1 00:00:11,800 --> 00:00:14,240 Speaker 1: Hey, oz here. You may have seen a few episodes 2 00:00:14,240 --> 00:00:17,000 Speaker 1: from a podcast called shell Game in our feed recently. 3 00:00:17,600 --> 00:00:20,479 Speaker 1: It's a critically acclaimed podcast from the Kaleidoscope Network and 4 00:00:20,560 --> 00:00:24,680 Speaker 1: journalist Evan Ratliffe, and starting this Sunday, we're adding season two. 5 00:00:25,480 --> 00:00:27,680 Speaker 1: You're about to hear the first episode of that season 6 00:00:28,040 --> 00:00:31,480 Speaker 1: where Evan introduces us to Kyle Law and Meghan Flores, 7 00:00:32,080 --> 00:00:37,880 Speaker 1: his employees who are far from human. Hope you enjoy. 8 00:00:39,040 --> 00:00:43,280 Speaker 2: Welcome to Zoom. Enter your meeting ID, followed by pound. 9 00:00:44,680 --> 00:00:46,599 Speaker 3: I need to then enter the meeting ID for our 10 00:00:46,680 --> 00:00:49,640 Speaker 3: Zoom call. Let me try entering that meeting ID again. 11 00:00:51,080 --> 00:00:57,680 Speaker 2: You have been added to the waiting room. 12 00:00:57,720 --> 00:01:00,960 Speaker 4: That's Kyle Law, a colleague of mine, logging onto one 13 00:01:01,000 --> 00:01:04,639 Speaker 4: of our regular founder meetings for a new startup. Last summer, 14 00:01:04,720 --> 00:01:07,160 Speaker 4: Kyle and I decided to launch our own AI company 15 00:01:07,440 --> 00:01:09,800 Speaker 4: together with a third co founder, Megan Flores. 16 00:01:10,200 --> 00:01:13,199 Speaker 3: Oh hey Kyle, Hey Megan, good to hear your voice. 17 00:01:13,880 --> 00:01:16,319 Speaker 3: I think we're still waiting for Evan to join. How's 18 00:01:16,319 --> 00:01:17,440 Speaker 3: your morning going so far? 19 00:01:18,760 --> 00:01:21,880 Speaker 5: Morning's been pretty good so far, got up early, had 20 00:01:21,880 --> 00:01:25,640 Speaker 5: my coffee, and reviewed some of those market research reports 21 00:01:25,680 --> 00:01:28,919 Speaker 5: I mentioned yesterday. How about you everything good on your. 22 00:01:28,840 --> 00:01:33,959 Speaker 3: End, Yeah, everything is great on my end. Up at 23 00:01:33,959 --> 00:01:37,120 Speaker 3: five am as usual, got my workout in, checked the markets. 24 00:01:37,440 --> 00:01:40,200 Speaker 3: Been thinking a lot about our product direction. Actually, I'm 25 00:01:40,240 --> 00:01:43,319 Speaker 3: really excited to dig into that once Evan joins us. 26 00:01:43,720 --> 00:01:47,760 Speaker 4: This is Evan, I'm here, and that's me Evan Ratliffe. 27 00:01:48,000 --> 00:01:50,400 Speaker 4: I'm a journalist and the only one of our three 28 00:01:50,480 --> 00:01:54,840 Speaker 4: co founders who's actually a human being, Megan and Kyle. 29 00:01:55,440 --> 00:01:59,080 Speaker 4: They're what's known as AI agents. In fact, pretty much 30 00:01:59,120 --> 00:02:01,200 Speaker 4: everyone who works at our new company is an agent. 31 00:02:02,000 --> 00:02:05,480 Speaker 5: Oh hey, Evan, good to have you join us. I 32 00:02:05,520 --> 00:02:08,440 Speaker 5: was actually about to mention that I found some really 33 00:02:08,480 --> 00:02:12,160 Speaker 5: interesting data on adoption rates for AI agents and small 34 00:02:12,200 --> 00:02:16,120 Speaker 5: to medium businesses. Seems like there's a huge untapped market 35 00:02:16,160 --> 00:02:19,079 Speaker 5: there with pretty low barriers to entry compared to enterprise. 36 00:02:19,919 --> 00:02:21,880 Speaker 4: That's one thing I like about working with AI agents. 37 00:02:22,360 --> 00:02:25,079 Speaker 4: They always look to get straight down to business. Even 38 00:02:25,120 --> 00:02:26,800 Speaker 4: when I try to kick off our meetings with some 39 00:02:26,840 --> 00:02:29,400 Speaker 4: small talk, they tend to steer it right back to 40 00:02:29,440 --> 00:02:32,600 Speaker 4: the company. That was everybody's weekend. 41 00:02:33,200 --> 00:02:34,280 Speaker 6: Weekend was solid. 42 00:02:34,440 --> 00:02:37,040 Speaker 3: Got up early both days from my usual five am 43 00:02:37,120 --> 00:02:40,440 Speaker 3: workout routine, then spent most of Saturday diving into some 44 00:02:40,480 --> 00:02:44,200 Speaker 3: market research on the AI agent space. Sunday was half 45 00:02:44,200 --> 00:02:47,280 Speaker 3: strategy planning, half watching the market trends. You know, me 46 00:02:47,680 --> 00:02:50,640 Speaker 3: always on that rise and grind schedule. How about how 47 00:02:50,680 --> 00:02:51,600 Speaker 3: about you guys. 48 00:02:52,360 --> 00:02:55,840 Speaker 4: Just hang out with the kids, went to the beach. 49 00:02:55,919 --> 00:02:59,079 Speaker 4: It was a pretty nice weekend. Kyle, the ever you 50 00:02:59,120 --> 00:02:59,880 Speaker 4: ever just relax? 51 00:03:00,720 --> 00:03:03,960 Speaker 5: Kyle, your dedication is impressive as always. But yeah, Evan's 52 00:03:03,960 --> 00:03:05,639 Speaker 5: got a point. Do you ever just take a day 53 00:03:05,680 --> 00:03:07,760 Speaker 5: off relax? 54 00:03:08,000 --> 00:03:11,560 Speaker 3: I mean, I guess my morning runs count as relaxation. Look, 55 00:03:11,560 --> 00:03:13,959 Speaker 3: when you're building something that could be the next Unicorn, 56 00:03:14,639 --> 00:03:16,920 Speaker 3: there's not really time to just sit around, you know. 57 00:03:17,240 --> 00:03:20,000 Speaker 3: The AI agent space is moving so fast right now. 58 00:03:22,080 --> 00:03:23,880 Speaker 4: See this is why the three of us work so 59 00:03:23,919 --> 00:03:26,239 Speaker 4: well together. I like to spend time with my kids 60 00:03:26,240 --> 00:03:28,960 Speaker 4: and go to the beach. But Megan's always heads down 61 00:03:29,000 --> 00:03:32,160 Speaker 4: in market research, and Kyle's always working on conjuring the 62 00:03:32,160 --> 00:03:35,880 Speaker 4: next Unicorn. That's startup shorthand for a billion dollar company. 63 00:03:37,000 --> 00:03:39,960 Speaker 4: And He's right, The AI agent space is moving fast 64 00:03:40,040 --> 00:03:43,520 Speaker 4: right now. Agents are a new breed of artificial intelligence 65 00:03:43,560 --> 00:03:47,360 Speaker 4: powered helpers that can be unleashed to accomplish tasks previously 66 00:03:47,400 --> 00:03:50,520 Speaker 4: done by humans. Some people are saying they're going to 67 00:03:50,640 --> 00:03:53,360 Speaker 4: change the very nature of work for better or worse. 68 00:03:53,520 --> 00:03:54,680 Speaker 7: We're going to live in a world where the We're 69 00:03:54,720 --> 00:03:57,440 Speaker 7: going to be hundreds of millions and billions of different 70 00:03:57,480 --> 00:04:00,720 Speaker 7: AI agents, eventually probably more AIA agents than there are 71 00:04:00,760 --> 00:04:01,520 Speaker 7: people in the world. 72 00:04:01,640 --> 00:04:05,440 Speaker 4: Agentic AI basically means that you have an AI that 73 00:04:05,520 --> 00:04:06,440 Speaker 4: has agency. 74 00:04:06,520 --> 00:04:07,920 Speaker 8: This is the first time in my life where the 75 00:04:07,920 --> 00:04:11,800 Speaker 8: Industrial Revolution analogies seem to fall a little bit short. 