[
  {
    "startMs": 1309,
    "text": "[music]"
  },
  {
    "startMs": 12720,
    "text": ">> Joining us on stage is the co-founder of"
  },
  {
    "startMs": 15120,
    "text": "Instagram [music]"
  },
  {
    "startMs": 16280,
    "text": "and a member of technical staff at"
  },
  {
    "startMs": 18160,
    "text": "Anthropic."
  },
  {
    "startMs": 19760,
    "text": "Mike Krieger."
  },
  {
    "startMs": 36360,
    "text": ">> How's everybody doing? I mean, good"
  },
  {
    "startMs": 37640,
    "text": "morning."
  },
  {
    "startMs": 39160,
    "text": "Nice."
  },
  {
    "startMs": 40520,
    "text": "Um"
  },
  {
    "startMs": 41640,
    "text": ">> Mike, thank you for releasing Fable just"
  },
  {
    "startMs": 43440,
    "text": "in time for us."
  },
  {
    "startMs": 44520,
    "text": ">> Exactly for the conference. We timed it."
  },
  {
    "startMs": 46505,
    "text": ">> [laughter]"
  },
  {
    "startMs": 47280,
    "text": ">> Um we're we're so glad to have you. Uh"
  },
  {
    "startMs": 49920,
    "text": "you're uh one of the preeminent builders"
  },
  {
    "startMs": 52800,
    "text": "and you're a leading labs at Anthropic."
  },
  {
    "startMs": 55600,
    "text": "Um how has your"
  },
  {
    "startMs": 57480,
    "text": "model usage changed as as you've, you"
  },
  {
    "startMs": 59880,
    "text": "know, seen models internally grow?"
  },
  {
    "startMs": 62280,
    "text": ">> Yeah, I mean, for me it's been like both"
  },
  {
    "startMs": 63840,
    "text": "the model shift and then my role shift."
  },
  {
    "startMs": 65800,
    "text": "So, I for like the first 2 years I was"
  },
  {
    "startMs": 67520,
    "text": "at Anthropic, I was chief product"
  },
  {
    "startMs": 68640,
    "text": "officer. And then I kept seeing people"
  },
  {
    "startMs": 70520,
    "text": "build with the models and the FOMO just"
  },
  {
    "startMs": 72760,
    "text": "kept increasing because I was you know,"
  },
  {
    "startMs": 75320,
    "text": "use the models as much as possible. But"
  },
  {
    "startMs": 76920,
    "text": "for example, on product strategy I would"
  },
  {
    "startMs": 78600,
    "text": "write a strategy doc and then have"
  },
  {
    "startMs": 80360,
    "text": "Claude critique it and maybe you can use"
  },
  {
    "startMs": 82040,
    "text": "a workflow, but it's not quite the same"
  },
  {
    "startMs": 84280,
    "text": "as like building in that pure way. And I"
  },
  {
    "startMs": 86840,
    "text": "was like spending all my weekends trying"
  },
  {
    "startMs": 88080,
    "text": "to build with it and I realized, \"Okay,"
  },
  {
    "startMs": 89440,
    "text": "I actually just need to shift. It's like"
  },
  {
    "startMs": 91280,
    "text": "way too interesting a time.\" And it's"
  },
  {
    "startMs": 92400,
    "text": "actually an interesting trend I've seen"
  },
  {
    "startMs": 93800,
    "text": "now like several people that were CTOs"
  },
  {
    "startMs": 96760,
    "text": "at other places are like now joining as"
  },
  {
    "startMs": 99080,
    "text": "ICs at Anthropic and other places. But I"
  },
  {
    "startMs": 101520,
    "text": "made a role shift and it was actually"
  },
  {
    "startMs": 103320,
    "text": "right around the time where we started"
  },
  {
    "startMs": 104760,
    "text": "getting sort of internal snapshots of"
  },
  {
    "startMs": 107440,
    "text": "what became Mythos and Fable. And what"
  },
  {
    "startMs": 110280,
    "text": "was really interesting watching that"
  },
  {
    "startMs": 112360,
    "text": "sort of shift was um that"
  },
  {
    "startMs": 115360,
    "text": "kind of change between I have an idea,"
  },
  {
    "startMs": 117800,
    "text": "I'm going to like sort of break it down"
  },
  {
    "startMs": 119200,
    "text": "in my head much more how I would do"
  },
  {
    "startMs": 120880,
    "text": "engineering normally, and then kind of"
  },
  {
    "startMs": 122960,
    "text": "iterate through these different steps to"
  },
  {
    "startMs": 124600,
    "text": "moving to much more of the paradigm of"
  },
  {
    "startMs": 126520,
    "text": "I'm going to describe the goal, like go"
  },
  {
    "startMs": 128560,
    "text": "off and work on it, and then like we can"
  },
  {
    "startMs": 130160,
    "text": "talk about what trade-offs you you know,"
  },
  {
    "startMs": 131560,
    "text": "surface some questions along the way,"
  },
  {
    "startMs": 132680,
    "text": "but then"
  },
  {
    "startMs": 133720,
    "text": "figure out what where you landed and"
  },
  {
    "startMs": 135120,
    "text": "where we can go from there. I find it's"
  },
  {
    "startMs": 137000,
    "text": "hard. I don't know if people have this"
  },
  {
    "startMs": 137920,
    "text": "experience where"
  },
  {
    "startMs": 139360,
    "text": "people"
  },
  {
    "startMs": 140040,
    "text": "people's only been re-enabled for a"
  },
  {
    "startMs": 141200,
    "text": "couple of days. People's definitely way"
  },
  {
    "startMs": 142880,
    "text": "way smarter than me. So, sometimes it'll"
  },
  {
    "startMs": 144240,
    "text": "finish work and be like, here's the"
  },
  {
    "startMs": 145640,
    "text": "trade-offs I made. I'm like, can you"
  },
  {
    "startMs": 147160,
    "text": "explain it to me like I'm a little"
  },
  {
    "startMs": 149120,
    "text": "dumber than you are because I need you"
  },
  {
    "startMs": 150680,
    "text": "to like sort of break this down for me."
  },
  {
    "startMs": 152440,
    "text": "But, that's been one sort of big change"
  },
  {
    "startMs": 154200,
    "text": "is sort of moving from that task"
  },
  {
    "startMs": 155880,
    "text": "delegation to like express the end state"
  },
  {
    "startMs": 157920,
    "text": "and then have it go and and cook on it."
  },
  {
    "startMs": 160240,
    "text": ">> Yeah,"
  },
  {
    "startMs": 161040,
    "text": "we're all learning how to delegate"
  },
  {
    "startMs": 162720,
    "text": "better. Tariq did us a huge favor"
  },
  {
    "startMs": 164880,
    "text": "yesterday."
  },
  {
    "startMs": 166360,
    "text": "We"
  },
  {
    "startMs": 167200,
    "text": "Did you want to read it in the"
  },
  {
    "startMs": 168240,
    "text": "newspaper?"
  },
  {
    "startMs": 170240,
    "text": "You know, that we have we have"
  },
  {
    "startMs": 171880,
    "text": "write-ups of talks now in in like the"
  },
  {
    "startMs": 173640,
    "text": "next day's newspaper."
  },
  {
    "startMs": 174840,
    "text": ">> He said, be unreasonable."
  },
  {
    "startMs": 177080,
    "text": "In what ways you know, you know, have"
  },
  {
    "startMs": 178640,
    "text": "you been more ambitious?"
  },
  {
    "startMs": 179880,
    "text": ">> Yeah, you're prompting."
