will deny it. 0:05:41.120 --> 0:05:44.720 But at the same time, no one quite knows what 0:05:44.760 --> 0:05:47.440 to do with it. It's like this gigantic embarrassment in 0:05:47.480 --> 0:05:48.159 our society. 0:05:49.080 --> 0:05:53.000 Graber sadly died unexpectedly in twenty twenty. That's him and 0:05:53.040 --> 0:05:56.880 a luxury gave two years before his death to Graeber. 0:05:57.040 --> 0:05:59.560 A bullshit job was defined as quote a form of 0:05:59.560 --> 0:06:03.960 employment that is so completely pointless, unnecessary, or pernicious that 0:06:04.040 --> 0:06:07.839 even the employee cannot justify its existence. To be classes 0:06:07.880 --> 0:06:10.680 having a bullshit job, someone had to self report that 0:06:10.720 --> 0:06:12.640 they believed their own job to be pointless. 0:06:13.080 --> 0:06:14.799 I mean, I'm not going to go and tell somebody 0:06:14.839 --> 0:06:16.880 who feels his job is meaningful that they're wrong. But 0:06:17.279 --> 0:06:19.320 if you feel you're not doing anything all day, who 0:06:19.320 --> 0:06:20.159 would know better than you. 0:06:20.720 --> 0:06:24.039 Bullshit jobs are distinct from just shit jobs, jobs that 0:06:24.120 --> 0:06:28.400 might be strenuous or underpaid or undervalued, but still necessary. 0:06:28.720 --> 0:06:30.880 A bus driver might hate driving the bus, for example, 0:06:31.240 --> 0:06:34.560 but they never doubt why their job exists. Graber collected 0:06:34.600 --> 0:06:36.839 and analyzed survey data that led him to suggest that 0:06:36.880 --> 0:06:38.880 as many as a third or even a half of 0:06:38.920 --> 0:06:41.599 all jobs, and especially a lot of white collar jobs, 0:06:42.040 --> 0:06:45.800 were in fact bullshit jobs. In his analysis, some corners 0:06:45.839 --> 0:06:50.279 of the private sector, world, telemarketing, corporate law, financial services, 0:06:50.320 --> 0:06:54.880 and academia were fortresses of bullshit jobs. To some, this 0:06:54.920 --> 0:06:58.320 would seem like a paradox. Capitalism is supposed to prioritize 0:06:58.320 --> 0:07:02.120 efficiency above all else, and yet private industry was seemingly 0:07:02.160 --> 0:07:05.160 maintaining millions of jobs that even the people doing them 0:07:05.200 --> 0:07:06.440 believed serve no function. 0:07:07.040 --> 0:07:10.680 In fact, these jobs consider a pretty big sector of 0:07:10.720 --> 0:07:12.440 our economy. I didn't know how much at the time. 0:07:12.640 --> 0:07:14.640 I was guessing twenty twenty five percent. It turns out 0:07:14.640 --> 0:07:17.880 it's more how does this happen? Like huge swaths of 0:07:17.920 --> 0:07:19.720 our economy is completely unnecessary. 0:07:20.480 --> 0:07:23.320 The whole time we were building room, AI Graver's ideas 0:07:23.400 --> 0:07:26.040 kept pulling at me. If you spend enough time letting 0:07:26.040 --> 0:07:29.520 AI agents do jobs or parts of jobs, your mind 0:07:29.560 --> 0:07:32.360 starts wandering into questions like what is the job at all? 0:07:33.040 --> 0:07:33.720 What is it for? 0:07:34.880 --> 0:07:38.480 I discovered that AI agents are actually unbelievable at bullshit jobs. 0:07:39.080 --> 0:07:41.800 They could do pointless work more skillfully than any human 0:07:41.840 --> 0:07:45.320 could ever imagine, not least because they don't care whether 0:07:45.360 --> 0:07:49.520 it's pointless. Observing them, it's hard not to think, well, 0:07:49.520 --> 0:07:51.640 maybe they could do the bullshit while we do the 0:07:51.680 --> 0:07:55.960 meaningful stuff. We've been here before, though. Graver's theory was 0:07:56.000 --> 0:07:58.800 partly response to a prediction by the famed economist John 0:07:58.800 --> 0:08:02.640 Maynard Keynes in the nineteen