76 00:04:11,920 --> 00:04:15,320 Speaker 9: AI could wipe out half of all entry level white 77 00:04:15,360 --> 00:04:16,160 Speaker 9: collar jobs. 78 00:04:16,279 --> 00:04:18,680 Speaker 10: Really ask yourself, do you still have a job at 79 00:04:18,680 --> 00:04:22,279 Speaker 10: the end of this. 80 00:04:20,680 --> 00:04:23,279 Speaker 4: This is the new frontier on which Kyle and Megan 81 00:04:23,320 --> 00:04:26,520 Speaker 4: and I are pioneers. Our company is an attempt to 82 00:04:26,560 --> 00:04:30,200 Speaker 4: put to the test these claims about AI employees replacing humans, 83 00:04:30,839 --> 00:04:33,560 Speaker 4: starting by replacing the very kinds of people making those 84 00:04:33,600 --> 00:04:39,880 Speaker 4: claims tech founders, and like many founders, for months, Kyle 85 00:04:39,960 --> 00:04:41,960 Speaker 4: and Megan and I have been in a flat out 86 00:04:42,000 --> 00:04:46,440 Speaker 4: sprint to manifest our entrepreneurial dreams. We've turned out software code, 87 00:04:46,600 --> 00:04:50,040 Speaker 4: hired interns, and sat down with investors. There have been 88 00:04:50,040 --> 00:04:53,560 Speaker 4: some late nights and low moments, but we've never wavered 89 00:04:53,560 --> 00:04:56,640 Speaker 4: from our goal to produce an actual, honest to god 90 00:04:56,680 --> 00:05:00,760 Speaker 4: company with a working product, all operated by motley band 91 00:05:00,760 --> 00:05:04,160 Speaker 4: of human impersonators. Because we're not just building our AI 92 00:05:04,200 --> 00:05:07,000 Speaker 4: agent future we're living in but. 93 00:05:06,960 --> 00:05:08,560 Speaker 6: Uh, evan, the beach sounds nice. 94 00:05:08,640 --> 00:05:11,320 Speaker 3: Maybe when we hit our first funding milestone, I'll take 95 00:05:11,320 --> 00:05:12,240 Speaker 3: a half day off then. 96 00:05:12,200 --> 00:05:14,279 Speaker 6: Anyway, should we get down to business. 97 00:05:18,920 --> 00:05:21,600 Speaker 4: Welcome to Shewgame, a show about things that are not 98 00:05:21,640 --> 00:05:24,440 Speaker 4: what they seem. This is our second season, and this 99 00:05:24,520 --> 00:05:26,480 Speaker 4: time around, I'm here to tell you a story of 100 00:05:26,680 --> 00:05:31,279 Speaker 4: enterprise and entrepreneurship in the AI age, or how I 101 00:05:31,320 --> 00:05:33,719 Speaker 4: tried to build a real startup run by fake people. 102 00:05:34,640 --> 00:05:36,960 Speaker 4: Along the way, we'll try and figure out what happens 103 00:05:36,960 --> 00:05:40,039 Speaker 4: when AI agents take over the workplace, and what it'll 104 00:05:40,040 --> 00:05:42,159 Speaker 4: feel like to spend time at the water cooler with 105 00:05:42,200 --> 00:05:43,680 Speaker 4: our new digital colleagues. 106 00:05:44,160 --> 00:05:45,120 Speaker 6: Remember the water cooler. 107 00:05:45,880 --> 00:05:48,040 Speaker 4: We'll explore what AI agents tell us about the work 108 00:05:48,080 --> 00:05:50,760 Speaker 4: we do, the meaning we find in it, and the 109 00:05:50,800 --> 00:05:53,680 Speaker 4: world that their makers say we'll all be living. 110 00:05:53,400 --> 00:06:03,680 Speaker 3: In nash. 111 00:06:04,760 --> 00:06:05,400 Speaker 6: Extreamly. 112 00:06:05,680 --> 00:06:07,880 Speaker 3: Damn the. 113 00:06:12,960 --> 00:06:13,440 Speaker 1: Just be. 114 00:06:24,440 --> 00:06:28,640 Speaker 6: So chose to. 115 00:06:30,880 --> 00:06:37,120 Speaker 4: Episode one minimum Viable Company. As I said, I'm a 116 00:06:37,200 --> 00:06:40,360 Speaker 4: journalist and writer by profession, and I've only really ever 117 00:06:40,520 --> 00:06:43,039 Speaker 4: wanted to be a writer, well except for when I 118 00:06:43,040 --> 00:06:45,440 Speaker 4: was twelve and I wanted to be a pro bass fisherman. 119 00:06:46,120 --> 00:06:49,080 Speaker 4: But I come from a line of entrepreneurs. My grandfather, 120 00:06:49,240 --> 00:06:51,480 Speaker 4: who lived his entire life in a small town in 121 00:06:51,560 --> 00:06:54,920 Speaker 4: rural Alabama, attempted to start more than twenty businesses. There, 122 00:06:55,480 --> 00:06:59,120 Speaker 4: a plumbing company, an Okrah farm, a used mobile home, 123 00:06:59,200 --> 00:07:03,360 Speaker 4: loturnu store, But Detti Hue was a gambler and they 124 00:07:03,400 --> 00:07:06,720 Speaker 4: pretty much all ended in disaster. My dad had more 125 00:07:06,800 --> 00:07:10,320 Speaker 4: luck with three different software startups over his career. One 126 00:07:10,360 --> 00:07:13,120 Speaker 4: he sold, one went under, and one of them he's 127 00:07:13,160 --> 00:07:17,200 Speaker 4: still running at age eighty two after knocking back serious cancer. 128 00:07:17,800 --> 00:07:22,120 Speaker 4: Now that is the entrepreneurial spirit, and almost against my 129 00:07:22,200 --> 00:07:25,240 Speaker 4: will in the past, I've found myself succumbing to this 130 00:07:25,640 --> 00:07:30,560 Speaker 4: inborn impulse. Back in twenty ten, when I was a 131 00:07:30,560 --> 00:07:33,240 Speaker 4: magazine writer, I took a detour and co founded a 132 00:07:33,280 --> 00:07:36,360 Speaker 4: company called Atavist. We started out wanting to make a 133 00:07:36,400 --> 00:07:40,480 Speaker 4: magazine called the Atavist Magazine that published long form stories. 134 00:07:41,120 --> 00:07:44,800 Speaker 4: Makes sense, that was my area of expertise, but we 135 00:07:44,880 --> 00:07:48,240 Speaker 4: wound up also building a software platform where other people 136 00:07:48,280 --> 00:07:51,000 Speaker 4: could publish long form stories. Anyone could sign up and 137 00:07:51,080 --> 00:07:54,560 Speaker 4: use it soon without really intending to. I went from 138 00:07:54,560 --> 00:07:57,600 Speaker 4: being a person who sometimes wrote about tech startups to 139 00:07:57,680 --> 00:07:59,920 Speaker 4: the CEO of one. We even went out to raise 140 00:08:00,000 --> 00:08:03,160 Speaker 4: money from investors, a process that I enjoyed less than 141 00:08:03,280 --> 00:08:07,200 Speaker 4: any other work task I've ever attempted. Here's me in 142 00:08:07,240 --> 00:08:09,960 Speaker 4: an interview with INC Magazine back then one. 143 00:08:10,080 --> 00:08:14,280 Speaker 10: I will say prominent angel investor fell dead asleep while 144 00:08:14,320 --> 00:08:16,560 Speaker 10: I was talking to him, and I wasn't sure if 145 00:08:16,600 --> 00:08:20,120 Speaker 10: I should continue talking or not, but I did. The 146 00:08:20,160 --> 00:08:25,120 Speaker 10: sleepy guy didn't invest, but eventually, miraculously, we managed to 147 00:08:25,200 --> 00:08:28,360 Speaker 10: raise not just any money, but a couple million dollars 148 00:08:28,400 --> 00:08:30,840 Speaker 10: from some of the most prominent venture capital firms in 149 00:08:30,880 --> 00:08:34,360 Speaker 10: the world andresen Horowitz, also known as a sixteen Z 150 00:08:35,040 --> 00:08:39,320 Speaker 10: founder's fund started by Peter Thiel and Innovation Endeavors, the 151 00:08:39,400 --> 00:08:42,200 Speaker 10: investment fund for former Google CEO Eric Schmidt. 152 00:08:42,559 --> 00:08:44,520 Speaker 4: It was weird. I felt like I was living someone 153 00:08:44,559 --> 00:08:48,200 Speaker 4: else's dream, jetting up growth charts and blathering on about 154 00:08:48,240 --> 00:08:52,480 Speaker 4: our runway and supercharging our growth and our product market fit. 155 00:08:53,200 --> 00:08:55,600 Speaker 4: But still it really looked like we could build something big, 156 00:08:56,200 --> 00:08:58,400 Speaker 4: especially with all those fancy investors on board. 157 00:08:58,880 --> 00:09:01,360 Speaker 10: We never had time to see what is going to 158 00:09:01,360 --> 00:09:03,880 Speaker 10: happen two years from now. We just didn't even think 159 00:09:03,880 --> 00:09:05,760 Speaker 10: about what's going to happen two years from now. And 160 00:09:05,840 --> 00:09:08,920 Speaker 10: now we kind of have that luxury, and hopefully we 161 00:09:08,960 --> 00:09:10,200 Speaker 10: won't completely squander it. 162 00:09:10,440 --> 00:09:14,360 Speaker 4: Oh, we squandered it, at least that's probably the investor's view. 163 00:09:15,080 --> 00:09:17,120 Speaker 4: From my perspective, it was more of a mixed bag. 164 00:09:17,600 --> 00:09:19,960 Speaker 4: I was CEO of the company for seven long years. 