  },
  {
    "startMs": 180920,
    "text": ">> I love that I I mean, I love that"
  },
  {
    "startMs": 182320,
    "text": "framing. We actually just hit this"
  },
  {
    "startMs": 184440,
    "text": "today. I'm one of the labs initiatives I"
  },
  {
    "startMs": 186320,
    "text": "have is internal product, and somebody"
  },
  {
    "startMs": 189840,
    "text": "was like, hey, it doesn't work the way I"
  },
  {
    "startMs": 191360,
    "text": "want it to,"
  },
  {
    "startMs": 193080,
    "text": "and can you make some changes? And I"
  },
  {
    "startMs": 195160,
    "text": "realized, oh, I'm just going to go ask"
  },
  {
    "startMs": 196840,
    "text": "Claude to do this. Like, why don't you"
  },
  {
    "startMs": 198280,
    "text": "ask Claude? And this was a non-technical"
  },
  {
    "startMs": 199800,
    "text": "person. So, I actually think as an"
  },
  {
    "startMs": 201360,
    "text": "industry or even as a product team, we"
  },
  {
    "startMs": 203800,
    "text": "have to teach people to be more"
  },
  {
    "startMs": 205239,
    "text": "unreasonable in their usage, and it's"
  },
  {
    "startMs": 207000,
    "text": "sort of hard to imagine. I think that"
  },
  {
    "startMs": 209160,
    "text": "that"
  },
  {
    "startMs": 209880,
    "text": "if I can digress for a second on product"
  },
  {
    "startMs": 211440,
    "text": "design, I think right now the like kind"
  },
  {
    "startMs": 213560,
    "text": "of first generation of AI products, we"
  },
  {
    "startMs": 215080,
    "text": "put them too much in a box and constrain"
  },
  {
    "startMs": 217160,
    "text": "their their sort of access to tools or"
  },
  {
    "startMs": 220239,
    "text": "kind of degrees of freedom, which means"
  },
  {
    "startMs": 221800,
    "text": "it was much harder to be unreasonable,"
  },
  {
    "startMs": 223160,
    "text": "right? When you say, do this thing for"
  },
  {
    "startMs": 225080,
    "text": "me, and then it would be like, well, I"
  },
  {
    "startMs": 226959,
    "text": "can't. I can barely like I can write"
  },
  {
    "startMs": 229080,
    "text": "code, but I can't really run it, or I"
  },
  {
    "startMs": 230560,
    "text": "can kind of introspect my environment,"
  },
  {
    "startMs": 232160,
    "text": "but not really."
  },
  {
    "startMs": 233680,
    "text": "Um and I think as you see our own like"
  },
  {
    "startMs": 235240,
    "text": "product progression even with things"
  },
  {
    "startMs": 236720,
    "text": "like co-work where like, you know, does"
  },
  {
    "startMs": 238920,
    "text": "every single like knowledge worker need"
  },
  {
    "startMs": 241000,
    "text": "a virtual machine that can write bash?"
  },
  {
    "startMs": 242920,
    "text": "Like, on the face of it, no, but then"
  },
  {
    "startMs": 244880,
    "text": "when you realize, oh, actually, that way"
  },
  {
    "startMs": 246440,
    "text": "it can remediate an issue where, oh, I"
  },
  {
    "startMs": 248320,
    "text": "tried to parse a PDF using our built-in"
  },
  {
    "startMs": 250000,
    "text": "PDF parser. I hit this yesterday and it"
  },
  {
    "startMs": 251800,
    "text": "was like, ah, I can't parse it this way."
  },
  {
    "startMs": 253320,
    "text": "Well, okay, well, I can probably write a"
  },
  {
    "startMs": 254800,
    "text": "script that can do this as well. Um so,"
  },
  {
    "startMs": 257040,
    "text": "I think that's it. My most unreasonable"
  },
  {
    "startMs": 258840,
    "text": "thing though was uh one of our labs"
  },
  {
    "startMs": 260720,
    "text": "projects I wrote in Python like near and"
  },
  {
    "startMs": 262560,
    "text": "dear to my heart. All of Instagram was"
  },
  {
    "startMs": 263919,
    "text": "in Python. But I think they're finally"
  },
  {
    "startMs": 265640,
    "text": "converting it to PHP now that they have"
  },
  {
    "startMs": 267960,
    "text": "um like models that can do it. I know."
  },
  {
    "startMs": 270518,
    "text": ">> [laughter]"
  },
  {
    "startMs": 271720,
    "text": ">> Tokens. Um and uh for deployment I"
  },
  {
    "startMs": 274800,
    "text": "realized that Cloud Code had like"
  },
  {
    "startMs": 275960,
    "text": "figured out a better deployment story"
  },
  {
    "startMs": 277320,
    "text": "with Bun. And I was like, okay, I need"
  },
  {
    "startMs": 278760,
    "text": "to port this whole thing from Python to"
  },
  {
    "startMs": 280680,
    "text": "TypeScript. Like, as a, you know, if I"
  },
  {
    "startMs": 282120,
    "text": "put on my like 2010s engineering hat or"
  },
  {
    "startMs": 284000,
    "text": "even my early 20 20s, like, that's a"
  },
  {
    "startMs": 285760,
    "text": "dumb idea. Like, who would ever port"
  },
  {
    "startMs": 287560,
    "text": "like, at that point, you know, a couple"
  },
  {
    "startMs": 288960,
    "text": "hundred thousands of lines of code. Um"
  },
  {
    "startMs": 291160,
    "text": "but I was like, I think this is doable"
  },
  {
    "startMs": 292480,
    "text": "now and I basically created this dynamic"
  },
  {
    "startMs": 294080,
    "text": "workflow setup and over the weekend had"
  },
  {
    "startMs": 296040,
    "text": "it port the whole thing, like, verify"
  },
  {
    "startMs": 298520,
    "text": "it, double-check it, then read both code"
  },
  {
    "startMs": 300520,
    "text": "like this basically churn and churn and"
  },
  {
    "startMs": 301960,
    "text": "churn and then came back Monday to a"
  },
  {
    "startMs": 304200,
    "text": "completed workflow that was a ported"
  },
  {
    "startMs": 306160,
    "text": "version of that thing. So, that probably"
  },
  {
    "startMs": 307960,
    "text": "ranks on like the more unreasonable"
  },
  {
    "startMs": 309360,
    "text": "things. Like, yeah, just port this"
  },
  {
    "startMs": 310440,
    "text": "entire Python code base to TypeScript,"
  },
  {
    "startMs": 312840,
    "text": "get it working, get it deployable in,"
  },
  {
    "startMs": 314680,
    "text": "you know, a weekend."
  },
  {
    "startMs": 316280,
    "text": ">> Yeah, I mean, a lot of people are"
  },
  {
    "startMs": 317360,
    "text": "talking about the the Bun Zig to Rust"
  },
  {
    "startMs": 320120,
    "text": "version. I think a lot of people are"
  },
  {
    "startMs": 321960,
    "text": "also like, well, it's a compiler, it's a"
  },
  {
    "startMs": 324120,
    "text": "it's a runtime, it's got lots of tests,"
  },
  {
    "startMs": 326080,
    "text": "easy to do. Can you port Instagram,"
  },
  {
    "startMs": 329440,
    "text": "which you would know very well, to PHP"
  },
  {
    "startMs": 332200,
    "text": "like that? Like a like a product."
  },
  {
    "startMs": 333760,
    "text": ">> Yeah, I mean, I think the product side"
  },
  {
    "startMs": 335240,
    "text": "of it it's even I don't know if it's"
  },
  {
    "startMs": 337000,
    "text": "easier or harder. One of the things we"
  },
  {
    "startMs": 337919,
    "text": "did at Instagram, this is when Python 3"
  },
  {
    "startMs": 339680,
    "text": "came out and we were able to add type"
  },
  {
    "startMs": 341600,
    "text": "hints for the first time and it was"
  },
  {
    "startMs": 343000,
    "text": "people had a lot of internal"
  },
  {
    "startMs": 343960,
    "text": "conversations like, are we going to run"
  },
  {
    "startMs": 345120,
    "text": "out of steam on Python? And my"
  },
  {
    "startMs": 346840,
    "text": "perspective was always like, I think we"
  },
  {
    "startMs": 348280,
    "text": "can take this way further than we think"
  },
  {
    "startMs": 349920,
    "text": "we can, uh, but I think types are going"
  },
  {
    "startMs": 351720,
    "text": "to help us not sort of be in our own"
  },
  {
    "startMs": 353640,
    "text": "way. And we built this thing called"
  },
  {
    "startMs": 356280,
    "text": "Monkey Type where we basically like"
  },
  {
    "startMs": 357520,
    "text": "captured runtime type like basically the"
  },
  {
    "startMs": 360240,
    "text": "types that were actually getting used in"
  },
  {
    "startMs": 361240,
    "text": "production and then map those back to to"
  },
  {
    "startMs": 363680,
    "text": "the types in the code base. And I think"
  },
  {
    "startMs": 365880,
    "text": "because of that sort of pattern, I think"
  },
  {
    "startMs": 368120,
    "text": "there's really interesting ways in which"
  },
  {
    "startMs": 369560,
    "text": "if you're doing sort of conversion or"
  },
  {
    "startMs": 371320,
    "text": "sort of cross compiling using LLMs, you"
  },
  {
    "startMs": 373880,
    "text": "can also lean on production data a lot"
  },
  {
    "startMs": 375600,
    "text": "more or run sort of like segmented"
  },
  {
    "startMs": 377160,
    "text": "tests. I think that like there's a lot"
  },
  {
    "startMs": 378600,
    "text": "of, uh, things you can do there. But"
  },
  {
    "startMs": 380200,
    "text": "yeah, I think it's, I mean, the sky's"
  },
  {
    "startMs": 381680,
    "text": "the limit there as well. I think the"
  },
  {
    "startMs": 382960,
    "text": "hardest part is always finding the"
  },
  {
    "startMs": 383960,
    "text": "boundary around where you can start"
  },
  {
    "startMs": 385240,
    "text": "doing it incrementally without trying to"
  },
  {
    "startMs": 386560,
    "text": "boil the whole ocean and like swap it"
  },
  {
    "startMs": 388320,
    "text": "overnight."
