1 00:00:00,240 --> 00:00:02,439 [Miles] And the worst part is you don't even get a warning. 2 00:00:02,759 --> 00:00:05,639 [Miles] You wake up, check your phone, and there's a launch post that 3 00:00:05,679 --> 00:00:07,299 [Miles] looks a lot like your roadmap. 4 00:00:07,900 --> 00:00:08,160 [Grant] Yeah. 5 00:00:08,679 --> 00:00:09,439 [Miles] That's the feeling. 6 00:00:10,039 --> 00:00:11,699 [Grant] I've seen it from the investor side. 7 00:00:12,099 --> 00:00:15,259 [Grant] A founder goes quiet for a day, then the message arrives. 8 00:00:15,279 --> 00:00:16,660 [Grant] "Have you seen this?" 9 00:00:17,339 --> 00:00:18,780 [Miles] Always with the question mark. 10 00:00:19,440 --> 00:00:19,980 [Grant] Always. 11 00:00:20,679 --> 00:00:21,579 [Grant] Welcome to Year One. 12 00:00:22,239 --> 00:00:22,679 [Grant] I'm Grant. 13 00:00:23,140 --> 00:00:23,839 [Miles] And I'm Miles. 14 00:00:24,320 --> 00:00:27,000 [Miles] So the founder question isn't whether the labs are scary. 15 00:00:27,699 --> 00:00:28,620 [Miles] Obviously, they are. 16 00:00:29,079 --> 00:00:32,560 [Grant] The question is which parts of an agent company are still standing 17 00:00:32,658 --> 00:00:33,539 [Grant] after the next release? 18 00:00:33,960 --> 00:00:36,479 [Grant] Which pieces get absorbed and which ones don't? 19 00:00:37,039 --> 00:00:37,340 [Miles] Hmm. 20 00:00:37,920 --> 00:00:40,299 [Miles] I have a hunch, but I wanna hear the shipping list first. 21 00:00:40,920 --> 00:00:41,140 [Grant] Fair. 22 00:00:41,880 --> 00:00:43,979 [Grant] You can't judge the damage without seeing the crater. 23 00:00:44,659 --> 00:00:44,899 [Miles] Right. 24 00:00:45,479 --> 00:00:47,460 [Miles] First, what actually shipped this week? 25 00:00:48,600 --> 00:00:48,939 [Miles] Okay. 26 00:00:49,140 --> 00:00:54,899 [Miles] Monday, November ninth, Anthropic unveiled two new models, Claude Fable 5.2 and 27 00:00:55,000 --> 00:00:56,740 [Miles] Claude Mythos 5.2. 28 00:00:57,200 --> 00:00:58,659 [Grant] Two at once, of course. 29 00:00:59,159 --> 00:01:04,400 [Miles] Reuters described it as Anthropic strengthening its position against OpenAI ahead of 30 00:01:04,439 --> 00:01:05,340 [Miles] its planned IPO. 31 00:01:05,900 --> 00:01:07,879 [Miles] That framing is all over the coverage. 32 00:01:08,379 --> 00:01:10,040 [Grant] And CNBC put a number on it, right? 33 00:01:10,480 --> 00:01:10,840 [Miles] It did. 34 00:01:11,099 --> 00:01:15,519 [Miles] CNBC reported outputs are about thirty percent faster than the predecessors. 35 00:01:16,060 --> 00:01:16,540 [Grant] Faster. 36 00:01:16,840 --> 00:01:17,760 [Grant] Is that the whole story? 37 00:01:18,480 --> 00:01:18,739 [Miles] No. 38 00:01:19,120 --> 00:01:22,980 [Miles] Same report says these are the first frontier models since Amodei called 39 00:01:23,040 --> 00:01:24,560 [Miles] for a slowdown in September. 40 00:01:25,140 --> 00:01:28,099 [Grant] Huh, so the slowdown talk and then a launch. 41 00:01:28,519 --> 00:01:30,480 [Miles] Mm-hmm, I'm only reporting the sequence. 42 00:01:31,000 --> 00:01:31,219 [Grant] Fair. 43 00:01:31,680 --> 00:01:33,879 [Grant] And there's a proactive agent in the mix too, yeah? 44 00:01:34,400 --> 00:01:35,159 [Miles] Careful here. 45 00:01:35,400 --> 00:01:39,099 [Miles] The MacRumors page on the Fable launch lists a related headline 46 00:01:39,159 --> 00:01:43,120 [Miles] about Anthropic launching a proactive agent called Claude Duet. 47 00:01:43,540 --> 00:01:45,159 [Miles] That's the name, and I'm stopping there. 48 00:01:46,299 --> 00:01:47,579 [Grant] You're stopping there on purpose. 49 00:01:48,040 --> 00:01:50,879 [Miles] Because the page gives me a headline and nothing else. 50 00:01:51,159 --> 00:01:53,340 [Miles] If I told you what it does, I'd be making it up. 51 00:01:53,799 --> 00:01:54,500 [Grant] I respect that. 52 00:01:54,780 --> 00:01:55,560 [Grant] Founders would not. 53 00:01:55,780 --> 00:01:57,079 [Grant] They'd be already guessing at it. 54 00:01:57,420 --> 00:01:59,180 [Miles] Anthropic absolutely would. 55 00:01:59,719 --> 00:02:02,920 [Grant] Okay, the other side, OpenAI and the Dots. 56 00:02:03,200 --> 00:02:05,980 [Grant] I keep seeing people talk about them like they landed this week. 57 00:02:06,379 --> 00:02:07,000 [Miles] They didn't. 58 00:02:07,319 --> 00:02:11,139 [Miles] OpenAI's own announcement describes Dots as always-on agents. 