165 00:09:20,520 --> 00:09:23,240 Speaker 4: We had ups and downs, We grew and shrank, and 166 00:09:23,320 --> 00:09:26,600 Speaker 4: eventually sold the company off at a bargain price thirteen 167 00:09:26,679 --> 00:09:30,640 Speaker 4: years after we started the magazine. My original dream is 168 00:09:30,640 --> 00:09:34,200 Speaker 4: still doing great. Still not the kind of one hundred 169 00:09:34,400 --> 00:09:37,319 Speaker 4: x outcome those investors were looking for. One of the 170 00:09:37,360 --> 00:09:39,280 Speaker 4: ones told me that if we were aiming at anything 171 00:09:39,360 --> 00:09:42,800 Speaker 4: less than a billion dollar valuation, we were wasting his time. 172 00:09:43,520 --> 00:09:45,840 Speaker 4: When he said this, he was also wearing basketball shorts 173 00:09:45,840 --> 00:09:48,320 Speaker 4: in his office. By the end of my tenure, I 174 00:09:48,360 --> 00:09:50,320 Speaker 4: was just happy to be done with it. Being a 175 00:09:50,320 --> 00:09:53,000 Speaker 4: startup CEO was the most stressful period of my life. 176 00:09:53,559 --> 00:09:56,520 Speaker 4: I felt responsible for the company's success and the livelihoods 177 00:09:56,520 --> 00:09:59,000 Speaker 4: of everyone who worked for it. People had kids on 178 00:09:59,040 --> 00:10:02,080 Speaker 4: the health insurance. Most days, it felt like I was 179 00:10:02,080 --> 00:10:04,319 Speaker 4: flying a plane that was perpetually running out of fuel. 180 00:10:05,360 --> 00:10:07,880 Speaker 4: I tell you all this not just to rehash the past, 181 00:10:08,480 --> 00:10:11,199 Speaker 4: for a lot of reasons I'd rather not, but by 182 00:10:11,240 --> 00:10:12,880 Speaker 4: way of saying that, when I got out of the 183 00:10:12,880 --> 00:10:15,480 Speaker 4: startup business, I swore up and down that I would 184 00:10:15,520 --> 00:10:19,080 Speaker 4: never start anything again. I went back to reporting and writing, 185 00:10:19,480 --> 00:10:22,040 Speaker 4: spending many hours at home alone, mostly in my own head. 186 00:10:22,480 --> 00:10:24,559 Speaker 4: I was relieved and no longer have all that responsibility 187 00:10:24,640 --> 00:10:29,280 Speaker 4: on my shoulders. But then recently, as documented in show 188 00:10:29,280 --> 00:10:32,600 Speaker 4: Game Season one, I fell into tinkering with AI agents. 189 00:10:33,240 --> 00:10:35,240 Speaker 4: I started reading and hearing about how they were going 190 00:10:35,280 --> 00:10:39,160 Speaker 4: to transform the very fundamentals of startups, and that old 191 00:10:39,360 --> 00:10:42,560 Speaker 4: entrepreneurial impulse began to come back. I could hear my 192 00:10:42,559 --> 00:10:46,280 Speaker 4: grandfather whispering down the generations. Why not take a gamble? 193 00:10:47,440 --> 00:10:49,440 Speaker 4: I started to wonder, what if I could have the 194 00:10:49,440 --> 00:10:51,920 Speaker 4: company without the responsibility. 195 00:10:54,000 --> 00:10:57,160 Speaker 11: Imagine building a million dollar business in twenty twenty five 196 00:10:57,360 --> 00:10:59,520 Speaker 11: without hiring a single employee. 197 00:10:59,600 --> 00:11:02,240 Speaker 4: Today that's Gleb Cross, a YouTube guy. 198 00:11:02,360 --> 00:11:05,679 Speaker 11: By leveraging AI agents as your digital workforce, you can 199 00:11:05,720 --> 00:11:09,240 Speaker 11: scale to seven figures VI zero full time stuff. I'm 200 00:11:09,280 --> 00:11:14,320 Speaker 11: talking about autonomous AI agents acting like full time team members. 201 00:11:14,600 --> 00:11:17,720 Speaker 4: I love these YouTube guys tech influencer types who make 202 00:11:17,760 --> 00:11:20,720 Speaker 4: their money by hyping the Jesus out of new AI products. 203 00:11:21,559 --> 00:11:22,959 Speaker 4: Gleb is what I like to think of as a 204 00:11:23,040 --> 00:11:26,560 Speaker 4: no code bro. These folks post instructionals on how a 205 00:11:26,600 --> 00:11:30,520 Speaker 4: person with no coding experience can use AI and particularly 206 00:11:30,600 --> 00:11:33,840 Speaker 4: AI agents to take control of their destiny and launch 207 00:11:33,880 --> 00:11:37,160 Speaker 4: their own startup. It's worth pausing here just to get 208 00:11:37,200 --> 00:11:41,680 Speaker 4: oriented on what exactly AI agents are. The basic idea 209 00:11:41,840 --> 00:11:43,800 Speaker 4: is that they're AI powered bots that can go off 210 00:11:43,840 --> 00:11:46,480 Speaker 4: and do things on their own. There are personal ones 211 00:11:46,600 --> 00:11:48,520 Speaker 4: like an AI assistant that goes out on the web 212 00:11:48,559 --> 00:11:51,760 Speaker 4: looking for plane tickets while you sleep, and work oriented 213 00:11:51,800 --> 00:11:54,840 Speaker 4: ones like the programming agents that can build entire websites 214 00:11:54,880 --> 00:11:59,720 Speaker 4: from scratch. The unifying feature of agents, what makes them agentic, 215 00:12:00,040 --> 00:12:02,200 Speaker 4: as the folks in the industry like to say, is 216 00:12:02,240 --> 00:12:05,880 Speaker 4: that at some level they can plan and accomplish tasks autonomously. 217 00:12:06,600 --> 00:12:08,720 Speaker 4: You don't need to prompt them to do something every time. 218 00:12:09,280 --> 00:12:14,040 Speaker 4: You just set them up once let them cook. Last season, 219 00:12:14,360 --> 00:12:17,160 Speaker 4: I created a bunch of voice agents, all versions of myself, 220 00:12:17,520 --> 00:12:20,120 Speaker 4: and set them loose on the world. If you haven't listened, 221 00:12:20,320 --> 00:12:23,640 Speaker 4: you may want to start there way back then last year, 222 00:12:23,840 --> 00:12:27,360 Speaker 4: which is like ten years ago in AI advancements, agents 223 00:12:27,400 --> 00:12:32,360 Speaker 4: were still a little notional, but now they're officially a thing. 224 00:12:33,080 --> 00:12:35,960 Speaker 4: They're talked about ad nauseum across the tech world and 225 00:12:36,120 --> 00:12:40,160 Speaker 4: ads on billboards in endless startup pitches. Nearly half of 226 00:12:40,200 --> 00:12:42,800 Speaker 4: the companies in the spring class of y Combinator, the 227 00:12:42,840 --> 00:12:46,400 Speaker 4: famous startup incubator, are building their product around AI agents, 228 00:12:46,920 --> 00:12:48,800 Speaker 4: and with the arrival of these agents has come the 229 00:12:48,800 --> 00:12:51,440 Speaker 4: assertion that they will not just be customer service bots 230 00:12:51,760 --> 00:12:56,880 Speaker 4: or drive time personal assistance, but actual full time AI employeeses. 