  },
  {
    "startMs": 389240,
    "text": ">> Yeah, I mean, your users are your test"
  },
  {
    "startMs": 390720,
    "text": "ultimately and, um,"
  },
  {
    "startMs": 392680,
    "text": "you know, I we I also read another"
  },
  {
    "startMs": 394040,
    "text": "article in the newspaper about how you"
  },
  {
    "startMs": 395400,
    "text": "could just use rollouts and sometimes"
  },
  {
    "startMs": 397520,
    "text": "you don't really know, uh, what you're"
  },
  {
    "startMs": 399400,
    "text": "going to need it for, but when that"
  },
  {
    "startMs": 400960,
    "text": "infrastructure exists for your"
  },
  {
    "startMs": 402240,
    "text": "experiments and to roll things out, it's"
  },
  {
    "startMs": 404560,
    "text": "enabled so much."
  },
  {
    "startMs": 405560,
    "text": ">> Yeah, I mean, I always found this was"
  },
  {
    "startMs": 406640,
    "text": "advice we got. It's like we launched"
  },
  {
    "startMs": 408000,
    "text": "Instagram and the happened to be the"
  },
  {
    "startMs": 409880,
    "text": "first week everything melted cuz we"
  },
  {
    "startMs": 411400,
    "text": "didn't really know what we were doing on"
  },
  {
    "startMs": 412400,
    "text": "the back end side of things. And, uh,"
  },
  {
    "startMs": 414840,
    "text": "coincidentally that week there was like"
  },
  {
    "startMs": 416240,
    "text": "a lunch that one of our investors just"
  },
  {
    "startMs": 417800,
    "text": "scheduled like not even for us. It was"
  },
  {
    "startMs": 419600,
    "text": "just a"
  },
  {
    "startMs": 420520,
    "text": "infrastructure lunch. And we ended up"
  },
  {
    "startMs": 422040,
    "text": "spending we totally like monopolized"
  },
  {
    "startMs": 424040,
    "text": "that conversation cuz everybody had"
  },
  {
    "startMs": 425280,
    "text": "their own opinion about how we could fix"
  },
  {
    "startMs": 426560,
    "text": "our scaling. Um, and like the two pieces"
  },
  {
    "startMs": 428640,
    "text": "of advice I got there is like 2010 that"
  },
  {
    "startMs": 430600,
    "text": "I like will forever retain is like, um,"
  },
  {
    "startMs": 433560,
    "text": "like basically like"
  },
  {
    "startMs": 435040,
    "text": "pre-measure everything that you think"
  },
  {
    "startMs": 436560,
    "text": "you might even remotely need because the"
  },
  {
    "startMs": 438840,
    "text": "worst thing is an outage where you're"
  },
  {
    "startMs": 439880,
    "text": "like, well, is this like number normal"
  },
  {
    "startMs": 442640,
    "text": "or is it high? And like, oh, I don't"
  },
  {
    "startMs": 443960,
    "text": "know because I don't have data until I"
  },
  {
    "startMs": 445320,
    "text": "just added this metric. And the other"
  },
  {
    "startMs": 446720,
    "text": "one is being like really thoughtful"
  },
  {
    "startMs": 447880,
    "text": "about knobs and feature flags. So even,"
  },
  {
    "startMs": 449919,
    "text": "you know, early Instagram we had like a"
  },
  {
    "startMs": 451120,
    "text": "very, uh, simple but really effective"
  },
  {
    "startMs": 453919,
    "text": "like way in which you could do like ramp"
  },
  {
    "startMs": 455680,
    "text": "outs and rollouts. And dynamic config"
  },
  {
    "startMs": 457160,
    "text": "too where, you know, a lot of our"
  },
  {
    "startMs": 459000,
    "text": "runtime configurations had to be"
  },
  {
    "startMs": 460560,
    "text": "changed, you know, in a matter of"
  },
  {
    "startMs": 461760,
    "text": "seconds so that we could handle load and"
  },
  {
    "startMs": 463400,
    "text": "being able to like do that in a first"
  },
  {
    "startMs": 464680,
    "text": "class way was was really important. I'm"
  },
  {
    "startMs": 466360,
    "text": "seeing that definitely in in AI as well"
  },
  {
    "startMs": 468120,
    "text": "where, you know, we're making all sorts"
  },
  {
    "startMs": 469560,
    "text": "of different trade-offs and having that"
  },
  {
    "startMs": 470720,
    "text": "kind of runtime configuration is super"
  },
  {
    "startMs": 472200,
    "text": "key."
  },
  {
    "startMs": 472400,
    "text": ">> Yeah."
  },
  {
    "startMs": 473240,
    "text": "Uh my my favorite scaling story of"
  },
  {
    "startMs": 474800,
    "text": "Instagram by the way, I think it's like"
  },
  {
    "startMs": 476040,
    "text": "your launch day when you you DDoS"
  },
  {
    "startMs": 477720,
    "text": "yourself with the email."
  },
  {
    "startMs": 478920,
    "text": ">> Yes."
  },
  {
    "startMs": 480240,
    "text": ">> Which I people should look up that story"
  },
  {
    "startMs": 481440,
    "text": "if uh if you haven't seen it. Um I"
  },
  {
    "startMs": 483440,
    "text": "wanted to go into tags. Uh"
  },
  {
    "startMs": 485560,
    "text": "very very major shift. Uh it's it's how"
  },
  {
    "startMs": 487520,
    "text": "60 something percent of your code is"
  },
  {
    "startMs": 489080,
    "text": "written today."
  },
  {
    "startMs": 489760,
    "text": ">> Yeah."
  },
  {
    "startMs": 490400,
    "text": ">> Um how do you square that with"
  },
  {
    "startMs": 493240,
    "text": "everything you just said where it's like"
  },
  {
    "startMs": 494560,
    "text": "very dynamic? Like you don't actually"
  },
  {
    "startMs": 496120,
    "text": "ship one app, you ship one app with"
  },
  {
    "startMs": 497600,
    "text": "3,000 flags."
  },
  {
    "startMs": 498720,
    "text": ">> Yeah."
  },
  {
    "startMs": 499040,
    "text": ">> And like, well, what are you working on"
  },
  {
    "startMs": 500120,
    "text": "today? I don't know. Like it's it's for"
  },
  {
    "startMs": 501440,
    "text": "this segment of the population."