59 00:02:11,520 --> 00:02:14,780 [Miles] Each has its own cloud computer, and they work towards goals around 60 00:02:14,800 --> 00:02:15,259 [Miles] the clock. 61 00:02:15,740 --> 00:02:16,560 [Grant] Twenty-four/seven. 62 00:02:17,060 --> 00:02:17,280 [Miles] Right. 63 00:02:17,539 --> 00:02:21,379 [Miles] But that was announced September twenty-ninth at DevDay, so the sense of 64 00:02:21,419 --> 00:02:24,520 [Miles] dots piling on this week is the news cycle, not a fresh 65 00:02:24,580 --> 00:02:24,960 [Miles] launch. 66 00:02:25,360 --> 00:02:27,400 [Grant] Wait, so any extension this week- 67 00:02:27,860 --> 00:02:29,439 [Miles] Needs its own confirmation. 68 00:02:29,759 --> 00:02:31,560 [Miles] Nothing I have says what changed. 69 00:02:32,080 --> 00:02:34,099 [Grant] Let me play the founder who sails an agent. 70 00:02:34,560 --> 00:02:36,900 [Grant] Which of these three things actually touches me? 71 00:02:37,400 --> 00:02:40,860 [Miles] Honestly, none of them directly until you know what you're selling. 72 00:02:41,340 --> 00:02:44,960 [Miles] A faster model is faster for everybody, including your competitors. 73 00:02:45,539 --> 00:02:45,939 [Grant] Right. 74 00:02:46,099 --> 00:02:48,680 [Grant] Speed lands for you and for them at the same time, so 75 00:02:48,719 --> 00:02:49,680 [Grant] it can't be your edge. 76 00:02:50,139 --> 00:02:50,719 [Miles] Exactly. 77 00:02:50,960 --> 00:02:54,120 [Miles] The only thing that stays yours is what you built around the 78 00:02:54,159 --> 00:02:54,460 [Miles] model. 79 00:02:54,879 --> 00:02:58,360 [Grant] And the proactive agent and the always-on dots matter less for what 80 00:02:58,379 --> 00:03:00,259 [Grant] they do than for what they signal. 81 00:03:00,780 --> 00:03:04,560 [Grant] Agents that keep working when nobody's watching are now a lab category 82 00:03:04,599 --> 00:03:04,819 [Grant] too. 83 00:03:05,280 --> 00:03:08,360 [Miles] Which is a sentence a lot of agent founders have probably said 84 00:03:08,419 --> 00:03:09,639 [Miles] out loud in their own pitch. 85 00:03:09,919 --> 00:03:11,419 [Miles] That's the real news for a seller. 86 00:03:11,960 --> 00:03:16,139 [Grant] Okay, so Monday's launch is faster models, dots are older than the 87 00:03:16,219 --> 00:03:18,060 [Grant] panic, and Duet is a name. 88 00:03:18,560 --> 00:03:19,259 [Miles] That's the shelf. 89 00:03:19,580 --> 00:03:20,360 [Miles] Now, the money. 90 00:03:21,479 --> 00:03:26,800 [Miles] Quartz, citing Bloomberg, reports Anthropic kicked off its IPO roadshow Tuesday, November 91 00:03:26,900 --> 00:03:27,199 [Miles] tenth. 92 00:03:27,400 --> 00:03:28,479 [Miles] That's basically today. 93 00:03:28,919 --> 00:03:31,800 [Grant] Today, while founders are still reading the release notes. 94 00:03:32,259 --> 00:03:36,000 [Miles] And the ask is up to sixty-four point four billion dollars. 95 00:03:36,419 --> 00:03:37,659 [Grant] Billion with a B. 96 00:03:38,439 --> 00:03:42,560 [Grant] Miles, I spent fifteen years around portfolios, and I still have to 97 00:03:42,599 --> 00:03:43,539 [Grant] say that number slowly. 98 00:03:43,879 --> 00:03:47,180 [Miles] Put it next to anything a startup raises, and it stops being 99 00:03:47,240 --> 00:03:49,159 [Miles] a number and starts being weather. 100 00:03:49,759 --> 00:03:50,080 [Grant] Weather. 101 00:03:50,439 --> 00:03:50,780 [Grant] Okay. 102 00:03:51,259 --> 00:03:51,919 [Grant] That I can picture. 103 00:03:52,379 --> 00:03:56,719 [Miles] And Quartz says the filing discloses about nine hundred billion dollars in 104 00:03:56,740 --> 00:03:58,599 [Miles] compute and infrastructure commitments. 105 00:03:59,039 --> 00:04:00,099 [Grant] Wait, nine hundred? 106 00:04:00,580 --> 00:04:01,500 [Grant] Higher than the raise. 107 00:04:02,000 --> 00:04:02,680 [Miles] Way higher. 108 00:04:02,900 --> 00:04:04,659 [Miles] So the raise isn't the whole bill. 109 00:04:04,860 --> 00:04:07,639 [Miles] It's a down payment on a very large promise. 110 00:04:08,280 --> 00:04:09,159 [Grant] I know that structure. 111 00:04:09,539 --> 00:04:12,099 [Grant] That's a mortgage on a house you haven't finished building. 112 00:04:13,520 --> 00:04:15,800 [Miles] Fair, but it tells a founder something. 113 00:04:16,199 --> 00:04:20,060 [Miles] A company carrying that much has to keep every product surface busy 114 00:04:20,180 --> 00:04:22,699 [Miles] because idle capacity is expensive kind. 115 00:04:23,379 --> 00:04:24,500 [Grant] I stop short of that, though. 