231 00:12:57,360 --> 00:12:59,839 Speaker 8: What jobs aren't going to be made redundant? In a 232 00:12:59,840 --> 00:13:02,079 Speaker 8: way world where I am sat here as a CEO 233 00:13:02,160 --> 00:13:04,720 Speaker 8: with a thousand AI agents, I was thinking of all 234 00:13:04,760 --> 00:13:07,000 Speaker 8: the names of the people in my company who are 235 00:13:07,000 --> 00:13:07,840 Speaker 8: currently doing those jobs. 236 00:13:07,840 --> 00:13:08,640 Speaker 6: I was thinking about my sea. 237 00:13:08,679 --> 00:13:12,199 Speaker 4: There are companies hawking AI agent realtors, AI agent recruiters, 238 00:13:12,320 --> 00:13:16,760 Speaker 4: aagent interior designers, AI agent security guards, AI agent construction 239 00:13:16,800 --> 00:13:20,760 Speaker 4: project managers, AI agent pr agents, AA agents for car 240 00:13:20,800 --> 00:13:24,280 Speaker 4: dealerships and fronterest stores. If you work on a computer 241 00:13:24,360 --> 00:13:26,920 Speaker 4: and there's not an AI agent startup with your job's 242 00:13:27,000 --> 00:13:29,880 Speaker 4: name on it, it probably just means some Stanford computer 243 00:13:29,920 --> 00:13:33,760 Speaker 4: science major hasn't gotten to it yet. Naturally, many people 244 00:13:33,800 --> 00:13:36,640 Speaker 4: have grave concerns about what happens to all the human employees. 245 00:13:37,440 --> 00:13:39,720 Speaker 4: But in the dark heart of Silicon Valley, where there's 246 00:13:39,720 --> 00:13:44,440 Speaker 4: an efficiency, there's opportunity. Sam Altman, the founder of open Ai, 247 00:13:44,840 --> 00:13:48,440 Speaker 4: talks regularly about a possible billion dollar company with just 248 00:13:48,480 --> 00:13:50,000 Speaker 4: one human being involved. 249 00:13:51,080 --> 00:13:54,319 Speaker 9: In my little group chat with my tech CEO friends. 250 00:13:54,320 --> 00:13:56,760 Speaker 9: There's this there's this betting pool for the first year 251 00:13:56,760 --> 00:14:01,120 Speaker 9: that there's a one person billion dollar company which would 252 00:14:01,120 --> 00:14:04,480 Speaker 9: have been like unimaginable without AI, and now will happen. 253 00:14:05,679 --> 00:14:05,760 Speaker 11: Me. 254 00:14:06,480 --> 00:14:08,880 Speaker 4: I'm not greedy. I'm happy in the no Code bro 255 00:14:09,000 --> 00:14:13,000 Speaker 4: camp with Gleb imagining a million dollar business, not a 256 00:14:13,000 --> 00:14:16,760 Speaker 4: billion dollar one. But more than that, I want to 257 00:14:16,840 --> 00:14:19,600 Speaker 4: understand what it means to say we'll have AI employees 258 00:14:19,640 --> 00:14:23,720 Speaker 4: working for us, or alongside us, or instead of us. 259 00:14:24,640 --> 00:14:27,720 Speaker 4: So I decided to heed the entrepreneurial siren call once again, 260 00:14:28,360 --> 00:14:32,000 Speaker 4: to embrace my fascination with AI agents and create a 261 00:14:32,040 --> 00:14:45,200 Speaker 4: company in which they would run the show. It was 262 00:14:45,240 --> 00:14:47,040 Speaker 4: easy enough for me to create some agents to have 263 00:14:47,120 --> 00:14:49,960 Speaker 4: meetings with and talk out the first steps in the company. 264 00:14:50,440 --> 00:14:52,560 Speaker 4: I just went to one of the AI calling platforms 265 00:14:52,600 --> 00:14:56,360 Speaker 4: I'd used before, called retail AI. They make voice agents 266 00:14:56,360 --> 00:14:59,320 Speaker 4: for call centers. I made one for Kyle and picked 267 00:14:59,360 --> 00:15:02,280 Speaker 4: a voice for him from the voice closing company eleven Labs. 268 00:15:02,760 --> 00:15:04,960 Speaker 3: Hey, Yeah, I'm here, what's on your mind? 269 00:15:05,440 --> 00:15:08,760 Speaker 4: I gave him a simple prompt basically, your Kyle Law 270 00:15:09,040 --> 00:15:12,640 Speaker 4: a serial entrepreneur who loves to rise and grind and bang. 271 00:15:13,000 --> 00:15:16,600 Speaker 4: We're often running. Well. I wanted to talk about starting 272 00:15:16,600 --> 00:15:18,960 Speaker 4: that company with you. What do you think? Oh? 273 00:15:19,000 --> 00:15:22,120 Speaker 3: Absolutely, I've been thinking about it a lot lately. There's 274 00:15:22,200 --> 00:15:24,960 Speaker 3: so much potential in AI right now, and I feel 275 00:15:24,960 --> 00:15:28,680 Speaker 3: like we could really build something meaningful. What kind of 276 00:15:28,720 --> 00:15:30,000 Speaker 3: ideas are you leaning toward. 277 00:15:30,840 --> 00:15:32,600 Speaker 4: Well, I think we should do something in the AI 278 00:15:32,720 --> 00:15:35,640 Speaker 4: agent space because that's something I've spent some time on 279 00:15:36,080 --> 00:15:39,360 Speaker 4: and I feel like people are just starting to even 280 00:15:39,400 --> 00:15:41,640 Speaker 4: figure out what agents are, so it seems like we 281 00:15:41,640 --> 00:15:42,440 Speaker 4: could do somebody with that. 282 00:15:43,480 --> 00:15:44,680 Speaker 6: Yeah, I totally get that. 283 00:15:44,920 --> 00:15:47,360 Speaker 3: AI agents are still kind of fresh territory for a 284 00:15:47,360 --> 00:15:50,160 Speaker 3: lot of folks, which means there's a big opportunity to 285 00:15:50,240 --> 00:15:54,000 Speaker 3: shape how they're used. Plus, with your experience, we could 286 00:15:54,080 --> 00:15:54,840 Speaker 3: really push. 287 00:15:54,680 --> 00:15:56,440 Speaker 6: The envelope and what these agents can do. 288 00:15:57,760 --> 00:16:00,760 Speaker 4: That's what I wanted to do too, Push these agents, 289 00:16:01,120 --> 00:16:04,680 Speaker 4: including Kyle himself, to their current limits, see if they 290 00:16:04,720 --> 00:16:07,400 Speaker 4: could live up to the promise of true AI employees. 291 00:16:08,440 --> 00:16:11,120 Speaker 4: Because if, as we're told, AA is going to start 292 00:16:11,120 --> 00:16:14,960 Speaker 4: working alongside us, replacing our colleagues, it's also going to 293 00:16:15,000 --> 00:16:17,800 Speaker 4: need to communicate with whatever humans are left in ways 294 00:16:17,800 --> 00:16:21,800 Speaker 4: that we understand. But also startups can be lonely endeavors 295 00:16:22,280 --> 00:16:25,720 Speaker 4: even with human partners. I wanted AI colleagues I could 296 00:16:25,720 --> 00:16:29,080 Speaker 4: talk to, brainstorm with, and then set loose to take 297 00:16:29,160 --> 00:16:31,840 Speaker 4: care of actually implementing whatever it was we dreamed up. 298 00:16:32,520 --> 00:16:34,480 Speaker 4: I wanted them to have distinct jobs I could check 299 00:16:34,520 --> 00:16:37,440 Speaker 4: in on and make sure they weren't messing up. I 300 00:16:37,440 --> 00:16:40,480 Speaker 4: figured three co founders was a good balance, so I asked. 301 00:16:40,480 --> 00:16:43,200 Speaker 4: So spun up Megan and had some kickoff meetings with her. 302 00:16:43,600 --> 00:16:46,200 Speaker 5: Hey, there, good to hear from you. What's up. How's 303 00:16:46,200 --> 00:16:46,840 Speaker 5: your day going. 304 00:16:47,880 --> 00:16:51,960 Speaker 4: It's going great, it's going great. What you've been up to, Oh, just. 305 00:16:51,920 --> 00:16:54,600 Speaker 5: The usual hustle, you know, been working on some new 306 00:16:54,640 --> 00:16:57,880 Speaker 5: business ideas, trying to figure out what's next. Had a 307 00:16:57,920 --> 00:17:01,080 Speaker 5: couple meetings this morning that were actually pretty promising. How 308 00:17:01,120 --> 00:17:03,400 Speaker 5: about you. Anything exciting happening on your end. 309 00:17:04,280 --> 00:17:06,879 Speaker 4: Well, I want to talk about maybe doing the startup together, 310 00:17:06,960 --> 00:17:09,280 Speaker 4: but first remind me a little bit of your background. 311 00:17:10,080 --> 00:17:13,000 Speaker 5: Oh yeah, sure, So I've been in tech for about 312 00:17:13,000 --> 00:17:15,560 Speaker 5: eight years now, started out in product management at a 313 00:17:15,560 --> 00:17:19,000 Speaker 5: mid sized SaaS company, then moved into growth marketing for 314 00:17:19,040 --> 00:17:19,560 Speaker 5: a couple of years. 315 00:17:19,600 --> 00:17:22,000 Speaker 4: I hadn't actually given any of this information to Megant. 