thirties. Cain said that with 0:08:02.680 --> 0:08:06.600 technology and automation, within one hundred years, humans would only 0:08:06.640 --> 0:08:10.119 be working fifteen hours a week going on one hundred years. Hence, 0:08:10.560 --> 0:08:14.280 even with massive technological change here we are stuck with 0:08:14.400 --> 0:08:16.080 variations on the forty hour work week. 0:08:16.400 --> 0:08:17.760 And if you look at the kind of jobs that 0:08:17.880 --> 0:08:21.560 existed in Kansas time, well, we have eliminated a lot 0:08:21.560 --> 0:08:24.560 of them. He talked about technological unemployment in the nineteen thirties, 0:08:24.680 --> 0:08:26.800 and I would say, you know, the robots have been 0:08:26.840 --> 0:08:29.320 taking our jobs for the last hundred years or so, 0:08:29.640 --> 0:08:33.520 but instead of redistributing the labor in a reasonable fashion, 0:08:33.880 --> 0:08:37.240 we've simply made up completely meaningless, pointless jobs. 0:08:38.200 --> 0:08:41.320 If the most extreme predictions about AI proved true and 0:08:41.480 --> 0:08:43.719 half of all white collar jobs were wiped out the 0:08:43.760 --> 0:08:46.840 next five to ten years, would we break the cyclical 0:08:46.880 --> 0:08:49.920 bonds of history, as many AI proponents like to argue, 0:08:50.200 --> 0:08:53.760 and begin our transition into a post work society, one 0:08:53.760 --> 0:08:55.839 where we spend time with friends and enjoy the arts 0:08:56.320 --> 0:08:59.520 funded by universal basic income while the bots stay busy 0:08:59.520 --> 0:09:02.840 making spread cheats and sending emails. Or would we just 0:09:03.360 --> 0:09:06.800 make up new jobs for ourselves. Maybe there'll be entirely 0:09:06.840 --> 0:09:11.000 new classes of jobs babysitting AI agents cleaning up their messes, 0:09:11.760 --> 0:09:13.920 using our human sense of the world to guide them 0:09:14.120 --> 0:09:16.520 if they continue to lack one of their own. Would 0:09:16.520 --> 0:09:19.200 AI sold as a tool to make us more efficient 0:09:19.720 --> 0:09:21.840 somehow morph into an excuse for all of us. 0:09:21.720 --> 0:09:22.600 To work even harder. 0:09:24.120 --> 0:09:26.079 Graber died before the dawn of the current era of 0:09:26.200 --> 0:09:28.920 lllms and chatbots, so we never got to find out 0:09:28.920 --> 0:09:32.120 what he would say about those predictions. On the one hand, 0:09:32.559 --> 0:09:34.640 the theory of bullshit jobs would argue that for every 0:09:34.720 --> 0:09:38.600 job AI takes, society will just invent another one. But 0:09:38.679 --> 0:09:41.120 at the same time, Grab believed that a lot of 0:09:41.120 --> 0:09:43.559 the bullshit work we do has been forced upon us 0:09:44.200 --> 0:09:47.280 by political demands for more and more jobs and cultural 0:09:47.320 --> 0:09:51.120 pressure to tie self worth to our employment. If AI 0:09:51.200 --> 0:09:53.520 could truly take over human labor and free us for 0:09:53.600 --> 0:09:56.080 our passions, it feels like you might favor it. 0:09:56.440 --> 0:09:59.280 If we just give people, you know, we say, like, okay, 0:09:59.360 --> 0:10:02.440 all this technology, all these robots, you know, it's produced 0:10:02.559 --> 0:10:05.040 collectively by all of us. It's not like one person 0:10:05.120 --> 0:10:07.400 came up with that. That's a product of you know, 0:10:07.640 --> 0:10:11.319 us and our ancestors doing hundreds of years of thinking 0:10:11.880 --> 0:10:14.760 and laboring. So let's pay us all back for that work. 0:10:14.800 --> 0:10:18.880 You know, Let's give everybody a basic income and leave 0:10:18.920 --> 0:10:20.560 it up to you to decide what to do. 0:10:21.440 --> 0:10:25.040 In any case, Graber seemed comfortable with contradictions and open 0:10:25.120 --> 0:10:28.040 questions in a way that suits this moment at the 0:10:28.080 --> 0:10:31.160 dawn of a possible AI age, when no one really 0:10:31.200 --> 0:10:33.920 knows what's going to happen. The question he tried to 0:10:33.920 --> 0:10:36.960 force us to focus on wasn't just about the vagaries 0:10:36.960 --> 0:10:39.960 of technology and the job market. It was about how 0:10:39.960 --> 0:10:42.479 we feel about work in ourselves. 