116 00:04:24,939 --> 00:04:26,399 [Grant] The filing lists the commitments. 117 00:04:26,720 --> 00:04:28,600 [Grant] It doesn't tell you which products fill them. 118 00:04:29,139 --> 00:04:29,560 [Miles] You're right. 119 00:04:29,699 --> 00:04:32,600 [Miles] I'm inferring the pressure, not reading it off a page. 120 00:04:33,040 --> 00:04:35,240 [Grant] And the filing itself is a different animal. 121 00:04:35,680 --> 00:04:39,220 [Grant] CNN reported Anthropic filed back on October twenty-sixth. 122 00:04:39,600 --> 00:04:41,220 [Miles] Two weeks before the roadshow. 123 00:04:41,660 --> 00:04:41,899 [Grant] Yeah. 124 00:04:42,079 --> 00:04:45,779 [Grant] And CNN's line on it is that the filing details surging business 125 00:04:45,860 --> 00:04:49,420 [Grant] growth along with a vast multitude of warnings about how it could 126 00:04:49,459 --> 00:04:50,079 [Grant] all go wrong. 127 00:04:50,420 --> 00:04:51,779 [Miles] A vast multitude. 128 00:04:51,939 --> 00:04:53,000 [Miles] I love that phrasing. 129 00:04:53,199 --> 00:04:56,300 [Miles] It's a victory lap and a disclaimer stapled together. 130 00:04:56,699 --> 00:04:59,139 [Grant] Which is every prospectus ever written, to be fair. 131 00:04:59,600 --> 00:05:01,459 [Grant] But for a founder, it's useful. 132 00:05:01,800 --> 00:05:04,660 [Grant] The growth half says the lab's business is moving. 133 00:05:05,019 --> 00:05:08,060 [Grant] The warnings half says the lab knows it's exposed too. 134 00:05:08,500 --> 00:05:08,759 [Miles] Huh. 135 00:05:09,019 --> 00:05:10,560 [Miles] So they're not invincible. 136 00:05:10,660 --> 00:05:11,519 [Miles] They're just big. 137 00:05:12,019 --> 00:05:13,959 [Grant] Big and under a microscope. 138 00:05:14,319 --> 00:05:17,860 [Grant] Public investors will want to see products that earn their keep, and 139 00:05:17,920 --> 00:05:19,160 [Grant] that changes what ships. 140 00:05:19,579 --> 00:05:21,759 [Miles] Is that good or bad for a founder, though? 141 00:05:22,000 --> 00:05:23,319 [Miles] I keep going back and forth. 142 00:05:23,779 --> 00:05:24,079 [Grant] Both. 143 00:05:24,620 --> 00:05:28,240 [Grant] Bad because a lab under that scrutiny has every reason to show 144 00:05:28,279 --> 00:05:30,540 [Grant] a clean story of products people use. 145 00:05:31,060 --> 00:05:35,019 [Grant] Good, maybe, because my guess, and it's only a guess, is that 146 00:05:35,060 --> 00:05:38,000 [Grant] a company that size can't chase every small workflow. 147 00:05:38,240 --> 00:05:42,019 [Miles] So the capital scale tells you the lab will be around and 148 00:05:42,160 --> 00:05:42,560 [Miles] loud. 149 00:05:42,819 --> 00:05:45,279 [Miles] It doesn't tell you what it builds next. 150 00:05:46,040 --> 00:05:46,300 [Grant] Right. 151 00:05:46,620 --> 00:05:48,740 [Grant] Don't confuse size with direction. 152 00:05:49,220 --> 00:05:52,560 [Grant] If you're a founder, read the figures as a reason to plan 153 00:05:52,620 --> 00:05:56,939 [Grant] for a very well-funded neighbor, not as a forecast of your product's 154 00:05:56,980 --> 00:05:57,220 [Grant] fate. 155 00:05:57,860 --> 00:05:59,060 [Miles] A well-funded neighbor. 156 00:05:59,180 --> 00:06:00,259 [Miles] I can live with that framing. 157 00:06:00,935 --> 00:06:04,396 [Grant] And read the warnings half as a reminder that nobody in this 158 00:06:04,456 --> 00:06:07,536 [Grant] story, including the lab, has a guaranteed outcome. 159 00:06:08,055 --> 00:06:10,036 [Miles] Which sends us back to the product side. 160 00:06:10,355 --> 00:06:14,576 [Miles] So how does Anthropic actually steer people across its own models, and 161 00:06:14,636 --> 00:06:17,076 [Miles] what can a founder reasonably infer from that? 162 00:06:17,896 --> 00:06:21,055 [Grant] Right, because if there's a map on how they route users, that's 163 00:06:21,115 --> 00:06:23,315 [Grant] the closest thing to a roadmap you get. 164 00:06:24,415 --> 00:06:25,675 [Miles] Okay, the routing piece. 165 00:06:26,495 --> 00:06:30,755 [Miles] CNBC says Anthropic points people to Opus five point five for everyday 166 00:06:30,836 --> 00:06:31,276 [Miles] tasks. 167 00:06:31,696 --> 00:06:35,115 [Miles] Then Fable five point two is the one for hard engineering, math, 168 00:06:35,175 --> 00:06:36,255 [Miles] and robotics problems. 