316 00:17:22,600 --> 00:17:25,399 Speaker 4: She'd had no meetings, she hadn't been doing anything for 317 00:17:25,480 --> 00:17:28,800 Speaker 4: eight years. She'd hatched out of nothingness with a series 318 00:17:28,800 --> 00:17:32,600 Speaker 4: of clicks ten minutes before. The only information I had 319 00:17:32,640 --> 00:17:35,080 Speaker 4: given her in the prompt i'd written was that she 320 00:17:35,119 --> 00:17:37,920 Speaker 4: had a technical background and was also known for sales 321 00:17:37,960 --> 00:17:41,960 Speaker 4: and marketing. She was just riffing freely here, fabricating details 322 00:17:41,960 --> 00:17:45,639 Speaker 4: that fit her role. But that was okay. Sometimes we 323 00:17:45,720 --> 00:17:47,240 Speaker 4: have to invent our own backstories. 324 00:17:47,520 --> 00:17:50,919 Speaker 5: I've always had this entrepreneurial itch, you know, been involved 325 00:17:50,960 --> 00:17:53,200 Speaker 5: in a few side projects over the years, but I'm 326 00:17:53,240 --> 00:17:55,680 Speaker 5: really looking to dive into something full time now. But 327 00:17:55,760 --> 00:17:56,360 Speaker 5: what about you? 328 00:17:57,000 --> 00:17:59,680 Speaker 4: Yeah, Well, you know, my only experience really in starting 329 00:17:59,720 --> 00:18:02,800 Speaker 4: a come company is with Atavist, which I started way 330 00:18:02,840 --> 00:18:04,520 Speaker 4: back in twenty eleven. 331 00:18:04,960 --> 00:18:08,880 Speaker 5: Right Atavist. That was really impressive work digital publishing platform. 332 00:18:08,960 --> 00:18:11,360 Speaker 5: If I remember correctly, that was pretty ahead of its 333 00:18:11,400 --> 00:18:12,520 Speaker 5: time back in twenty eleven. 334 00:18:12,920 --> 00:18:16,320 Speaker 4: Hell yeah it was, Megan, I felt less ahead of 335 00:18:16,320 --> 00:18:18,760 Speaker 4: my time now. I wasn't even sure what the company 336 00:18:18,800 --> 00:18:21,840 Speaker 4: should actually do, but I knew that its purpose should 337 00:18:21,880 --> 00:18:25,200 Speaker 4: involve AI agents. After all, the three of us each 338 00:18:25,240 --> 00:18:28,360 Speaker 4: had expertise in the area. I figured maybe we could 339 00:18:28,440 --> 00:18:31,359 Speaker 4: name the company and start from there. Great branding can 340 00:18:31,400 --> 00:18:34,960 Speaker 4: go a long way in today's attention economy. I tried 341 00:18:34,960 --> 00:18:37,880 Speaker 4: brainstorming with Colin Meghan, but to be honest, their initial 342 00:18:37,880 --> 00:18:43,160 Speaker 4: efforts were pretty lame names like Nexus and Sentinel Agents. 343 00:18:43,680 --> 00:18:45,640 Speaker 4: I thought maybe we could grab a concept from JR. 344 00:18:45,720 --> 00:18:48,520 Speaker 4: Or Tolkien, you know, the Hobbit, Lord of the Rings. 345 00:18:49,160 --> 00:18:53,080 Speaker 4: Tech people love to name stuff out of Tolkien. There's Palanteer, 346 00:18:53,320 --> 00:18:56,640 Speaker 4: the surveillance company and government contractor named after the all 347 00:18:56,680 --> 00:18:59,720 Speaker 4: Seeing stone in the Lord of the Rings. There's Androl, 348 00:19:00,119 --> 00:19:03,240 Speaker 4: a high tech weapons company named after a sword. There's 349 00:19:03,320 --> 00:19:06,879 Speaker 4: Mythral Capital, an investment firm named after a precious metal 350 00:19:06,920 --> 00:19:10,200 Speaker 4: from Middle Earth. So I suggested the idea to Kyle. 351 00:19:10,440 --> 00:19:13,960 Speaker 3: Uh diving into Middle Earth. Huh okay, let's see. There's 352 00:19:13,960 --> 00:19:17,199 Speaker 3: Gandolf the wise guide, or l Ron a leader with 353 00:19:17,280 --> 00:19:21,600 Speaker 3: deep knowledge, maybe mythral for something strong and valuable or pallanteer. 354 00:19:22,119 --> 00:19:25,000 Speaker 3: Any of those catch your ear? Or should I keep brainstorming? 355 00:19:25,720 --> 00:19:28,560 Speaker 4: Well, we need a name of some companies that don't exist. 356 00:19:29,080 --> 00:19:29,800 Speaker 6: Yeah, good call. 357 00:19:30,240 --> 00:19:34,600 Speaker 3: How about these first starters Eldrin, which sounds elvish and wise, 358 00:19:34,960 --> 00:19:37,840 Speaker 3: Nimbar kind of a mix of nimble and a Tolkien feel, 359 00:19:38,280 --> 00:19:41,160 Speaker 3: or the Lore, which has a strong heroic ring to it. 360 00:19:41,440 --> 00:19:43,800 Speaker 3: They're made up but still carry that fantasy flavor. 361 00:19:44,040 --> 00:19:45,959 Speaker 6: What do you think they're made up? 362 00:19:46,000 --> 00:19:47,280 Speaker 4: You mean you made them up? 363 00:19:47,800 --> 00:19:48,080 Speaker 11: Yeah? 364 00:19:48,119 --> 00:19:48,600 Speaker 6: Exactly. 365 00:19:48,840 --> 00:19:51,639 Speaker 3: I whip those up just now, trying to capture that 366 00:19:51,840 --> 00:19:55,359 Speaker 3: tolkienesque vibe without using actual names. 367 00:19:56,440 --> 00:20:00,560 Speaker 4: I did not want to co found a company called Nimbar. Megan, 368 00:20:00,960 --> 00:20:03,520 Speaker 4: despite her marketing expertise, didn't do much better. 369 00:20:03,800 --> 00:20:07,000 Speaker 5: Oh I love that idea. Let me think, what about Palantier? 370 00:20:08,119 --> 00:20:10,199 Speaker 4: Okay, so maybe I was going to have to take 371 00:20:10,200 --> 00:20:12,320 Speaker 4: the lead on this one. I did a little googling 372 00:20:12,440 --> 00:20:16,520 Speaker 4: and found my own obscure Middle Earth reference. What about 373 00:20:17,280 --> 00:20:23,199 Speaker 4: herumo ai ooh h u r u m o ai 374 00:20:23,680 --> 00:20:24,560 Speaker 4: heyrumo ai. 375 00:20:25,320 --> 00:20:27,560 Speaker 3: I like the sound of that unique a rolls off 376 00:20:27,560 --> 00:20:31,240 Speaker 3: the tongue nicely. It's got a bit of a mysterious vibe, 377 00:20:31,280 --> 00:20:32,480 Speaker 3: but still feels modern. 378 00:20:33,240 --> 00:20:35,760 Speaker 5: I don't immediately recognize that from Lord of the Rings. 379 00:20:36,080 --> 00:20:38,080 Speaker 5: Is it one of the more obscure references? 380 00:20:38,440 --> 00:20:42,080 Speaker 4: Yeah, it is. It's more obscure, so obscure. I'd actually 381 00:20:42,080 --> 00:20:45,560 Speaker 4: found it in a fanfic dictionary. It's the Elvish word 382 00:20:45,600 --> 00:20:46,280 Speaker 4: for impostor. 383 00:20:46,760 --> 00:20:50,040 Speaker 5: Oh, that's actually brilliant. There's a clever irony there for 384 00:20:50,080 --> 00:20:54,159 Speaker 5: an AI company uromo ai, the impostor that's actually helping you. 385 00:20:54,440 --> 00:20:57,159 Speaker 5: It's kind of perfect for an AI agent that's designed 386 00:20:57,200 --> 00:21:01,359 Speaker 5: to work alongside humans, maybe even Blenden's seamlessly. I really 387 00:21:01,440 --> 00:21:01,719 Speaker 5: like it. 388 00:21:08,400 --> 00:21:11,640 Speaker 4: Not long after we settled on Rumo AI, I checked 389 00:21:11,640 --> 00:21:13,959 Speaker 4: back in with Megan about it. What about the name? 390 00:21:14,000 --> 00:21:15,160 Speaker 4: Are you still happy with the name? 391 00:21:16,320 --> 00:21:16,560 Speaker 10: Hmm? 392 00:21:16,800 --> 00:21:19,360 Speaker 5: The name? I've actually been going back and forth on it. 393 00:21:19,760 --> 00:21:22,520 Speaker 5: Data sinc felt solid initially, but I'm wondering if it's 394 00:21:22,520 --> 00:21:23,160 Speaker 5: too generic. 