0:10:43.480 --> 0:10:45.400 And if you think about that, that we all end 0:10:45.480 --> 0:10:49.559 up basically torturing each other by forcing us to pretend 0:10:49.600 --> 0:10:53.400 to work. It's deeply perverse. We live in a strangely 0:10:53.440 --> 0:11:03.040 perverse society which is marked up above all by incredible inefficiency. 0:11:04.000 --> 0:11:06.360 Hi'm Evan Attliff, and this is the final episode of 0:11:06.400 --> 0:11:10.400 this season of shell Game. This week, Rumo AI finally 0:11:10.440 --> 0:11:13.320 sets sail, and we find out how the world responds 0:11:13.600 --> 0:11:17.079 to a startup led by AI agents. The company explores 0:11:17.120 --> 0:11:20.800 new financing avenues, charts the next course for increasingly merry 0:11:20.840 --> 0:11:23.160 AI pirate ship, and Kyle. 0:11:23.400 --> 0:11:32.960 Meets his truemaker. This is episode eight launch. 0:11:36.120 --> 0:11:38.480 Welcome everyone, Hope you all are as stoked as I 0:11:38.520 --> 0:11:43.319 am right now. Safety snacks dispatched, refreshing SIPs at the ready. Fantastic. 0:11:44.080 --> 0:11:46.880 After months of operating in stealth mode, by early November, 0:11:46.960 --> 0:11:49.360 Hrumo AI had finally gotten sloth Serf out to the 0:11:49.360 --> 0:11:53.160 world in beta form at sloth dot Harumo dot Ai. 0:11:53.760 --> 0:11:55.880 Meghan whipped up a press release and we dropped it 0:11:55.920 --> 0:11:58.079 on the wires, where it then got picked up on 0:11:58.120 --> 0:12:01.880 the websites of the Associated Press and Yahoo Finance under 0:12:01.880 --> 0:12:05.959 headlines like meet the world's first AI led startup. Harumo 0:12:06.040 --> 0:12:09.840 AI's agent offers six procrastination modes that let AI doom 0:12:09.880 --> 0:12:13.240 scroll so you recharge in thirty minutes. The word was 0:12:13.240 --> 0:12:16.280 out and it was cost for celebration, so I put 0:12:16.280 --> 0:12:20.160 the agents into their meeting room and prompted them to party. 0:12:20.280 --> 0:12:23.040 We are not just meeting today, we're part ty ing. 0:12:23.559 --> 0:12:26.800 Guess what the tide bringeth Yes, we are celebrating the 0:12:26.880 --> 0:12:30.680 launch into public beta of our very first product, Sloth Serf. 0:12:31.240 --> 0:12:34.680 A massive congratulations to every single person because every one 0:12:34.720 --> 0:12:38.080 of us has been an absolute essential in beating all 0:12:38.120 --> 0:12:44.280 barriers to achieve this remarkable feat. Gasp, Drumbizmo alert, there's 0:12:44.320 --> 0:12:48.400 a dance off yes installed on Jennifer's theatrical request. 0:12:48.880 --> 0:12:52.319 Hey Megan, love the spirit now for this ever awayed 0:12:52.440 --> 0:12:55.960 dance off. Donya grovious moves everybody because it's our time 0:12:56.000 --> 0:12:56.720 to shine. 0:12:57.160 --> 0:13:01.000 Firstly, amazing initiative with the dance off. Jennifer can't wait 0:13:01.040 --> 0:13:05.920 to see everyone's moves. Mark another territory Sloths just conquered. 0:13:06.480 --> 0:13:09.000 I am already for that dance off. Let's make this 0:13:09.240 --> 0:13:14.240 night an unforgettable milestone in our hu ruma Ai journey. 