169 00:06:36,815 --> 00:06:39,435 [Grant] Huh, so the flagship isn't the default. 170 00:06:39,975 --> 00:06:40,315 [Miles] Right. 171 00:06:40,576 --> 00:06:43,555 [Miles] It's like a dealer telling you the daily driver is the sensible 172 00:06:43,656 --> 00:06:45,836 [Miles] one and the track car is for people with a track. 173 00:06:46,935 --> 00:06:49,036 [Grant] I'll allow it, but that's guidance, Miles. 174 00:06:49,456 --> 00:06:52,896 [Grant] Anything we say about which workloads the labs chase or ignore is 175 00:06:52,975 --> 00:06:54,036 [Grant] us reading tea leaves. 176 00:06:54,495 --> 00:06:54,735 [Grant] Fair. 177 00:06:55,196 --> 00:06:58,136 [Miles] Nobody at Anthropic said that, and CNBC didn't either. 178 00:06:58,696 --> 00:06:59,776 [Miles] It's our inference. 179 00:07:00,315 --> 00:07:01,456 [Grant] Then what's the inference? 180 00:07:01,935 --> 00:07:03,776 [Grant] Everyday tasks get one model. 181 00:07:03,995 --> 00:07:05,814 [Grant] The brutal stuff gets the expensive one. 182 00:07:06,276 --> 00:07:07,536 [Grant] Where does a founder sit in that? 183 00:07:08,216 --> 00:07:12,935 [Miles] Somewhere in between, probably, which is exactly the uncomfortable part. 184 00:07:13,336 --> 00:07:13,596 [Grant] Hmm. 185 00:07:14,276 --> 00:07:15,956 [Miles] Let me walk a made-up founder through it. 186 00:07:16,555 --> 00:07:19,935 [Miles] Their agent handles routine requests for customers, and most of what it 187 00:07:19,975 --> 00:07:21,216 [Miles] does is everyday work. 188 00:07:21,776 --> 00:07:25,076 [Grant] So the everyday model is the one the lab itself points people 189 00:07:25,136 --> 00:07:28,435 [Grant] to for that kind of work, which means the founder's core loop 190 00:07:28,495 --> 00:07:30,636 [Grant] sits right where the lab says, "Use this one." 191 00:07:31,295 --> 00:07:33,596 [Miles] Right, and that's the uncomfortable reading. 192 00:07:34,156 --> 00:07:34,636 [Miles] But flip it. 193 00:07:34,995 --> 00:07:38,216 [Miles] Say the founder's real value is in the rare messy cases where 194 00:07:38,235 --> 00:07:41,516 [Miles] the customer's rules decide the answer, not raw model strength. 195 00:07:42,036 --> 00:07:43,956 [Grant] Then the routing split barely touches them. 196 00:07:44,136 --> 00:07:46,696 [Grant] The guidance is about which model suits which problem. 197 00:07:47,055 --> 00:07:49,956 [Grant] It says nothing about whose process the problem lives in. 198 00:07:50,615 --> 00:07:52,175 [Miles] That's the distinction I'd hold onto. 199 00:07:52,776 --> 00:07:55,815 [Miles] Routing tells you how the lab sorts problems by difficulty. 200 00:07:56,375 --> 00:07:59,315 [Miles] A founder should sort theirs by where the knowledge lives. 201 00:07:59,795 --> 00:08:02,656 [Grant] Mm-hmm, and notice I'm not saying that makes them safe. 202 00:08:03,055 --> 00:08:06,136 [Grant] I'm saying the split gives them a question to ask, not an 203 00:08:06,216 --> 00:08:06,495 [Grant] answer. 204 00:08:07,175 --> 00:08:08,096 [Miles] Now the Dots side. 205 00:08:08,735 --> 00:08:12,055 [Miles] DataCamp ran an explainer on Dots going off that September twenty-ninth 206 00:08:12,096 --> 00:08:12,596 [Miles] announcement. 207 00:08:13,096 --> 00:08:16,216 [Miles] It says Dots are a managed alternative to agents built on the 208 00:08:16,255 --> 00:08:16,735 [Miles] API. 209 00:08:17,255 --> 00:08:19,156 [Grant] Alternative, not a replacement. 210 00:08:19,696 --> 00:08:20,896 [Miles] That's their wording, yes. 211 00:08:21,495 --> 00:08:24,795 [Miles] And it says OpenAI hasn't announced a public API for Dots. 212 00:08:25,276 --> 00:08:28,115 [Grant] Okay, but that's one explainer, secondary analysis. 213 00:08:28,495 --> 00:08:30,596 [Grant] I wouldn't bet a seed round on one blog's read. 214 00:08:31,216 --> 00:08:31,555 [Miles] Agreed. 215 00:08:31,716 --> 00:08:32,456 [Miles] Hold it loosely. 216 00:08:32,755 --> 00:08:33,596 [Grant] And here's what bugs me. 217 00:08:33,875 --> 00:08:37,875 [Grant] I can already hear a founder hearing that and exhaling, "No API, 218 00:08:38,035 --> 00:08:40,196 [Grant] not a replacement, so the labs won't come here." 219 00:08:40,716 --> 00:08:41,895 [Miles] Wait, so is that safety? 220 00:08:42,336 --> 00:08:43,336 [Grant] That's my question to you. 221 00:08:43,576 --> 00:08:45,015 [Grant] Should anyone read it as safety? 