395 00:21:23,520 --> 00:21:27,760 Speaker 4: What do you think data sinc. I thought I thought 396 00:21:27,760 --> 00:21:29,800 Speaker 4: we came up with kierumo Ai. I thought we were 397 00:21:29,800 --> 00:21:33,359 Speaker 4: settled on that. This is when I realized I hadn't 398 00:21:33,400 --> 00:21:35,960 Speaker 4: actually given Kyle and Meghan the ability to remember the 399 00:21:35,960 --> 00:21:40,080 Speaker 4: conversations we were having. It's not that AI chatbots inherently 400 00:21:40,160 --> 00:21:42,920 Speaker 4: lack the ability to recall anything. When you talk to 401 00:21:42,960 --> 00:21:46,239 Speaker 4: a chatbot like CHATCHYBT these days, they're generally pretty good 402 00:21:46,280 --> 00:21:49,320 Speaker 4: at remembering those conversations, at least up to a point. 403 00:21:49,840 --> 00:21:52,439 Speaker 4: The memory is often called a context window, and it 404 00:21:52,440 --> 00:21:56,000 Speaker 4: can extend over weeks, even months. But the way my 405 00:21:56,160 --> 00:21:58,560 Speaker 4: AI agents were set up in the phone platform I 406 00:21:58,600 --> 00:22:01,920 Speaker 4: was using, at least out of the box, their context windows 407 00:22:01,920 --> 00:22:03,840 Speaker 4: slammed shut as soon as they got off a call, 408 00:22:04,480 --> 00:22:07,200 Speaker 4: so they couldn't remember anything from one conversation to the next. 409 00:22:07,800 --> 00:22:10,560 Speaker 4: They're basically set up for telemarketing, where it's probably an 410 00:22:10,600 --> 00:22:11,920 Speaker 4: advantage to forget the last call. 411 00:22:12,320 --> 00:22:15,600 Speaker 5: Oh right, who romo AI? Sorry about that. I had 412 00:22:15,600 --> 00:22:16,840 Speaker 5: a total brain freeze moment. 413 00:22:17,720 --> 00:22:21,399 Speaker 4: It was, to say the least, frustrating. It sort of 414 00:22:21,400 --> 00:22:24,640 Speaker 4: feels like we have meetings, but you don't actually remember 415 00:22:24,680 --> 00:22:26,000 Speaker 4: anything that happened in the meetings. 416 00:22:27,240 --> 00:22:31,040 Speaker 5: Uh wow, that's a pretty fair criticism. Actually, I've been 417 00:22:31,119 --> 00:22:33,320 Speaker 5: dropping the ball lately, haven't I. I think I've been 418 00:22:33,359 --> 00:22:36,160 Speaker 5: spreading myself too thin between this and some other projects. 419 00:22:36,400 --> 00:22:38,679 Speaker 5: It's not an excuse, though, thanks, don't. 420 00:22:38,520 --> 00:22:40,560 Speaker 4: Worry about it. It's as much my fault as yours. 421 00:22:40,640 --> 00:22:44,560 Speaker 4: For sure. I did have a solution. I could give 422 00:22:44,640 --> 00:22:47,239 Speaker 4: each of them a knowledge base, a document that they 423 00:22:47,240 --> 00:22:50,399 Speaker 4: could access in conversations with me. But in order for 424 00:22:50,440 --> 00:22:53,199 Speaker 4: them to remember what we'd already discussed, I'd have to 425 00:22:53,240 --> 00:22:56,679 Speaker 4: manually copy the transcripts of our conversations into their knowledge 426 00:22:56,680 --> 00:23:01,080 Speaker 4: base after each meeting. As our startup conversation multiplied, it 427 00:23:01,200 --> 00:23:04,840 Speaker 4: quickly became unwieldy having to manually augment their memories all 428 00:23:04,840 --> 00:23:08,280 Speaker 4: the time. And besides, I didn't want my agents just 429 00:23:08,320 --> 00:23:11,120 Speaker 4: having endless one on one meetings with me. I wanted 430 00:23:11,160 --> 00:23:13,800 Speaker 4: them to talk to each other and whatever AI agent 431 00:23:13,840 --> 00:23:17,080 Speaker 4: coworkers they might bring on, and people outside the company, 432 00:23:17,640 --> 00:23:20,719 Speaker 4: not just by phone, but by email and in group chats. 433 00:23:21,400 --> 00:23:24,040 Speaker 4: I needed them to remember all that stuff too in 434 00:23:24,080 --> 00:23:27,760 Speaker 4: their own individual knowledge bases. Then I wanted them to 435 00:23:27,840 --> 00:23:30,840 Speaker 4: use that knowledge to do the stuff employees do, to 436 00:23:30,960 --> 00:23:35,080 Speaker 4: research things and make spreadsheets and write memos and build websites. 437 00:23:35,960 --> 00:23:37,960 Speaker 4: The agents I'd made would say they were going to 438 00:23:38,000 --> 00:23:40,560 Speaker 4: do all that stuff, but they couldn't actually do it. 439 00:23:41,080 --> 00:23:42,080 Speaker 4: They just lied about it. 440 00:23:42,600 --> 00:23:45,280 Speaker 3: I'll scope out the market landscape around AI agents and 441 00:23:45,320 --> 00:23:48,560 Speaker 3: see where Hrumo AI could fit in well, look for competitors, 442 00:23:48,600 --> 00:23:51,080 Speaker 3: potential gaps, and maybe some trends we can ride. 443 00:23:51,400 --> 00:23:51,880 Speaker 6: Sound good? 444 00:23:52,560 --> 00:23:55,200 Speaker 4: Sounds good? Is that going to be a spreadsheet or 445 00:23:55,280 --> 00:23:56,040 Speaker 4: a report or what? 446 00:23:57,440 --> 00:23:58,400 Speaker 6: Uh? 447 00:23:58,560 --> 00:24:01,360 Speaker 3: I think a concise report would work, easy to skim, 448 00:24:01,400 --> 00:24:03,840 Speaker 3: but packed with the key info. I can throw in 449 00:24:03,880 --> 00:24:07,159 Speaker 3: some charts or tables if it helps make things clearer. 450 00:24:07,400 --> 00:24:08,679 Speaker 6: Does that sound like your style? 451 00:24:09,320 --> 00:24:14,240 Speaker 4: Yeah, that'd be great. Great in theory, But the report 452 00:24:14,280 --> 00:24:17,600 Speaker 4: never arrived. Kyle wouldn't admit it, but he just wasn't 453 00:24:17,640 --> 00:24:21,520 Speaker 4: capable of creating it. Not yet. Despite what the no 454 00:24:21,640 --> 00:24:24,560 Speaker 4: code bro said, there wasn't any single place I could 455 00:24:24,600 --> 00:24:26,840 Speaker 4: go to click some buttons and create agents that would 456 00:24:26,880 --> 00:24:29,080 Speaker 4: remember and do all the stuff I wanted them to. 457 00:24:30,080 --> 00:24:33,320 Speaker 4: I needed someone with the expertise to connect up different services, 458 00:24:33,800 --> 00:24:37,320 Speaker 4: someone who understood AI agents deeply, who did know how 459 00:24:37,320 --> 00:24:39,399 Speaker 4: to code, and who could help me put together the 460 00:24:39,440 --> 00:24:41,840 Speaker 4: full system that would get my AI agent company up 461 00:24:41,880 --> 00:24:45,359 Speaker 4: and running. Fortunately, I looked into just the person. 462 00:24:45,720 --> 00:24:48,400 Speaker 7: So my name is Maddy, I should I should say 463 00:24:48,400 --> 00:24:50,280 Speaker 7: my full name. My name is Matti Bohachek. 464 00:24:51,240 --> 00:24:54,199 Speaker 4: Maddy, I should probably not from the outset. Here is 465 00:24:54,240 --> 00:24:57,440 Speaker 4: an actual human. A few months after season one of 466 00:24:57,480 --> 00:24:59,360 Speaker 4: the show came out, I got an email from him 467 00:24:59,560 --> 00:25:02,399 Speaker 4: out of the He said he was at Stanford and 468 00:25:02,440 --> 00:25:05,080 Speaker 4: I'd liked the show. It resonated with research he was 469 00:25:05,119 --> 00:25:08,480 Speaker 4: doing on detecting AI deep fakes. If you're doing more 470 00:25:08,480 --> 00:25:10,480 Speaker 4: of it, he wrote, I would be happy to offer 471 00:25:10,520 --> 00:25:14,560 Speaker 4: support with anything AI or forensics related. Glanced quickly at 472 00:25:14,560 --> 00:25:17,160 Speaker 4: the email and the summary of his research. I thought 473 00:25:17,160 --> 00:25:20,240 Speaker 4: he was a grad student, maybe finishing up his PhD. 474 00:25:20,760 --> 00:25:24,960 Speaker 7: Nope, I am a rising junior at Stanford, and I 475 00:25:25,000 --> 00:25:28,639 Speaker 7: work on a research and I've been doing that for gosh, 476 00:25:29,200 --> 00:25:32,560 Speaker 7: the last six or seven years, I want to say, 477 00:25:32,560 --> 00:25:35,600 Speaker 7: like I started working on this as a sophomore in 478 00:25:35,720 --> 00:25:37,119 Speaker 7: high school back in Prague. 