  },
  {
    "startMs": 503040,
    "text": ">> Yeah. Yeah, I mean, I think there's a"
  },
  {
    "startMs": 504600,
    "text": "bunch of things. So, like with I was"
  },
  {
    "startMs": 506680,
    "text": "really excited. I was talking to Swix"
  },
  {
    "startMs": 507800,
    "text": "earlier like, I'm really excited that we"
  },
  {
    "startMs": 508840,
    "text": "have tag out there because it is uh how"
  },
  {
    "startMs": 511240,
    "text": "we've been working for a while and I"
  },
  {
    "startMs": 513400,
    "text": "would get up on stages and people like,"
  },
  {
    "startMs": 515000,
    "text": "\"How do you work at Anthropic?\" And I'd"
  },
  {
    "startMs": 516240,
    "text": "be like, \"Oh, yeah, we use these things"
  },
  {
    "startMs": 518200,
    "text": "like that are not quite Claude code, but"
  },
  {
    "startMs": 520320,
    "text": "you know, uh but it's hard to describe"
  },
  {
    "startMs": 521880,
    "text": "it, but I mean, if you like got to poke"
  },
  {
    "startMs": 524120,
    "text": "into Anthropic, like you would see uh of"
  },
  {
    "startMs": 527080,
    "text": "course Claude code usage for things that"
  },
  {
    "startMs": 528520,
    "text": "are like more interactive or if you're"
  },
  {
    "startMs": 529920,
    "text": "kind of iterating on a particular uh"
  },
  {
    "startMs": 531880,
    "text": "sort of"
  },
  {
    "startMs": 532839,
    "text": "sort of specific thing where you want a"
  },
  {
    "startMs": 534080,
    "text": "lot of like a high sort of bandwidth"
  },
  {
    "startMs": 536320,
    "text": "back and forth, but most usage is"
  },
  {
    "startMs": 538120,
    "text": "actually much more delegating uh via"
  },
  {
    "startMs": 540800,
    "text": "tagging and via tag. And you can say"
  },
  {
    "startMs": 542640,
    "text": "like, \"Here's the And the reason it's"
  },
  {
    "startMs": 544360,
    "text": "really interesting is how multiplayer it"
  },
  {
    "startMs": 545920,
    "text": "is.\" And it reminds me sort of of like"
  },
  {
    "startMs": 547920,
    "text": "um actually like Midjourney, like the"
  },
  {
    "startMs": 549360,
    "text": "fact that everyone was on Discord seeing"
  },
  {
    "startMs": 551640,
    "text": "how other people were using it. I think"
  },
  {
    "startMs": 553360,
    "text": "it actually to your earlier question"
  },
  {
    "startMs": 554360,
    "text": "really helps with that unreasonableness"
  },
  {
    "startMs": 555920,
    "text": "or ambition where the first time you see"
  },
  {
    "startMs": 558040,
    "text": "somebody tag Claude and be like, \"Hey,"
  },
  {
    "startMs": 560480,
    "text": "you know, don't just fix this bug, but"
  },
  {
    "startMs": 561680,
    "text": "like now you are responsible for this"
  },
  {
    "startMs": 564240,
    "text": "part of the code base and I want you to"
  },
  {
    "startMs": 565760,
    "text": "monitor this feedback channel and"
  },
  {
    "startMs": 567560,
    "text": "proactively take on tasks and then fix"
  },
  {
    "startMs": 569760,
    "text": "them and then also take like, you know,"
  },
  {
    "startMs": 571400,
    "text": "if this API changes, do that.\" Like I"
  },
  {
    "startMs": 573520,
    "text": "saw somebody do that. I was like, \"Oh,"
  },
  {
    "startMs": 574520,
    "text": "wait, I've I've totally underutilizing"
  },
  {
    "startMs": 576320,
    "text": "this thing. I've just been using it as"
  },
  {
    "startMs": 577760,
    "text": "like a glorified Claude code and slack."
  },
  {
    "startMs": 579440,
    "text": "Like that's definitely a totally like"
  },
  {
    "startMs": 582240,
    "text": "sort of new version of it, right? And"
  },
  {
    "startMs": 584000,
    "text": "then more advanced version is really"
  },
  {
    "startMs": 585040,
    "text": "trying to start thinking of it as a"
  },
  {
    "startMs": 586400,
    "text": "teammate that is actually sort of holds"
  },
  {
    "startMs": 588680,
    "text": "context, has memory, and can be"
  },
  {
    "startMs": 590120,
    "text": "proactive. And that's just really"
  },
  {
    "startMs": 592200,
    "text": "changed how we operate internally. It's"
  },
  {
    "startMs": 594000,
    "text": "much more like this multiplayer async"
  },
  {
    "startMs": 597080,
    "text": "proactive way than it is a you know,"
  },
  {
    "startMs": 599160,
    "text": "most people often their own CLIs."
  },
  {
    "startMs": 601560,
    "text": ">> Are you bottlenecked by code review and"
  },
  {
    "startMs": 603680,
    "text": "get? Obviously, there is code review,"
  },
  {
    "startMs": 605920,
    "text": "but someone usually still looks at it."
  },
  {
    "startMs": 608280,
    "text": "Is there a world in which you just merge"
  },
  {
    "startMs": 610200,
    "text": "it in?"
  },
  {
    "startMs": 611040,
    "text": ">> Yeah, we're it's a really good question."
  },
  {
    "startMs": 613360,
    "text": "We are definitely still bottlenecked on"
  },
  {
    "startMs": 615400,
    "text": "reviews, especially for things that are"
  },
  {
    "startMs": 617160,
    "text": "like touching some architecture pieces."
  },
  {
    "startMs": 619440,
    "text": "And it's actually more subtle than just"
  },
  {
    "startMs": 620760,
    "text": "being bottlenecked on review, cuz that's"
  },
  {
    "startMs": 622680,
    "text": "you know, okay, we can carve out time"
  },
  {
    "startMs": 624200,
    "text": "differently. It's like bottlenecked on"
  },
  {
    "startMs": 626040,
    "text": "human ability to even"
  },
  {
    "startMs": 628200,
    "text": "like fully conceptualize what we're"
  },
  {
    "startMs": 630000,
    "text": "doing. So, one of the reasons we built"
  },
  {
    "startMs": 631560,
    "text": "Claude code artifacts that we shipped a"
  },
  {
    "startMs": 633000,
    "text": "couple weeks ago was partially for that,"
  },
  {
    "startMs": 635120,
    "text": "which is"
  },
  {
    "startMs": 636080,
    "text": "you would send somebody a PR, and then"
  },
  {
    "startMs": 638400,
    "text": "they'd be like, I don't know, man. This"
  },
  {
    "startMs": 639800,
    "text": "is like 2,000 lines of code. Like, it"
  },
  {
    "startMs": 642120,
    "text": "looks like code to me. And what we"
  },
  {
    "startMs": 643880,
    "text": "started doing instead is sharing much"
  },
  {
    "startMs": 645360,
    "text": "more like, here's a Claude code"
  },
  {
    "startMs": 647040,
    "text": "artifact. Like, here's the explanation."
  },
  {
    "startMs": 649200,
    "text": "Here's the intention of the the change."
  },
  {
    "startMs": 651320,
    "text": "Here's the trade-offs that were made."
  },
  {
    "startMs": 652720,
    "text": "And like, I think that's going to be"
  },
  {
    "startMs": 654560,
    "text": "much more be the trend by which we"
  },
  {
    "startMs": 656040,
    "text": "communicate, which is the code is"
  },
  {
    "startMs": 657720,
    "text": "ultimately, you know, verifiable using"
  },
  {
    "startMs": 660320,
    "text": "some things, but actually like"
  },
  {
    "startMs": 661760,
    "text": "discussing intent and trade-offs, and"
  },
  {
    "startMs": 663880,
    "text": "then measuring in production I think"
  },
  {
    "startMs": 665839,
    "text": "that at least the direction of travel"
  },
  {
    "startMs": 667160,
    "text": "we've we've gone. And I don't review"
  },
  {
    "startMs": 669160,
    "text": "when I get a pull request, I wish I"
  },
  {
    "startMs": 670920,
    "text": "could say I reviewed every line of code."
  },
  {
    "startMs": 672080,
    "text": "I definitely do not. I like actually"
  },
  {
    "startMs": 673600,
    "text": "talk to Claude about the the code and"
  },
  {
    "startMs": 675200,
    "text": "say, all right, like, these are the"
  },
  {
    "startMs": 676440,
    "text": "questions that I would have. Can you go"
  },
  {
    "startMs": 677720,
    "text": "investigate it? So, it is kind of"
  },
  {
    "startMs": 679320,
    "text": "Claude-powered code review, but still"
  },
  {
    "startMs": 680640,
    "text": "human-driven. And and for the really"
  },
  {
    "startMs": 682360,
    "text": "important ones. And for the ones that"
  },
  {
    "startMs": 683560,
    "text": "are like cosmetic visual changes, it's"
  },
  {
    "startMs": 685640,
    "text": "much more like look like we'll fix"
  },
  {
    "startMs": 687680,
    "text": "forward if we need to fix forward, you"
  },
  {
    "startMs": 689120,
    "text": "know."