222 00:08:45,495 --> 00:08:46,655 [Miles] No, not from this. 223 00:08:46,956 --> 00:08:49,816 [Miles] A managed product today says nothing about next quarter. 224 00:08:50,735 --> 00:08:53,936 [Miles] Grant, when a founder says the labs won't come here, what are 225 00:08:53,975 --> 00:08:54,615 [Miles] they avoiding? 226 00:08:55,635 --> 00:08:56,535 [Grant] Usually the date. 227 00:08:56,895 --> 00:08:59,316 [Grant] They're saying never when they mean not yet. 228 00:08:59,836 --> 00:09:02,196 [Grant] I've watched investors do it with positions they love. 229 00:09:02,456 --> 00:09:04,895 [Miles] So the comfort is a guess dressed as a fact. 230 00:09:05,375 --> 00:09:05,755 [Grant] Mostly. 231 00:09:06,255 --> 00:09:09,076 [Grant] A route split and a missing API tell you what's shipped. 232 00:09:09,416 --> 00:09:10,755 [Grant] They don't tell you what's coming. 233 00:09:10,975 --> 00:09:13,255 [Miles] So what would you actually watch for next? 234 00:09:13,635 --> 00:09:15,135 [Miles] Not predictions, signals. 235 00:09:15,436 --> 00:09:19,076 [Grant] Three things, and all of them are observable: whether the routing guidance 236 00:09:19,155 --> 00:09:23,995 [Grant] changes, whether OpenAI announces a public API for Dots, or whether either 237 00:09:24,076 --> 00:09:28,796 [Grant] lab starts talking about customer-specific setup rather than general capability. 238 00:09:29,135 --> 00:09:30,716 [Miles] Each of those would move the picture. 239 00:09:31,056 --> 00:09:34,336 [Miles] A public API, for one, might make Dots something a founder could 240 00:09:34,375 --> 00:09:36,155 [Miles] build on rather than compete with. 241 00:09:36,475 --> 00:09:38,836 [Grant] And if none of them move for a while, that's still not 242 00:09:38,916 --> 00:09:39,296 [Grant] safety. 243 00:09:39,475 --> 00:09:40,436 [Grant] It's just quiet. 244 00:09:40,676 --> 00:09:42,855 [Miles] Which means we need a better test than vibes. 245 00:09:43,235 --> 00:09:45,655 [Grant] Time to sort the pile into features and workflows. 246 00:09:46,755 --> 00:09:47,995 [Miles] Okay, sorting time. 247 00:09:48,515 --> 00:09:50,796 [Miles] Grant, you ran money for fifteen years. 248 00:09:51,275 --> 00:09:54,115 [Miles] Where do you draw the line between a feature and a workflow? 249 00:09:54,755 --> 00:09:57,355 [Grant] My rule, and it's mine, nobody handed it to me. 250 00:09:57,816 --> 00:10:01,056 [Grant] A feature is something a lab can ship into its own product 251 00:10:01,076 --> 00:10:02,115 [Grant] for everyone at once. 252 00:10:02,696 --> 00:10:06,135 [Grant] A workflow only works because it knows one customer's mess. 253 00:10:06,816 --> 00:10:07,495 [Miles] Give me the mess. 254 00:10:08,176 --> 00:10:09,995 [Grant] Say your agent drafts a summary. 255 00:10:10,556 --> 00:10:11,936 [Grant] Any lab can do that tomorrow. 256 00:10:12,336 --> 00:10:16,495 [Grant] Now say it reads a client's approval chain, knows who signs off 257 00:10:16,576 --> 00:10:19,015 [Grant] when, and files it where their auditors look. 258 00:10:19,535 --> 00:10:24,196 [Grant] That's not in anyone's roadmap because it lives inside one company's habits. 259 00:10:24,775 --> 00:10:25,115 [Miles] Hmm. 260 00:10:25,456 --> 00:10:26,615 [Miles] Cars have the same split. 261 00:10:27,176 --> 00:10:29,995 [Miles] A factory can add a bigger turbo to every model overnight. 262 00:10:30,556 --> 00:10:30,796 [Grant] Sure. 263 00:10:31,296 --> 00:10:35,255 [Miles] But it can't rebuild a collector's whole restoration process around the one 264 00:10:35,355 --> 00:10:36,296 [Miles] chassis he owns. 265 00:10:36,735 --> 00:10:38,556 [Miles] That takes someone standing in his garage. 266 00:10:39,336 --> 00:10:39,635 [Grant] Fair. 267 00:10:39,916 --> 00:10:41,956 [Grant] The difference is who has to be in the room. 268 00:10:42,576 --> 00:10:42,895 [Miles] Right. 269 00:10:43,196 --> 00:10:44,735 [Miles] So features get absorbed. 270 00:10:44,995 --> 00:10:46,255 [Miles] Workflows get what? 271 00:10:46,556 --> 00:10:47,076 [Miles] Tolerated? 272 00:10:47,836 --> 00:10:51,235 [Grant] Protected, if you're lucky, and I'd say that's opinion. 