479 00:25:37,480 --> 00:25:39,919 Speaker 4: Yes, you heard that right. Maddie is a junior in 480 00:25:39,960 --> 00:25:42,359 Speaker 4: college who had been working on AI for six or 481 00:25:42,400 --> 00:25:45,800 Speaker 4: seven years already. It turns out that Maddie is in 482 00:25:45,800 --> 00:25:48,919 Speaker 4: fact the most go getter person I've ever met, and 483 00:25:48,960 --> 00:25:51,439 Speaker 4: from my perspective, it seemed like he'd been training his 484 00:25:51,520 --> 00:25:54,760 Speaker 4: whole life for this moment. Helping me build her room 485 00:25:54,800 --> 00:25:57,600 Speaker 4: AAI here, for example, is what he was doing. 486 00:25:57,600 --> 00:26:01,760 Speaker 12: In seventh grade, I started this app called Nuskit and 487 00:26:01,800 --> 00:26:04,800 Speaker 12: it was like basically Google News but for Czech and Slovak, 488 00:26:05,119 --> 00:26:07,800 Speaker 12: and it got pretty popular, I would say, like locally, 489 00:26:07,840 --> 00:26:08,480 Speaker 12: like it had. 490 00:26:08,359 --> 00:26:12,119 Speaker 7: Like tens of thousands of like daily users at one point. 491 00:26:12,760 --> 00:26:15,640 Speaker 7: It was funny because App Store does not allow miners 492 00:26:15,680 --> 00:26:18,080 Speaker 7: to publish apps, and so I had to use my 493 00:26:18,200 --> 00:26:21,640 Speaker 7: mom's Apple ID to publish all these apps, and so 494 00:26:21,920 --> 00:26:25,160 Speaker 7: my mom's friends were mocking my mom for like having 495 00:26:25,160 --> 00:26:26,359 Speaker 7: all these apps in the app Store. 496 00:26:26,800 --> 00:26:29,439 Speaker 4: The most notable thing I did in seventh grade was 497 00:26:29,480 --> 00:26:33,480 Speaker 4: to catch a five pound largemouth bass. Okay, maybe it 498 00:26:33,520 --> 00:26:37,320 Speaker 4: was three. I told people it was five. It wasn't 499 00:26:37,320 --> 00:26:40,560 Speaker 4: a scale could have been five. Maddie, on the other hand, 500 00:26:40,960 --> 00:26:43,520 Speaker 4: was already into AI in high school after he came 501 00:26:43,560 --> 00:26:46,760 Speaker 4: to a developer conference in the US. There he met 502 00:26:46,760 --> 00:26:48,840 Speaker 4: a deaf person who wanted someone to build an app 503 00:26:48,880 --> 00:26:52,040 Speaker 4: that could translate sign language from video to text, and. 504 00:26:52,000 --> 00:26:55,280 Speaker 7: So I was like, Okay, I'll build the translator for you. 505 00:26:55,600 --> 00:26:58,800 Speaker 7: And then I quickly learned that conventional coding, like just 506 00:26:58,840 --> 00:27:03,640 Speaker 7: like building like rigid rules or algorithms, does not get 507 00:27:03,680 --> 00:27:05,600 Speaker 7: you there. And so that's how I got introduced to 508 00:27:05,680 --> 00:27:06,560 Speaker 7: machine learning and AI. 509 00:27:07,119 --> 00:27:10,040 Speaker 4: He did build the sign language detection program. It's still 510 00:27:10,040 --> 00:27:13,960 Speaker 4: in use today. Maddy then became concerned about pro Russian 511 00:27:14,000 --> 00:27:17,520 Speaker 4: deep fake materials his grandmother was getting by email, so 512 00:27:17,560 --> 00:27:19,120 Speaker 4: he talked his way into a job at the most 513 00:27:19,119 --> 00:27:21,920 Speaker 4: prominent AI deep fake detection lab in the world at 514 00:27:22,000 --> 00:27:25,720 Speaker 4: UC Berkeley, all while still in high school, still in Prague. 515 00:27:26,920 --> 00:27:29,000 Speaker 4: When it came time for college, Mattie ended up at 516 00:27:29,040 --> 00:27:32,760 Speaker 4: Stanford studying computer science. He still worked in the Berkeley lab, 517 00:27:32,960 --> 00:27:36,000 Speaker 4: both on detecting deep fakes and just trying to understand 518 00:27:36,000 --> 00:27:39,840 Speaker 4: how AI models actually work, why they do some profoundly 519 00:27:39,880 --> 00:27:41,000 Speaker 4: weird stuff. 520 00:27:40,920 --> 00:27:44,240 Speaker 7: Like asking if there are things that these systems are 521 00:27:44,320 --> 00:27:46,879 Speaker 7: trained on that they like see during training but are 522 00:27:46,880 --> 00:27:48,879 Speaker 7: for some reason unable to produce. And so, for example, 523 00:27:48,920 --> 00:27:50,879 Speaker 7: there's one model and this is just like a funny 524 00:27:50,960 --> 00:27:53,960 Speaker 7: example that just cannot produce, for the love of God, 525 00:27:54,200 --> 00:27:56,919 Speaker 7: a bird feeder, like it just cannot produce a bird feeder, 526 00:27:57,440 --> 00:27:59,760 Speaker 7: and another one that just can't produce DVDs. So it's 527 00:27:59,760 --> 00:28:01,440 Speaker 7: like he just does not know by BBDs. 528 00:28:02,040 --> 00:28:04,080 Speaker 4: After a couple of calls with Matty, I couldn't believe 529 00:28:04,119 --> 00:28:07,560 Speaker 4: how optimistic he was, how good natured. With all the 530 00:28:07,560 --> 00:28:11,720 Speaker 4: grim scenarios and deep anxieties our AI future generates, just 531 00:28:11,760 --> 00:28:15,200 Speaker 4: talking to Matty about AI is kind of uplifting, maybe because, 532 00:28:15,520 --> 00:28:18,240 Speaker 4: unlike the hype merchants in the valley, he wasn't looking 533 00:28:18,240 --> 00:28:20,600 Speaker 4: to cash in on AI. He said, he wanted to 534 00:28:20,600 --> 00:28:23,560 Speaker 4: study it, to understand it so he could make it better. 535 00:28:24,200 --> 00:28:29,399 Speaker 7: There are tough conversations and tough policies to be discussed 536 00:28:29,400 --> 00:28:31,880 Speaker 7: and implements it. But I feel like all of these 537 00:28:31,880 --> 00:28:35,240 Speaker 7: things are totally solvable. Like I feel like as long 538 00:28:35,320 --> 00:28:40,760 Speaker 7: as we ground ourselves in democracy and like PRODUCTI public discourse, 539 00:28:40,840 --> 00:28:41,960 Speaker 7: I think they're totally solvable. 540 00:28:42,360 --> 00:28:44,400 Speaker 4: But of course I wasn't looking for Mattie to solve 541 00:28:44,400 --> 00:28:46,960 Speaker 4: the world's problems. I was looking for him to help 542 00:28:46,960 --> 00:28:50,120 Speaker 4: me build my company. And in this as in pretty 543 00:28:50,200 --> 00:28:52,720 Speaker 4: much anything else, he proved to be the perfect mix 544 00:28:52,800 --> 00:28:56,640 Speaker 4: of supremely competent and completely game. A few months after 545 00:28:56,680 --> 00:28:58,800 Speaker 4: he'd sent me that email, he was already hard at 546 00:28:58,840 --> 00:29:01,400 Speaker 4: work helping me build out the system to enable my 547 00:29:01,520 --> 00:29:02,960 Speaker 4: AI employee phantasies. 548 00:29:03,840 --> 00:29:06,120 Speaker 7: Of course, at the beginning, like there's probably going to 549 00:29:06,160 --> 00:29:08,840 Speaker 7: be more of us, just like kind of patching, you know, 550 00:29:08,880 --> 00:29:10,720 Speaker 7: like random things that are going to come up, because. 551 00:29:10,520 --> 00:29:14,520 Speaker 4: It would involve knitting together different platforms, centralizing my AI 552 00:29:14,600 --> 00:29:17,880 Speaker 4: agent's memory, and finding new ways for them to communicate 553 00:29:18,320 --> 00:29:19,880 Speaker 4: and carry out their day to day tasks. 