  },
  {
    "startMs": 690120,
    "text": ">> Yeah, totally. I think a lot of people"
  },
  {
    "startMs": 691600,
    "text": "are here are trying to figure that out,"
  },
  {
    "startMs": 692840,
    "text": "too."
  },
  {
    "startMs": 694000,
    "text": "I wanted to talk also a little bit about"
  },
  {
    "startMs": 696400,
    "text": "Anthropic Labs in general."
  },
  {
    "startMs": 698400,
    "text": "Nilay Patel,"
  },
  {
    "startMs": 699720,
    "text": "who you've probably met before, loves to"
  },
  {
    "startMs": 701720,
    "text": "ask ask the question like draw the org"
  },
  {
    "startMs": 703440,
    "text": "chart."
  },
  {
    "startMs": 704520,
    "text": "Like how like people, you know, you ship"
  },
  {
    "startMs": 706800,
    "text": "your org chart. Like I think it's"
  },
  {
    "startMs": 708320,
    "text": "important like everyone knows cloud"
  },
  {
    "startMs": 709680,
    "text": "code. Now you've got tags."
  },
  {
    "startMs": 712320,
    "text": "Um How are you structuring the labs?"
  },
  {
    "startMs": 714480,
    "text": ">> Yeah, it's a good question. Because what"
  },
  {
    "startMs": 716120,
    "text": "we were trying to wrestle with was you"
  },
  {
    "startMs": 718480,
    "text": "want sort of people to be supported"
  },
  {
    "startMs": 721160,
    "text": "like, you know, I think the death of the"
  },
  {
    "startMs": 722960,
    "text": "engineering manager discipline has been"
  },
  {
    "startMs": 724440,
    "text": "greatly exaggerated. Like I think"
  },
  {
    "startMs": 725640,
    "text": "there's still a lot of coaching and"
  },
  {
    "startMs": 727560,
    "text": "interpersonal pieces and personal"
  },
  {
    "startMs": 729240,
    "text": "development that I think is still"
  },
  {
    "startMs": 730160,
    "text": "really, really important. But especially"
  },
  {
    "startMs": 732480,
    "text": "in a labs type group where like our"
  },
  {
    "startMs": 734800,
    "text": "whole cadence is two-week reviews where"
  },
  {
    "startMs": 737400,
    "text": "every project goes up for we call it"
  },
  {
    "startMs": 739480,
    "text": "persevere or pivot. So basically every"
  },
  {
    "startMs": 741200,
    "text": "project is up for review and either it's"
  },
  {
    "startMs": 743520,
    "text": "time to, you know, keep going,"
  },
  {
    "startMs": 745240,
    "text": "persevering, or you know, it's time to"
  },
  {
    "startMs": 746680,
    "text": "pivot it or even shut down. And, you"
  },
  {
    "startMs": 748840,
    "text": "know, we've shut down projects basically"
  },
  {
    "startMs": 751040,
    "text": "every single one of those cycles and"
  },
  {
    "startMs": 752360,
    "text": "it's like the more you do it, the less"
  },
  {
    "startMs": 753880,
    "text": "it's just like, \"Oh no, my project is"
  },
  {
    "startMs": 755360,
    "text": "shut down. I failed.\" It's like, \"No,"
  },
  {
    "startMs": 756960,
    "text": "that is definitely the intention of the"
  },
  {
    "startMs": 758480,
    "text": "labs team is to prototype quickly, try"
  },
  {
    "startMs": 759960,
    "text": "to ship internally, maybe get it to"
  },
  {
    "startMs": 761720,
    "text": "early access, and if it doesn't work,"
  },
  {
    "startMs": 763240,
    "text": "wind it down.\" But because of that kind"
  },
  {
    "startMs": 765280,
    "text": "of like rapid iteration, it means that"
  },
  {
    "startMs": 766920,
    "text": "if you align the org chart too much to"
  },
  {
    "startMs": 769400,
    "text": "the individual projects, you're going to"
  },
  {
    "startMs": 770680,
    "text": "end up like re-orging every two weeks,"
  },
  {
    "startMs": 772680,
    "text": "which would be a total nightmare. And so"
  },
  {
    "startMs": 774200,
    "text": "we've actually ended up with this"
  },
  {
    "startMs": 775160,
    "text": "interesting setup where like the the pod"
  },
  {
    "startMs": 777839,
    "text": "or the team that is working on a given"
  },
  {
    "startMs": 779720,
    "text": "we call them bets within labs,"
  },
  {
    "startMs": 781760,
    "text": "definitely just draws upon like all"
  },
  {
    "startMs": 783240,
    "text": "right, somebody from product, somebody"
  },
  {
    "startMs": 784720,
    "text": "from the eng team,"
  },
  {
    "startMs": 786280,
    "text": "you know, I'll jump in when it's a"
  },
  {
    "startMs": 787560,
    "text": "product I'm particularly interested in."
  },
  {
    "startMs": 789000,
    "text": "I'll come in and work together with the"
  },
  {
    "startMs": 790360,
    "text": "team on it. And that's the unit for that"
  },
  {
    "startMs": 792640,
    "text": "time. And there is the concept of a bet"
  },
  {
    "startMs": 794440,
    "text": "lead or a directly responsible"
  },
  {
    "startMs": 795760,
    "text": "individual. But the interesting thing is"
  },
  {
    "startMs": 797240,
    "text": "that they don't manage usually any of"
  },
  {
    "startMs": 798960,
    "text": "the other people, which kind of breaks"
  },
  {
    "startMs": 800960,
    "text": "the that kind of previous way in which a"
  },
  {
    "startMs": 802520,
    "text": "lot of these things were done. But I"
  },
  {
    "startMs": 803800,
    "text": "think it leaves it leaves us to be"
  },
  {
    "startMs": 804839,
    "text": "really flexible when you say, \"Okay,"
  },
  {
    "startMs": 806240,
    "text": "actually this project is not going to"
  },
  {
    "startMs": 807960,
    "text": "work out. Let's disband and keep going"
  },
  {
    "startMs": 809960,
    "text": "and it's not a big deal.\" And the engine"
  },
  {
    "startMs": 811520,
    "text": "manager is much more playing the like"
  },
  {
    "startMs": 813560,
    "text": "make sure every individual is assigned"
  },
  {
    "startMs": 815160,
    "text": "to the thing that they're most excited"
  },
  {
    "startMs": 816280,
    "text": "about and that they're working in the"
  },
  {
    "startMs": 817480,
    "text": "best way possible. Now, what we do sort"
  },
  {
    "startMs": 819720,
    "text": "of solidify is when there's a product"
  },
  {
    "startMs": 821160,
    "text": "that has like legs. Like Cloud Design"
  },
  {
    "startMs": 822840,
    "text": "for example, started in this sort of ad"
  },
  {
    "startMs": 824640,
    "text": "hoc sort of grouped way and then now"
  },
  {
    "startMs": 827360,
    "text": "that like we've shipped it, it's gotten"
  },
  {
    "startMs": 828680,
    "text": "traction, we've done like a big second"
  },
  {
    "startMs": 830400,
    "text": "release in June. Like it's becoming like"
  },
  {
    "startMs": 833160,
    "text": "we've hired people for that specific"
  },
  {
    "startMs": 835040,
    "text": "team and it has more of a of a"
  },
  {
    "startMs": 836880,
    "text": "structure. So, it's like loose until it"
  },
  {
    "startMs": 838720,
    "text": "gets solidified down the line."
  },
  {
    "startMs": 841000,
    "text": ">> What's the future of Cloud Design? I"
  },
  {
    "startMs": 842400,
    "text": "think a lot of people are very"
  },
  {
    "startMs": 843360,
    "text": "interested in It's one of your biggest"
  },
  {
    "startMs": 845480,
    "text": "launches this year."
  },
  {
    "startMs": 847520,
    "text": "Where does this go?"