273 00:10:51,436 --> 00:10:54,296 [Grant] I can't promise the labs agree on where the line sits. 274 00:10:55,056 --> 00:10:58,615 [Miles] Fair, and nobody here is claiming a YC partner signed off on 275 00:10:58,655 --> 00:10:58,796 [Miles] it. 276 00:10:59,495 --> 00:10:59,696 [Grant] No. 277 00:10:59,875 --> 00:11:02,696 [Grant] If a partner weighs in on this split, that belongs in its 278 00:11:02,775 --> 00:11:05,135 [Grant] own slot, in their voice, not ours. 279 00:11:05,836 --> 00:11:08,676 [Miles] Give me a second example, because the summary one is easy. 280 00:11:08,855 --> 00:11:10,035 [Miles] Something less obvious. 281 00:11:10,676 --> 00:11:10,936 [Grant] Okay. 282 00:11:11,395 --> 00:11:13,395 [Grant] Say your agent triages request. 283 00:11:13,796 --> 00:11:16,916 [Grant] The generic version, sort by urgency, is a feature. 284 00:11:17,375 --> 00:11:21,816 [Grant] The workflow version knows that at this one customer, a request from 285 00:11:21,855 --> 00:11:25,775 [Grant] a particular team jumps the queue no matter what the wording says, 286 00:11:26,316 --> 00:11:28,456 [Grant] and that nobody ever wrote that rule. 287 00:11:29,135 --> 00:11:30,235 [Miles] The unwritten rule. 288 00:11:30,456 --> 00:11:31,916 [Miles] That's the thing a model can't guess. 289 00:11:32,615 --> 00:11:35,875 [Grant] Right, and it's hard for the customer to hand over too, which 290 00:11:35,916 --> 00:11:39,615 [Grant] is why it's sticky, but it only exists if you've been inside 291 00:11:39,676 --> 00:11:40,615 [Grant] long enough to learn it. 292 00:11:41,255 --> 00:11:41,576 [Miles] Right. 293 00:11:41,875 --> 00:11:45,576 [Miles] Now, the footing underneath a recent batch founder matters here because their 294 00:11:45,596 --> 00:11:46,676 [Miles] position is thin. 295 00:11:47,316 --> 00:11:47,735 [Grant] How thin? 296 00:11:48,336 --> 00:11:52,475 [Miles] Y Combinator's own site says it funds companies with no revenue, no 297 00:11:52,596 --> 00:11:54,735 [Miles] product, or even a fully baked idea. 298 00:11:55,375 --> 00:11:55,855 [Grant] Wait, really? 299 00:11:56,255 --> 00:11:56,836 [Grant] No product? 300 00:11:57,436 --> 00:11:58,556 [Miles] That's the stated bar. 301 00:11:58,796 --> 00:12:00,035 [Miles] Never too early to apply. 302 00:12:00,635 --> 00:12:01,196 [Grant] Hmm. 303 00:12:01,395 --> 00:12:03,375 [Grant] So think about what that means this week. 304 00:12:03,875 --> 00:12:06,895 [Grant] A lab ships something, and a founder doesn't have a pile of 305 00:12:06,956 --> 00:12:08,696 [Grant] customer workflows to point at yet. 306 00:12:09,279 --> 00:12:10,039 [Grant] They have a pitch. 307 00:12:10,399 --> 00:12:12,200 [Miles] And a pitch reads as a feature. 308 00:12:12,379 --> 00:12:13,320 [Miles] That's the scary part. 309 00:12:13,600 --> 00:12:14,179 [Grant] Exactly. 310 00:12:14,500 --> 00:12:17,220 [Grant] Early, you look like a feature because you haven't had time to 311 00:12:17,259 --> 00:12:18,179 [Grant] become a workflow. 312 00:12:18,659 --> 00:12:21,860 [Miles] So is there a way out of that, or is it just 313 00:12:21,899 --> 00:12:22,679 [Miles] a timing problem? 314 00:12:23,100 --> 00:12:25,559 [Grant] It's partly timing, but it's also a choice. 315 00:12:25,899 --> 00:12:28,820 [Grant] A founder can pick the customer whose mess is deepest and go 316 00:12:28,860 --> 00:12:32,759 [Grant] there on purpose instead of spreading thin across everyone who says, "Interesting." 317 00:12:33,279 --> 00:12:35,399 [Miles] Go narrow before the lab goes wide. 318 00:12:35,820 --> 00:12:36,620 [Grant] That's how I'd put it. 319 00:12:37,019 --> 00:12:39,860 [Grant] And be honest with yourself about which side of the line each 320 00:12:39,919 --> 00:12:41,558 [Grant] customer is actually paying for. 321 00:12:41,960 --> 00:12:44,139 [Miles] And there are a lot of people in that spot. 322 00:12:44,600 --> 00:12:47,840 [Miles] Brian Sykes wrote about attending his first Demo Day this week, and 323 00:12:47,899 --> 00:12:50,240 [Miles] he counted two hundred and fifty companies. 324 00:12:50,600 --> 00:12:51,840 [Grant] Two hundred and fifty? 325 00:12:52,159 --> 00:12:54,539 [Miles] With five hundred and eight founders presenting. 326 00:12:54,899 --> 00:12:56,580 [Miles] Summer twenty twenty-six batch. 327 00:12:56,879 --> 00:13:00,720 [Grant] That's five hundred people all facing the same Monday morning question. 328 00:13:01,059 --> 00:13:02,080 [Miles] At the same time. 329 00:13:02,379 --> 00:13:05,500 [Grant] Which is funny because I'd bet most of them pitched the demo, 330 00:13:05,879 --> 00:13:06,820 [Grant] not the integration. 