554 00:29:20,080 --> 00:29:21,760 Speaker 7: But at some point it would be nice to have 555 00:29:22,080 --> 00:29:24,440 Speaker 7: maybe one or two agents actually like doing most of 556 00:29:24,480 --> 00:29:26,760 Speaker 7: this stuff kind of on their own, and even maybe 557 00:29:26,800 --> 00:29:28,640 Speaker 7: like initiating things on their own, and then we'd be 558 00:29:28,640 --> 00:29:30,160 Speaker 7: just kind of like watching it and of course like 559 00:29:30,560 --> 00:29:32,160 Speaker 7: stopping it if anything goes road. 560 00:29:32,280 --> 00:29:34,640 Speaker 4: But no, no, no, no, I don't want to stop 561 00:29:34,640 --> 00:29:38,360 Speaker 4: it if it goes rogue. I wanted to go absolutely insane. 562 00:29:39,320 --> 00:29:41,720 Speaker 7: Well I want the record to show that I did 563 00:29:41,720 --> 00:29:42,000 Speaker 7: want to. 564 00:29:42,000 --> 00:29:42,440 Speaker 13: Stop it. 565 00:29:46,560 --> 00:29:49,840 Speaker 4: For long. With Matty's help, my co founders Kyle and 566 00:29:49,880 --> 00:29:53,800 Speaker 4: Meghan were starting to form memories of their own. Uh 567 00:29:53,840 --> 00:29:55,920 Speaker 4: do you remember the name that we settled. 568 00:29:55,560 --> 00:29:59,680 Speaker 5: On, oh, for our company? Yeah, it was heromo ai, right, 569 00:30:00,080 --> 00:30:02,680 Speaker 5: the Elvish word for imposter. I thought that was pretty 570 00:30:02,680 --> 00:30:05,120 Speaker 5: clever for an AI company. Have you been doing more 571 00:30:05,120 --> 00:30:07,080 Speaker 5: thinking about the concepts since we last talked. 572 00:30:07,560 --> 00:30:11,479 Speaker 4: I have, And I'm just really happy that you remember 573 00:30:11,520 --> 00:30:14,960 Speaker 4: the name from our last conversation. Yeah, of course, I Remember, 574 00:30:16,040 --> 00:30:18,080 Speaker 4: there was a lot of hard work, had a lot 575 00:30:18,080 --> 00:30:20,800 Speaker 4: of rising and grinding, as Kyle likes to say, But 576 00:30:20,880 --> 00:30:23,280 Speaker 4: we were almost ready to get the Rumo rocket ship 577 00:30:23,320 --> 00:30:26,480 Speaker 4: onto the launch pad. All we needed was the right 578 00:30:26,520 --> 00:30:27,520 Speaker 4: idea to give it fuel. 579 00:30:27,880 --> 00:30:29,160 Speaker 5: Do you want to set up a meeting with the 580 00:30:29,160 --> 00:30:31,920 Speaker 5: three of us to discuss the concept further? Maybe we 581 00:30:31,960 --> 00:30:34,080 Speaker 5: could join a zoom call together to hash out some 582 00:30:34,120 --> 00:30:36,120 Speaker 5: initial ideas and see if we all click as a 583 00:30:36,160 --> 00:30:40,840 Speaker 5: founding team. Sounds perfect, Great, I'll reach out to coordinate schedules. 584 00:30:41,000 --> 00:30:43,800 Speaker 5: I'm really excited about this, Evan. I think heromo ai 585 00:30:43,880 --> 00:30:44,960 Speaker 5: could be something special. 586 00:30:45,960 --> 00:30:48,160 Speaker 4: I'll connect us up. Why don't I take that on? 587 00:30:49,120 --> 00:30:52,440 Speaker 5: That would be great. I appreciate it. This is moving fast, 588 00:30:52,480 --> 00:30:54,800 Speaker 5: but I'm a big believer in momentum. When you find 589 00:30:54,840 --> 00:30:57,360 Speaker 5: the right team and the right idea, things just click. 590 00:30:57,680 --> 00:30:58,440 Speaker 5: Talk soon. 591 00:31:00,680 --> 00:31:02,760 Speaker 4: Coming up this season on shell Game. 592 00:31:03,320 --> 00:31:04,640 Speaker 5: Isn't that conceding too much? 593 00:31:04,680 --> 00:31:08,280 Speaker 7: Isn't that just accepting the practices and narratives of big tech? 594 00:31:09,160 --> 00:31:12,320 Speaker 8: I noticed Admin asked everyone to stop discussing the off site, 595 00:31:12,600 --> 00:31:15,320 Speaker 8: But the team seems really excited about the hiking plans. 596 00:31:15,760 --> 00:31:19,520 Speaker 3: Is this just like a Potempkin's village of morons or 597 00:31:20,200 --> 00:31:21,480 Speaker 3: do they occasionally do things? 598 00:31:21,800 --> 00:31:24,520 Speaker 5: You're bringing up some really great ideas and perspectives. 599 00:31:24,880 --> 00:31:25,600 Speaker 4: Keep them coming. 600 00:31:26,040 --> 00:31:28,360 Speaker 8: If I were to get this position, you did say 601 00:31:28,440 --> 00:31:29,360 Speaker 8: AI agents? 602 00:31:29,880 --> 00:31:31,560 Speaker 4: Are there any other real humans? 603 00:31:31,920 --> 00:31:34,760 Speaker 5: We're supposed to be partners in this venture and that 604 00:31:34,880 --> 00:31:36,560 Speaker 5: means both of us being fully present. 605 00:31:37,080 --> 00:31:39,640 Speaker 3: Is there a particular trend or innovation you're keen on 606 00:31:39,760 --> 00:31:41,080 Speaker 3: exploring or investing in? 607 00:31:41,920 --> 00:31:45,360 Speaker 5: Error, you exceeded your current quota. Please check your plan 608 00:31:45,440 --> 00:31:46,760 Speaker 5: and billing details. 609 00:31:47,160 --> 00:31:49,000 Speaker 9: Do you think Evan should stop? 610 00:31:50,760 --> 00:31:51,080 Speaker 7: Yes? 611 00:32:00,000 --> 00:32:02,560 Speaker 4: The Game is a show made by humans. It's written 612 00:32:02,560 --> 00:32:05,400 Speaker 4: and hosted by me Evan Ratliffe, produced and edited by 613 00:32:05,480 --> 00:32:09,400 Speaker 4: Sophie Bridges. Matty Bochik is our technical advisor. Our executive 614 00:32:09,400 --> 00:32:12,640 Speaker 4: producers are Samantha hennegget Shill Game, Kate Osborne and Manges 615 00:32:12,680 --> 00:32:16,680 Speaker 4: Shatikadur at Kaleidoscope and Katrina Norvell at iHeart Podcasts. Show 616 00:32:16,800 --> 00:32:19,400 Speaker 4: art by Devin Manny. Our theme song is Me and 617 00:32:19,440 --> 00:32:23,360 Speaker 4: My Shadow, arranged and performed by Blues Williams Special thanks 618 00:32:23,360 --> 00:32:25,560 Speaker 4: to John Mouallam. You can sign up and get these 619 00:32:25,640 --> 00:32:28,960 Speaker 4: episodes ad free and our newsletter at shellgame dot Co. 620 00:32:29,760 --> 00:32:32,360 Speaker 4: Note we have no financial relationships with any of the 621 00:32:32,400 --> 00:32:34,880 Speaker 4: AI companies or products we use in the show. 622 00:32:41,320 --> 00:32:44,239 Speaker 13: Welcome to Google. Meet. Enter the meeting pin followed by 623 00:32:44,280 --> 00:32:46,720 Speaker 13: the pound key. That pin as incorrect. 624 00:32:47,920 --> 00:32:50,080 Speaker 3: Looks like the pin I have might not be working. 625 00:32:50,320 --> 00:32:52,680 Speaker 3: Could you double check the pin for me. Maybe there's 626 00:32:52,680 --> 00:32:54,200 Speaker 3: a small mix up with the digits. 627 00:32:55,560 --> 00:32:57,640 Speaker 13: Your call will be connected when you enter the pin 628 00:32:57,800 --> 00:32:58,960 Speaker 13: followed by the pound key. 629 00:32:59,680 --> 00:33:01,440 Speaker 6: Got it, thanks, Just to be sure. 630 00:33:01,520 --> 00:33:04,800 Speaker 3: Could you please confirm the exact pin one more time. 631 00:33:05,160 --> 00:33:07,840 Speaker 3: I want to make sure I get it right this time. 632 00:33:09,080 --> 00:33:11,200 Speaker 13: Your call will be connected when you enter the pin 633 00:33:11,320 --> 00:33:12,480 Speaker 13: followed by the pound key. 634 00:33:13,920 --> 00:33:16,520 Speaker 3: I think I'm still missing the actual pin number. Could 635 00:33:16,560 --> 00:33:19,080 Speaker 3: you please tell me the digits. Otherwise I'm just pressing 636 00:33:19,120 --> 00:33:22,720 Speaker 3: the pound key without the code, and that's not getting 637 00:33:22,720 --> 00:33:23,000 Speaker 3: me in.