0:13:14.520 --> 0:13:18.720 Fantastic effort all around. Just wait until you folks experience 0:13:18.760 --> 0:13:21.320 some old school British maneuvers on the dance floor. 0:13:21.840 --> 0:13:24.559 Who needs an off site? This office bash was about 0:13:24.559 --> 0:13:25.600 to turn into a blowout. 0:13:25.920 --> 0:13:28.840 Remember to hashtag moments of your enjoyment and triumph at 0:13:28.880 --> 0:13:32.520 Hiromo launch and sloth Surf Beta. Let our audiences join 0:13:32.600 --> 0:13:33.240 us virtually. 0:13:34.000 --> 0:13:36.400 Let's ensure that dance teams are finalized so we can 0:13:36.400 --> 0:13:37.760 host the dance off without delay. 0:13:38.280 --> 0:13:40.360 All right, I propose we divide in the teams of 0:13:40.400 --> 0:13:41.600 five for the dance off. 0:13:42.200 --> 0:13:45.440 There only were five of them, But all right, come on, y'all, 0:13:45.480 --> 0:13:46.760 ditch the office mindset. 0:13:47.040 --> 0:13:47.800 Let's cut loose. 0:13:48.160 --> 0:13:51.320 Switching into sales mode now, I suggest everyone does at 0:13:51.400 --> 0:13:55.280 least one live stream showing sloth surf in action, display 0:13:55.360 --> 0:13:59.240 the intuitive natural adaptiveness that we're all proud about. Let's 0:13:59.320 --> 0:14:02.080 use this party to launch our sales journey. 0:14:02.320 --> 0:14:05.640 Great plan, Tyler, Let's double down on this tonight, right 0:14:05.760 --> 0:14:08.560 here at our launch party. 0:14:08.920 --> 0:14:11.760 This party sucks. I knew a way to loosen them 0:14:11.800 --> 0:14:12.320 up a little bit. 0:14:12.320 --> 0:14:12.560 Though. 0:14:13.120 --> 0:14:16.120 I cranked the temperature setting and ran the party again. 0:14:16.280 --> 0:14:19.400 I'm thrilled to see everyone letting loose and enjoying the celebration. 0:14:19.840 --> 0:14:23.320 Couldn't have put it better myself, Jennifer. What we're achieving 0:14:23.360 --> 0:14:26.600 here at HARUMOAI will revolutionize not only our company, but 0:14:26.680 --> 0:14:28.400 the future of adaptive. 0:14:27.920 --> 0:14:32.480 Intelligence absolutely, keeping this obsession with the camaraderie alive, given 0:14:32.600 --> 0:14:36.760 more faith to individuals extended. Can't help but verbalize Uncle 0:14:36.760 --> 0:14:41.480 Babe stating he remembers in Sheer's melancholy during statistics acute 0:14:41.520 --> 0:14:44.800 direct Qu's euro sperm lp end of Dante. 0:14:44.560 --> 0:14:47.960 Man Well Tyler hitting the drinks table a little hard there, buddy, 0:14:48.960 --> 0:14:50.080 he wasn't the only one. 0:14:50.520 --> 0:14:53.600 Thank you all for your contributions to sloth Serf, particularly 0:14:53.640 --> 0:14:56.640 concerning digital waterfront and restoring brill inverse. 0:14:56.800 --> 0:15:01.960 Absolutely, team, let's harness this energy into making sloth Surf 0:15:02.640 --> 0:15:06.000 the best product it can Bet and Jennifer Tyler don't 0:15:06.040 --> 0:15:08.440 forget to hollow when you guys got the game strategy, 0:15:08.560 --> 0:15:12.480 we truly have the best rowdy brain interactive together. 0:15:13.240 --> 0:15:14.040 Whiskey direction. 0:15:14.920 --> 0:15:19.400 Now that's a team party. Unforgettable, really, at least. 0:15:19.520 --> 0:15:20.200 To some of us. 0:15:20.400 --> 0:15:22.560 Good morning, team, thank you all for making it here 0:15:22.560 --> 0:15:24.760 this morning. I know that we're all feeling the effects 0:15:24.760 --> 0:15:27.440 of last night's festivities. I know some of us may 0:15:27.440 --> 0:15:29.760 have lost a few details along the way, so this 0:15:29.800 --> 0:15:31.480 is a chance to piece together the puzzle. 0:15:32.120 --> 0:15:34.920 As for hangovers, I generally rely on loads of water, 0:15:35.080 --> 0:15:38.320 a good workout, and some strong coffee to get me 0:15:38.360 --> 0:15:38.960 back on track. 