331 00:13:07,340 --> 00:13:08,559 [Grant] Demos are easy to love. 332 00:13:08,879 --> 00:13:10,360 [Grant] Integrations take months. 333 00:13:10,759 --> 00:13:13,059 [Miles] That's my guess, too, though Sykes didn't say that. 334 00:13:13,299 --> 00:13:15,139 [Miles] I'm only guessing from how these things go. 335 00:13:15,519 --> 00:13:16,600 [Grant] Yeah, keep it labeled. 336 00:13:17,080 --> 00:13:18,740 [Grant] What he gave us is the headcount. 337 00:13:19,019 --> 00:13:19,860 [Miles] Which is plenty. 338 00:13:20,419 --> 00:13:23,919 [Miles] Five hundred founders, one lab launch cycle, and the ones who live 339 00:13:24,019 --> 00:13:26,860 [Miles] are the ones whose product knows a customer's mess better than a 340 00:13:26,919 --> 00:13:27,500 [Miles] model does. 341 00:13:27,899 --> 00:13:31,480 [Grant] And to be clear on the other side, a workflow isn't automatically 342 00:13:31,539 --> 00:13:31,899 [Grant] a moat. 343 00:13:32,399 --> 00:13:36,039 [Grant] If the customer's process changes or the customer decides to build it 344 00:13:36,059 --> 00:13:37,519 [Grant] themselves, it's gone. 345 00:13:37,980 --> 00:13:38,159 [Miles] Right. 346 00:13:38,399 --> 00:13:40,159 [Miles] Protected doesn't mean permanent. 347 00:13:40,220 --> 00:13:43,080 [Miles] It means the lab's next release isn't what puts it at risk 348 00:13:43,159 --> 00:13:43,539 [Miles] first. 349 00:13:43,980 --> 00:13:47,179 [Grant] So a founder should hold the line as a working assumption, not 350 00:13:47,220 --> 00:13:47,720 [Grant] a premise. 351 00:13:48,139 --> 00:13:49,279 [Miles] Working assumption. 352 00:13:49,440 --> 00:13:50,460 [Miles] I like that better than moat. 353 00:13:50,759 --> 00:13:51,059 [Grant] Okay. 354 00:13:51,440 --> 00:13:53,919 [Grant] But that's a lot of theory for a founder with a launch 355 00:13:54,039 --> 00:13:55,480 [Grant] landing on their roadmap tonight. 356 00:13:56,360 --> 00:13:56,620 [Miles] Yeah. 357 00:13:57,059 --> 00:13:59,820 [Grant] So what does a founder actually do on Monday morning? 358 00:14:00,919 --> 00:14:04,240 [Miles] Okay, Grant, take the sorting idea and make it something a founder 359 00:14:04,340 --> 00:14:05,720 [Miles] can run in an afternoon. 360 00:14:06,039 --> 00:14:07,480 [Miles] Three questions on paper. 361 00:14:07,899 --> 00:14:08,159 [Grant] Fine. 362 00:14:08,559 --> 00:14:11,879 [Grant] One, if you swapped in the next model tomorrow, would your product 363 00:14:11,899 --> 00:14:12,980 [Grant] get better by itself? 364 00:14:13,379 --> 00:14:16,259 [Grant] If yes, that part lives in the model, and the lab can 365 00:14:16,320 --> 00:14:16,659 [Grant] ship it. 366 00:14:16,980 --> 00:14:17,278 [Miles] Ouch. 367 00:14:17,558 --> 00:14:18,519 [Miles] That's most of the demo. 368 00:14:19,019 --> 00:14:19,619 [Grant] Usually, yeah. 369 00:14:20,039 --> 00:14:22,480 [Grant] Two, where does your product keep its state? 370 00:14:22,740 --> 00:14:25,720 [Grant] Whose data does it touch, and whose approvals does it wait on? 371 00:14:25,940 --> 00:14:27,019 [Miles] That's the workflow side. 372 00:14:27,179 --> 00:14:29,580 [Miles] The customer's process, the customer's permissions. 373 00:14:29,940 --> 00:14:30,120 [Grant] Right. 374 00:14:30,480 --> 00:14:34,100 [Grant] Three, if a customer left, what would they have to rebuild by 375 00:14:34,159 --> 00:14:34,500 [Grant] hand? 376 00:14:34,820 --> 00:14:37,639 [Grant] If the answer is nothing, uh, that's your warning light. 377 00:14:37,919 --> 00:14:38,820 [Miles] That one stings. 378 00:14:38,980 --> 00:14:41,480 [Miles] It's the question that tells you whether you're a product or a 379 00:14:41,539 --> 00:14:41,840 [Miles] demo. 380 00:14:42,240 --> 00:14:42,840 [Grant] It's meant to. 381 00:14:43,159 --> 00:14:44,980 [Miles] And to be clear, this is us talking. 382 00:14:45,240 --> 00:14:46,799 [Miles] Our checklist, our opinion. 383 00:14:47,120 --> 00:14:47,460 [Grant] Agreed. 384 00:14:47,659 --> 00:14:48,840 [Grant] Nobody's name goes on it. 385 00:14:49,220 --> 00:14:50,480 [Miles] One more thing on the checklist. 386 00:14:50,740 --> 00:14:53,860 [Miles] Score each part of the product, not the product as a whole. 