  },
  {
    "startMs": 848520,
    "text": ">> I think for me, I mean"
  },
  {
    "startMs": 850840,
    "text": "the things that are holding back Cloud"
  },
  {
    "startMs": 852960,
    "text": "Design from being even better is better"
  },
  {
    "startMs": 854880,
    "text": "interaction with our other surfaces. So,"
  },
  {
    "startMs": 856840,
    "text": "you know, I was designing something or I"
  },
  {
    "startMs": 858920,
    "text": "was talking to to Cloud Code the other"
  },
  {
    "startMs": 861000,
    "text": "day. I'm like, I want a really much more"
  },
  {
    "startMs": 863120,
    "text": "seamless like what I'm talking about the"
  },
  {
    "startMs": 865280,
    "text": "design for it, you know, and then after"
  },
  {
    "startMs": 866680,
    "text": "design back to that. I think in general"
  },
  {
    "startMs": 868440,
    "text": "it's I mean this goes back again to kind"
  },
  {
    "startMs": 870640,
    "text": "of unconstraining Cloud. Like the fact"
  },
  {
    "startMs": 873120,
    "text": "that our surfaces don't talk to each"
  },
  {
    "startMs": 874960,
    "text": "other as well as they could. I think"
  },
  {
    "startMs": 876040,
    "text": "really holds back a lot of interesting"
  },
  {
    "startMs": 877880,
    "text": "ideas around what we could do. So, I"
  },
  {
    "startMs": 879280,
    "text": "think that's one like kind of major area"
  },
  {
    "startMs": 881560,
    "text": "that we're looking at."
  },
  {
    "startMs": 883160,
    "text": "And then the other one is people like"
  },
  {
    "startMs": 884880,
    "text": "the lines between a Cloud Design and an"
  },
  {
    "startMs": 887760,
    "text": "app get blurry and blurrier over time."
  },
  {
    "startMs": 889520,
    "text": "Like I've seen people Of course there's"
  },
  {
    "startMs": 890960,
    "text": "no like persistence but build like fully"
  },
  {
    "startMs": 893360,
    "text": "functional like even games which is"
  },
  {
    "startMs": 895360,
    "text": "definitely not what we designed Cloud"
  },
  {
    "startMs": 896360,
    "text": "Design for but you can do it. It's just"
  },
  {
    "startMs": 898000,
    "text": "HTML and JavaScript."
  },
  {
    "startMs": 900200,
    "text": "So, blurring those lines even further"
  },
  {
    "startMs": 901839,
    "text": "and thinking through like what is the"
  },
  {
    "startMs": 903120,
    "text": "path from a like fully featured design"
  },
  {
    "startMs": 905200,
    "text": "that looks really well to really good to"
  },
  {
    "startMs": 907520,
    "text": "something that is maybe more like a"
  },
  {
    "startMs": 909080,
    "text": "artifact where you're actually able to"
  },
  {
    "startMs": 910400,
    "text": "go and you know, persist data and share"
  },
  {
    "startMs": 912400,
    "text": "it with others and build from there. So,"
  },
  {
    "startMs": 913800,
    "text": "I think that those lines get really"
  },
  {
    "startMs": 915200,
    "text": "interesting over time, too."
  },
  {
    "startMs": 916839,
    "text": ">> Yeah."
  },
  {
    "startMs": 918080,
    "text": "A big part of design is having taste. I"
  },
  {
    "startMs": 920520,
    "text": "actually asked Fable what Fable wants to"
  },
  {
    "startMs": 923040,
    "text": "ask you"
  },
  {
    "startMs": 924400,
    "text": "and this this this is what Fable came up"
  },
  {
    "startMs": 926120,
    "text": "with. You deleted almost all of Bourbon"
  },
  {
    "startMs": 928560,
    "text": "to get to Instagram which is like you"
  },
  {
    "startMs": 930080,
    "text": "had a whole you know solo mode whatever"
  },
  {
    "startMs": 932920,
    "text": "thing and you went to Instagram. What"
  },
  {
    "startMs": 935360,
    "text": "would you delete in AI or more spicy"
  },
  {
    "startMs": 937400,
    "text": "what would you delete in Claude?"
  },
  {
    "startMs": 938680,
    "text": ">> Oh, I like the spice."
  },
  {
    "startMs": 941440,
    "text": "I think I mean we have it's interesting"
  },
  {
    "startMs": 943120,
    "text": "we have a one of our slack channels is"
  },
  {
    "startMs": 944880,
    "text": "like project unship which is like what"
  },
  {
    "startMs": 946640,
    "text": "is in the product right now. It's"
  },
  {
    "startMs": 948520,
    "text": "I mean this is hard at Instagram. The"
  },
  {
    "startMs": 949760,
    "text": "Instagram we what"
  },
  {
    "startMs": 951880,
    "text": "things that had like four to five"
  },
  {
    "startMs": 953160,
    "text": "percent usage you're like oh that's"
  },
  {
    "startMs": 955160,
    "text": "really not very many but then you have"
  },
  {
    "startMs": 957000,
    "text": "like 20 features that each have four to"
  },
  {
    "startMs": 958440,
    "text": "five percent usage is like the classic"
  },
  {
    "startMs": 960040,
    "text": "Microsoft Word problem of like everybody"
  },
  {
    "startMs": 962800,
    "text": "uses some disjoint subset of the of the"
  },
  {
    "startMs": 965440,
    "text": "functionality. So that that's always the"
  },
  {
    "startMs": 967280,
    "text": "challenge. Now I think we're younger"
  },
  {
    "startMs": 969760,
    "text": "product so hopefully we have less of"
  },
  {
    "startMs": 971040,
    "text": "those things. Like we unship styles I"
  },
  {
    "startMs": 972840,
    "text": "think recently where it was like used by"
  },
  {
    "startMs": 974880,
    "text": "a small percentage of people and was not"
  },
  {
    "startMs": 977360,
    "text": "really AGI filled in a lot of ways it"
  },
  {
    "startMs": 979040,
    "text": "was like very sort of prescriptive in"
  },
  {
    "startMs": 980440,
    "text": "the way that it worked and skills very"
  },
  {
    "startMs": 981960,
    "text": "much better applications and then like"
  },
  {
    "startMs": 984000,
    "text": "that. So I think you have to be willing"
  },
  {
    "startMs": 985400,
    "text": "to take the primitives of like one"
  },
  {
    "startMs": 987720,
    "text": "generation of AI and like unship them or"
  },
  {
    "startMs": 989960,
    "text": "at least like supplement them or"
  },
  {
    "startMs": 991839,
    "text": "supplant them with the next one as well."
  },
  {
    "startMs": 993920,
    "text": "I think the biggest thing is I look at"
  },
  {
    "startMs": 995680,
    "text": "it and I've been spending some time like"
  },
  {
    "startMs": 997080,
    "text": "outside labs on some of this is like man"
  },
  {
    "startMs": 999320,
    "text": "like we're asking people to make like"
  },
  {
    "startMs": 1000920,
    "text": "code versus co-work versus like chat"
  },
  {
    "startMs": 1003800,
    "text": "distinctions and like one they don't"
  },
  {
    "startMs": 1005280,
    "text": "interoperate well and they can't"
  },
  {
    "startMs": 1006440,
    "text": "delegate to each other and two I think"
  },
  {
    "startMs": 1008200,
    "text": "the average person off the street could"
  },
  {
    "startMs": 1009520,
    "text": "not explain to you why those are all"
  },
  {
    "startMs": 1011960,
    "text": "different. So I think deleting some of"
  },
  {
    "startMs": 1013720,
    "text": "the product complexity within our our"
  },
  {
    "startMs": 1016120,
    "text": "code or our product I think is a a thing"
  },
  {
    "startMs": 1018600,
    "text": "that would would serve well. Also"
  },
  {
    "startMs": 1020120,
    "text": "because then Claude can do what it needs"
  },
  {
    "startMs": 1022360,
    "text": "to do and and do well. Like there's"
  },
  {
    "startMs": 1023440,
    "text": "nothing more frustrating than having a"
  },
  {
    "startMs": 1024800,
    "text": "co-work session where you're like great"
  },
  {
    "startMs": 1026520,
    "text": "I've mapped out exactly what I want you"
  },
  {
    "startMs": 1027920,
    "text": "to build and then be like can you please"
  },
  {
    "startMs": 1029920,
    "text": "like create a paragraph that I can paste"
  },
  {
    "startMs": 1031560,
    "text": "into Claude code? Like that is some 2020"
  },
  {
    "startMs": 1033880,
    "text": "you know kind of workflow there that"
  },
  {
    "startMs": 1035720,
    "text": "really shouldn't exist anymore."