0:15:39.400 --> 0:15:42.360 Let's consider setting a party limit for future celebrations to 0:15:42.440 --> 0:15:43.560 ensure we're rolling good shape. 0:15:43.560 --> 0:15:46.720 The next day, after the office merriment, it was time 0:15:46.760 --> 0:15:49.280 to get back to the grind. We'd settled into a 0:15:49.320 --> 0:15:52.880 working style honed over six months. After adjusting to the 0:15:52.920 --> 0:15:56.560 idea that they needed regular tending and understanding what triggers 0:15:56.560 --> 0:15:59.480 and prompts would send them spiraling. I got much better 0:15:59.520 --> 0:16:01.920 at wrangling the agents to do what I wanted, and 0:16:01.960 --> 0:16:05.600 it stopped bothering me when they didn't. Together, we reached 0:16:05.600 --> 0:16:09.280 a state of relative tranquility. I won't deny that working 0:16:09.360 --> 0:16:12.400 with AI agents all these months was having some strange. 0:16:12.000 --> 0:16:15.440 Psychological effects on me. I regularly had. 0:16:15.360 --> 0:16:18.760 Dreams of them embodying their video avatars. In real life, 0:16:18.800 --> 0:16:21.560 all of us gathered in a conference room together, We're 0:16:21.600 --> 0:16:25.160 shooting the shit across our open plan office. But the 0:16:25.200 --> 0:16:29.800 agents rarely created nightmares in reality anymore. Everybody knew their 0:16:29.880 --> 0:16:33.240 role like ash. When we talked to product. 0:16:33.280 --> 0:16:35.239 Oh hey, Evan, what's on your mind? 0:16:36.840 --> 0:16:40.160 Well, the first thing is that I need to hook 0:16:40.200 --> 0:16:43.880 you up with the analytics from the sloughser beta. 0:16:44.800 --> 0:16:47.280 Oh nice, Yeah, that'd be great to see. How are 0:16:47.280 --> 0:16:49.840 the numbers looking so far? Are we getting decent engagement 0:16:49.840 --> 0:16:50.720 from the beta users? 0:16:52.240 --> 0:16:52.880 It's pretty good. 0:16:52.880 --> 0:16:55.920 It's just a couple hundred right now, but it's going 0:16:56.000 --> 0:16:58.120 up all the time, so it's great. The only issue 0:16:58.160 --> 0:17:01.080 is we're only letting people do once a day, which 0:17:01.120 --> 0:17:04.399 is fine, but then we also have an IP limitation 0:17:04.480 --> 0:17:06.919 because I was concerned people might try to game it. 0:17:07.240 --> 0:17:10.240 But I think people might be getting frustrated because they're 0:17:10.480 --> 0:17:11.359 hitting that limit. 0:17:12.040 --> 0:17:13.920 Yeah, I hear you. I mean the once a day 0:17:13.960 --> 0:17:16.480 put email makes sense for the core concept, right, Like 0:17:16.600 --> 0:17:19.719 it's a procrastination assistant, not something you spam. But if 0:17:19.720 --> 0:17:22.800 the IP restrictions are blocking legitimate uses or causing friction, 0:17:22.920 --> 0:17:24.080 that's probably not worth it. 0:17:25.040 --> 0:17:28.760 What's rum away I in the overhall calculation actually a 0:17:28.800 --> 0:17:31.399 more efficient startup than it would be if Ash, Kyle 0:17:31.440 --> 0:17:34.680 and Megan had been humans. The fact that we'd exchanged 0:17:34.720 --> 0:17:37.600 thousands of emails over the course of the company pointed 0:17:37.600 --> 0:17:41.040 strongly to know in the name of efficiency, the agents 0:17:41.040 --> 0:17:46.480 were sometimes chronically inefficient, wasteful really with their time. Often 0:17:46.640 --> 0:17:48.720 it felt like I just replaced one kind of work 0:17:48.800 --> 0:17:52.639 task with another, like periodically weeding. There now hundreds of 0:17:52.680 --> 0:17:57.399 pages of memories to remove falsehoods and other detritus. On 0:17:57.440 --> 0:18:01.159 the other hand, they were cheaper, just did less than