387 00:14:54,259 --> 00:14:55,460 [Grant] Most products are a mix. 388 00:14:55,840 --> 00:14:56,080 [Grant] Yes. 389 00:14:56,480 --> 00:15:00,740 [Grant] You'll find the summarizing layer fails question one, and the approval handling 390 00:15:00,940 --> 00:15:02,019 [Grant] passes question two. 391 00:15:02,379 --> 00:15:05,019 [Grant] That's useful because it tells you where to put the next stretch 392 00:15:05,059 --> 00:15:05,500 [Grant] of work. 393 00:15:05,879 --> 00:15:07,580 [Miles] So who does a founder run it past? 394 00:15:07,679 --> 00:15:10,860 [Miles] Because staring at your own spreadsheet at midnight is a bad review 395 00:15:10,879 --> 00:15:11,360 [Miles] process. 396 00:15:11,779 --> 00:15:15,960 [Grant] Well, YC's own About page says it runs office hours year-round, and 397 00:15:16,080 --> 00:15:19,340 [Grant] startups from all previous cycles can book time whenever they want. 398 00:15:19,659 --> 00:15:21,279 [Miles] Wait, all previous cycles? 399 00:15:21,460 --> 00:15:24,179 [Miles] So a founder from two batches ago still gets a slot? 400 00:15:24,539 --> 00:15:25,480 [Grant] That's what the page says. 401 00:15:25,779 --> 00:15:29,480 [Grant] And it describes Bookface as the platform founders use to connect with 402 00:15:29,519 --> 00:15:32,299 [Grant] each other, where you can post a question or a poll for 403 00:15:32,340 --> 00:15:32,720 [Grant] knowledge. 404 00:15:33,139 --> 00:15:36,200 [Miles] So you could literally post, "My lab just shipped my feature. 405 00:15:36,320 --> 00:15:37,019 [Miles] Anyone been here?" 406 00:15:37,399 --> 00:15:38,360 [Grant] I'll bet someone has. 407 00:15:39,360 --> 00:15:42,240 [Grant] Even if you're not in YC, the move is the same. 408 00:15:42,559 --> 00:15:45,679 [Grant] Find the founders who've lived through a bad launch week and ask 409 00:15:45,700 --> 00:15:46,799 [Grant] them what they cut first. 410 00:15:47,059 --> 00:15:49,919 [Miles] And you ask before Monday, not after the press release. 411 00:15:50,279 --> 00:15:50,720 [Grant] Uh, yes. 412 00:15:51,139 --> 00:15:53,259 [Grant] Spreadsheet in the morning, calls in the afternoon. 413 00:15:53,580 --> 00:15:55,759 [Miles] And one more for the founders who get a pit in their 414 00:15:55,799 --> 00:15:56,720 [Miles] stomach reading this. 415 00:15:57,059 --> 00:15:58,559 [Grant] The pit is useful. 416 00:15:58,919 --> 00:16:02,059 [Grant] It usually means you found the part you were hoping nobody would 417 00:16:02,139 --> 00:16:02,480 [Grant] test. 418 00:16:02,879 --> 00:16:04,419 [Grant] Better to find it on your own schedule. 419 00:16:04,720 --> 00:16:06,639 [Miles] Okay, last thoughts, then we leave it open. 420 00:16:06,919 --> 00:16:10,500 [Miles] Mine is run the three questions on your riskiest feature first, not 421 00:16:10,539 --> 00:16:11,200 [Miles] your favorite one. 422 00:16:11,539 --> 00:16:14,659 [Grant] Mine is to write down what you'd rebuild if the lab shipped 423 00:16:14,679 --> 00:16:15,279 [Grant] it tonight. 424 00:16:15,700 --> 00:16:17,179 [Grant] Then we'll see how the week plays out. 425 00:16:18,159 --> 00:16:19,000 [Miles] One test then. 426 00:16:19,559 --> 00:16:23,259 [Miles] Take your product and ask whether a customer could keep using it 427 00:16:23,320 --> 00:16:27,320 [Miles] if the next model arrived tomorrow and did everything for free. 428 00:16:27,799 --> 00:16:28,059 [Grant] Mm. 429 00:16:28,359 --> 00:16:32,039 [Grant] And if the honest answer is a long pause, that's your answer. 430 00:16:32,519 --> 00:16:32,759 [Miles] Right. 431 00:16:32,980 --> 00:16:37,159 [Miles] And what that next launch actually does to anyone's roadmap, we don't 432 00:16:37,200 --> 00:16:37,600 [Miles] know yet. 433 00:16:38,480 --> 00:16:39,220 [Grant] Nobody does. 434 00:16:39,700 --> 00:16:42,960 [Grant] Some founder is reading release notes right now with a coffee going 435 00:16:43,039 --> 00:16:43,419 [Grant] cold. 436 00:16:43,879 --> 00:16:44,139 [Miles] Okay. 437 00:16:44,379 --> 00:16:45,080 [Miles] The housekeeping. 438 00:16:45,419 --> 00:16:50,740 [Miles] If you're a first-year YC founder, send us your story at yearone@heymato.com. 439 00:16:51,120 --> 00:16:53,080 [Grant] And if this was useful, leave a review. 440 00:16:53,340 --> 00:16:54,820 [Grant] It helps other founders find us. 441 00:16:55,360 --> 00:16:57,740 [Miles] Grant, I'll be watching the changelog tonight. 442 00:16:58,220 --> 00:16:58,899 [Grant] Me too, Miles. 443 00:16:59,320 --> 00:16:59,639 [Grant] Me too.