  },
  {
    "startMs": 1037400,
    "text": ">> Yeah. Um"
  },
  {
    "startMs": 1038839,
    "text": "I think"
  },
  {
    "startMs": 1039720,
    "text": "drawing lines on what you don't want to"
  },
  {
    "startMs": 1041199,
    "text": "do and also sort of leaving room for"
  },
  {
    "startMs": 1043040,
    "text": "others is interesting. Um a lot of"
  },
  {
    "startMs": 1044600,
    "text": "people today is like the startups day"
  },
  {
    "startMs": 1046560,
    "text": "for AI E or obviously very"
  },
  {
    "startMs": 1048120,
    "text": "sympathetically aligned to startups. Uh"
  },
  {
    "startMs": 1050040,
    "text": "but there's some anxiety in the room"
  },
  {
    "startMs": 1051440,
    "text": "because tomorrow's Anthropic could wake"
  },
  {
    "startMs": 1053840,
    "text": "up and publish"
  },
  {
    "startMs": 1055200,
    "text": "some markdown files that destroy my"
  },
  {
    "startMs": 1056680,
    "text": "industry. Um so"
  },
  {
    "startMs": 1058492,
    "text": ">> [laughter]"
  },
  {
    "startMs": 1059080,
    "text": ">> uh why should we not all just give up"
  },
  {
    "startMs": 1060840,
    "text": "and join Anthropic? Like why bother"
  },
  {
    "startMs": 1062080,
    "text": "starting any other company?"
  },
  {
    "startMs": 1064600,
    "text": ">> Um I I mean actually joined one of the"
  },
  {
    "startMs": 1067040,
    "text": "main reasons I joined Anthropic was"
  },
  {
    "startMs": 1068800,
    "text": "because I saw how much this was like,"
  },
  {
    "startMs": 1071240,
    "text": "you know, the models weren't that good"
  },
  {
    "startMs": 1072320,
    "text": "at coding up but they were getting"
  },
  {
    "startMs": 1073320,
    "text": "there. Like how much it would unlock"
  },
  {
    "startMs": 1075240,
    "text": "like whole like next generation of"
  },
  {
    "startMs": 1076720,
    "text": "startups. Not because it was going to"
  },
  {
    "startMs": 1078120,
    "text": "solve their ideation or their taste, but"
  },
  {
    "startMs": 1079880,
    "text": "because like it would make"
  },
  {
    "startMs": 1081440,
    "text": "experimentation way simpler and and"
  },
  {
    "startMs": 1083760,
    "text": "would get you to move faster. And I"
  },
  {
    "startMs": 1084880,
    "text": "still like really believe that. And I"
  },
  {
    "startMs": 1086840,
    "text": "mean it's the reality of, you know,"
  },
  {
    "startMs": 1089480,
    "text": "uh"
  },
  {
    "startMs": 1090800,
    "text": "And we saw this with like Instagram."
  },
  {
    "startMs": 1092040,
    "text": "Like we would get questions from"
  },
  {
    "startMs": 1092960,
    "text": "investors like, well, what happens when"
  },
  {
    "startMs": 1094720,
    "text": "Google launches a photos product? It's"
  },
  {
    "startMs": 1096360,
    "text": "like Google's going to launch a very"
  },
  {
    "startMs": 1097400,
    "text": "googly photos product and it's going to"
  },
  {
    "startMs": 1099160,
    "text": "have to be bound by the integrations"
  },
  {
    "startMs": 1100600,
    "text": "that they already have and it's going to"
  },
  {
    "startMs": 1102040,
    "text": "be like it's going to play to their"
  },
  {
    "startMs": 1103600,
    "text": "strengths. And I think that is going to"
  },
  {
    "startMs": 1104760,
    "text": "be true. And I'm not like giving advice"
  },
  {
    "startMs": 1106200,
    "text": "on how to compete with Anthropic, I"
  },
  {
    "startMs": 1107240,
    "text": "guess in a way, but like it's actually"
  },
  {
    "startMs": 1108360,
    "text": "not because we're also a platform which"
  },
  {
    "startMs": 1109760,
    "text": "is like there's so much I think room to"
  },
  {
    "startMs": 1112000,
    "text": "be like laser obsessed with your"
  },
  {
    "startMs": 1114360,
    "text": "particular vertical or your industry or"
  },
  {
    "startMs": 1116640,
    "text": "group of people that you know really"
  },
  {
    "startMs": 1117720,
    "text": "well in a way that like none of the labs"
  },
  {
    "startMs": 1119400,
    "text": "are ever going to get to that level of"
  },
  {
    "startMs": 1121680,
    "text": "uh of understanding and like therefore"
  },
  {
    "startMs": 1123600,
    "text": "get that kind of adoption and user love"
  },
  {
    "startMs": 1126360,
    "text": "and and build that out. Now, it's"
  },
  {
    "startMs": 1127880,
    "text": "definitely harder in the age where like"
  },
  {
    "startMs": 1129880,
    "text": "the models can just do a lot and so"
  },
  {
    "startMs": 1131280,
    "text": "there's, you know, some of these things"
  },
  {
    "startMs": 1133160,
    "text": "can be like skillified and like maybe"
  },
  {
    "startMs": 1135160,
    "text": "don't need their own dedicated product."
  },
  {
    "startMs": 1137080,
    "text": "But I think it's like the hard stuff is"
  },
  {
    "startMs": 1138360,
    "text": "still hard. It's like understanding the"
  },
  {
    "startMs": 1139760,
    "text": "needs of people, like figuring out how"
  },
  {
    "startMs": 1141320,
    "text": "you're going to reach them, uh listening"
  },
  {
    "startMs": 1143280,
    "text": "to them and iterating on them really"
  },
  {
    "startMs": 1144360,
    "text": "quickly. Like"
  },
  {
    "startMs": 1145680,
    "text": "it is still the case that like a group"
  },
  {
    "startMs": 1147520,
    "text": "of four or five people obsessed with a"
  },
  {
    "startMs": 1149640,
    "text": "problem is going to move faster than"
  },
  {
    "startMs": 1151400,
    "text": "those same people at any other kind of"
  },
  {
    "startMs": 1153400,
    "text": "organization that are like, you know,"
  },
  {
    "startMs": 1154560,
    "text": "subject just to the complexity. I just"
  },
  {
    "startMs": 1155960,
    "text": "mentioned the like the fact that we"
  },
  {
    "startMs": 1157280,
    "text": "have, you know, a lot of different"
  },
  {
    "startMs": 1158840,
    "text": "products that kind of interoperate. Like"
  },
  {
    "startMs": 1160080,
    "text": "that's a interesting constraint that we"
  },
  {
    "startMs": 1161560,
    "text": "have to work through. It's an advantage"
  },
  {
    "startMs": 1162680,
    "text": "in other ways, right? So, yeah, I'm"
  },
  {
    "startMs": 1164560,
    "text": "still like very long and bullish on"
  },
  {
    "startMs": 1166120,
    "text": "startups and um it's just"
  },
  {
    "startMs": 1169480,
    "text": "it tapers over the fact that like"
  },
  {
    "startMs": 1171240,
    "text": "writing code was never the like the"
  },
  {
    "startMs": 1173080,
    "text": "limiting part. You know, maybe it was on"
  },
  {
    "startMs": 1174680,
    "text": "the timeline perspective, but it was"
  },
  {
    "startMs": 1175960,
    "text": "never like the thing that was going to"
  },
  {
    "startMs": 1177120,
    "text": "like make or break a startup. It's"
  },
  {
    "startMs": 1178840,
    "text": "really that space and user"
  },
  {
    "startMs": 1180320,
    "text": "understanding."
  },
  {
    "startMs": 1181160,
    "text": ">> Yeah. Uh domain knowledge."
  },
  {
    "startMs": 1183080,
    "text": ">> Yeah."