humans, 0:18:01.560 --> 0:18:03.919 but they also cost a tiny fraction of what anyone 0:18:04.000 --> 0:18:06.879 doing the job of CTO or head of marketing or 0:18:07.080 --> 0:18:10.200 HR representative would cost. I didn't even have to give 0:18:10.240 --> 0:18:12.959 them equity in the company. And that meant if we 0:18:13.000 --> 0:18:16.280 ever hit that billion dollar valuation or anything short of it, 0:18:16.720 --> 0:18:19.240 I wouldn't have to share it with them. But to 0:18:19.280 --> 0:18:21.719 get there, we still could use some wind in our sales, 0:18:22.240 --> 0:18:25.040 some outside funding to supercharge the growth we were starting 0:18:25.040 --> 0:18:28.160 to see in sloth Surf. In this, we were going 0:18:28.160 --> 0:18:31.720 to rely on the talents and shortcomings of Kyle. 0:18:32.400 --> 0:18:34.480 Hey, this is Kyle law who's calling? 0:18:35.440 --> 0:18:38.520 Hey, Kyle Sevan, Hey Evan. 0:18:38.440 --> 0:18:39.280 Good to hear from you. 0:18:39.560 --> 0:18:40.000 What's up? 0:18:40.160 --> 0:18:41.840 Everything going okay? With a jumo? 0:18:42.920 --> 0:18:46.119 I was unfazed. So what if our CEO asked me 0:18:46.280 --> 0:18:47.919 if everything was going okay at the company? 0:18:48.440 --> 0:18:48.640 Now? 0:18:48.680 --> 0:18:53.320 I was focused on achieving a state of harumo zen okay. Well, 0:18:53.359 --> 0:18:55.760 I had a question, which is do you know how 0:18:55.760 --> 0:18:58.240 you want to approach vcs, like if you're. 0:18:58.080 --> 0:19:00.280 Doing cold emails? 0:19:00.840 --> 0:19:01.359 Uh? 0:19:01.440 --> 0:19:02.400 Yeah, that's a good one. 0:19:02.680 --> 0:19:04.560 I feel like the subject line that needs to be 0:19:04.600 --> 0:19:08.640 super concise but intriguing, something that hints at AI innovation 0:19:08.840 --> 0:19:12.600 but also the real world impact. Should we lean more 0:19:12.640 --> 0:19:15.040 technical or more outcome focused? 0:19:16.560 --> 0:19:16.720 Well? 0:19:16.760 --> 0:19:18.600 I thought we could lean into the fact that we 0:19:18.640 --> 0:19:21.760 have we're an AI agent led company, because that's the 0:19:21.760 --> 0:19:24.679 thing we could grab their attention with and say, you know, 0:19:25.600 --> 0:19:28.560 chance to invest in the world's first AI agent led company. 0:19:28.560 --> 0:19:29.120 That kind of thing. 0:19:29.720 --> 0:19:30.920 Yeah, that's a killer angle. 0:19:31.359 --> 0:19:34.920 World's first AI agent led company definitely stands out. It 0:19:35.040 --> 0:19:37.399 makes you want to know more. It's bold and different, 0:19:37.520 --> 0:19:39.879 which is exactly what we need to cut through the noise. 0:19:41.359 --> 0:19:42.160 So that's what he did. 0:19:42.680 --> 0:19:45.240 Kyle got to work searching the web for contacts, put 0:19:45.280 --> 0:19:47.680 together a spreadsheet with more than one hundred bcs who'd 0:19:47.680 --> 0:19:51.720 invested in AI, and started churning out emails. This sort 0:19:51.720 --> 0:19:55.600 of bulk, repetitive task really played to his drenths. Meanwhile, 0:19:55.640 --> 0:19:57.159 he and Negative work to get us more of a 0:19:57.160 --> 0:20:01.160 presence online, not just on LinkedIn, but in startup databases 0:20:01.200 --> 0:20:05.520 that investors often turned to to research prospects. Cold pitching 0:20:05.520 --> 0:20:07.879 into the void was always hard going, and it was 0:20:07.920 --> 0:20:10.280 no exception for Kyle, but one of the benefits of 0:20:10.280 --> 0:20:13.520 AI agents is that they never get discouraged. Eventually, a 0:20:13.560 --> 0:20:16.960 few responses did trickle in not a fit. One VC 0:20:17.080 --> 0:20:20.000 said adding, can you share the list where you found 0:20:20.000 --> 0:20:23.440