  },
  {
    "startMs": 1183640,
    "text": ">> Uh today is also our day for vertical"
  },
  {
    "startMs": 1186320,
    "text": "AI. Uh one of our"
  },
  {
    "startMs": 1188560,
    "text": "uh returning speakers and top speakers,"
  },
  {
    "startMs": 1190040,
    "text": "Chris Lovejoy, uh was always talking"
  },
  {
    "startMs": 1191960,
    "text": "about vertical AI. He was in from"
  },
  {
    "startMs": 1193840,
    "text": "interior in the healthcare space. And"
  },
  {
    "startMs": 1195880,
    "text": "then recently I was I invited him back"
  },
  {
    "startMs": 1197600,
    "text": "and turned out he you guys just hired"
  },
  {
    "startMs": 1199360,
    "text": "him for your uh healthcare efforts. Um"
  },
  {
    "startMs": 1202280,
    "text": "we also our next big one is also"
  },
  {
    "startMs": 1203720,
    "text": "finance. You know, we have a yeah, a"
  },
  {
    "startMs": 1205960,
    "text": "finance track. You guys just had a huge"
  },
  {
    "startMs": 1207840,
    "text": "finance event in New York City. Um and"
  },
  {
    "startMs": 1210160,
    "text": "where our next uh AI is is sort of"
  },
  {
    "startMs": 1212320,
    "text": "finance focused. What are you seeing"
  },
  {
    "startMs": 1213640,
    "text": "there? Any you know, any potential uh"
  },
  {
    "startMs": 1215680,
    "text": "for Claude? Obviously a lot of Excel"
  },
  {
    "startMs": 1217160,
    "text": "Excel spreadsheets."
  },
  {
    "startMs": 1218160,
    "text": ">> Yeah. No, I think that there's there's a"
  },
  {
    "startMs": 1219560,
    "text": "lot in there, too. And that's like an"
  },
  {
    "startMs": 1221120,
    "text": "area where uh you could see the model"
  },
  {
    "startMs": 1223520,
    "text": "get clearly better at it like sort of"
  },
  {
    "startMs": 1226200,
    "text": "generation to generation. And there's,"
  },
  {
    "startMs": 1228160,
    "text": "you know, there's some good sort of"
  },
  {
    "startMs": 1229560,
    "text": "vertical specific uh uh finance startups"
  },
  {
    "startMs": 1232360,
    "text": "that have like done their own um"
  },
  {
    "startMs": 1234880,
    "text": "evals, which has also been interesting"
  },
  {
    "startMs": 1236040,
    "text": "to to track. And it's not like we're"
  },
  {
    "startMs": 1237280,
    "text": "like sort of playing to the eval, but it"
  },
  {
    "startMs": 1238840,
    "text": "is a useful sort of barometer around"
  },
  {
    "startMs": 1240240,
    "text": "like is this actually getting better um"
  },
  {
    "startMs": 1242200,
    "text": "at these finance use cases. I think the"
  },
  {
    "startMs": 1244080,
    "text": "interesting blend that's going to happen"
  },
  {
    "startMs": 1245920,
    "text": "um is this mix of, again, the model"
  },
  {
    "startMs": 1248680,
    "text": "having the flexibility to like dive in"
  },
  {
    "startMs": 1250320,
    "text": "and create just-in-time analyses or"
  },
  {
    "startMs": 1252640,
    "text": "dashboards or workflows with like some"
  },
  {
    "startMs": 1254720,
    "text": "sense of like what is the not immutable,"
  },
  {
    "startMs": 1257400,
    "text": "but at least like verified sort of set"
  },
  {
    "startMs": 1259480,
    "text": "of data. And so like uh"
  },
  {
    "startMs": 1261640,
    "text": "set having all of that be totally free"
  },
  {
    "startMs": 1263840,
    "text": "form, I think is a recipe for confusion"
  },
  {
    "startMs": 1265840,
    "text": "and is like not what most companies in"
  },
  {
    "startMs": 1267840,
    "text": "the financial services space want. So,"
  },
  {
    "startMs": 1270000,
    "text": "finding that right uh sort of cut line"
  },
  {
    "startMs": 1272080,
    "text": "where you have verifiability and audit"
  },
  {
    "startMs": 1274800,
    "text": "logging and and sort of data provenance"
  },
  {
    "startMs": 1276720,
    "text": "here, but not in a way that constrains"
  },
  {
    "startMs": 1278760,
    "text": "the kinds of applications that you can"
  },
  {
    "startMs": 1280160,
    "text": "build on top, I think is a lot of the"
  },
  {
    "startMs": 1281880,
    "text": "art that we're seeing in that space as"
  },
  {
    "startMs": 1284000,
    "text": "well. Um and I think, you know, if you"
  },
  {
    "startMs": 1286360,
    "text": "solve it well, you can you can get the"
  },
  {
    "startMs": 1288320,
    "text": "best of both worlds. The hard part is a"
  },
  {
    "startMs": 1290680,
    "text": "lot of the systems that were built to do"
  },
  {
    "startMs": 1292320,
    "text": "the verifiability audibility like are"
  },
  {
    "startMs": 1294080,
    "text": "kind of almost by design not super"
  },
  {
    "startMs": 1296360,
    "text": "flexible in terms of agentic workloads"
  },
  {
    "startMs": 1298200,
    "text": "on top. So, I think there's opportunity"
  },
  {
    "startMs": 1299800,
    "text": "at both sides of the stack there."
  },
  {
    "startMs": 1301320,
    "text": ">> Yeah. Um I think I I also agree we'll be"
  },
  {
    "startMs": 1303720,
    "text": "exploring that in in New York. Um the"
  },
  {
    "startMs": 1306080,
    "text": "last thing I want to end on is on mental"
  },
  {
    "startMs": 1308040,
    "text": "health, which we don't talk about enough"
  },
  {
    "startMs": 1310040,
    "text": "in technical conference conferences. Um"
  },
  {
    "startMs": 1312520,
    "text": "you've seen a lot of hyper growth."
  },
  {
    "startMs": 1314240,
    "text": "People are just always refreshing their"
  },
  {
    "startMs": 1316080,
    "text": "timelines and it's exhausting."
  },
  {
    "startMs": 1318800,
    "text": "Um"
  },
  {
    "startMs": 1319480,
    "text": "how do you advise people who are working"
  },
  {
    "startMs": 1321120,
    "text": "996 to avoid burnout?"
  },
  {
    "startMs": 1323240,
    "text": ">> Yeah. I mean, I think this is a hard"
  },
  {
    "startMs": 1324800,
    "text": "one. I mean, and"
  },
  {
    "startMs": 1326680,
    "text": "it is, I'm sure you are experiencing"
  },
  {
    "startMs": 1328240,
    "text": "this cuz you're all working in this"
  },
  {
    "startMs": 1329280,
    "text": "industry like"
  },
  {
    "startMs": 1330680,
    "text": "it is"
  },
  {
    "startMs": 1331800,
    "text": "you know, multiples more intense and"
  },
  {
    "startMs": 1334120,
    "text": "things move much more quickly at an"
  },
  {
    "startMs": 1335400,
    "text": "Instagram like our the two things that"
  },
  {
    "startMs": 1337440,
    "text": "we were thinking about was like, what is"
  },
  {
    "startMs": 1338920,
    "text": "Apple going to announce at WWDC and is"
  },
  {
    "startMs": 1340600,
    "text": "it going to like totally mess us up or"
  },
  {
    "startMs": 1342200,
    "text": "boost us, right? So, that's like once a"
  },
  {
    "startMs": 1343840,
    "text": "year. Um or, you know, maybe a"
  },
  {
    "startMs": 1345440,
    "text": "competitor launches every three or four"
  },
  {
    "startMs": 1347080,
    "text": "months, right? And uh it is definitely"
  },
  {
    "startMs": 1349360,
    "text": "not that. It's"
  },
  {
    "startMs": 1350560,
    "text": "a topic we we do a when we do do our"
  },
  {
    "startMs": 1352520,
    "text": "weekly all hands. Usually on Wednesdays"
  },
  {
    "startMs": 1354640,
    "text": "and we have a slide that's like the week"
  },
  {
    "startMs": 1355800,
    "text": "in AI at Pinterest is and it's only"
  },
  {
    "startMs": 1357440,
    "text": "Wednesday and like and inevitably like"
  },
  {
    "startMs": 1360040,
    "text": "some competitor has shipped a new model"
  },
  {
    "startMs": 1361720,
    "text": "and like there's been a like new product"
  },
  {
    "startMs": 1363640,
    "text": "and maybe there's some interesting thing"
  },
  {
    "startMs": 1364760,
    "text": "happening"
  },
  {
    "startMs": 1365840,
    "text": "um uh on the regulation side. Like if"
  },
  {
    "startMs": 1367480,
    "text": "things are moving really, really"
  },
  {
    "startMs": 1368520,
    "text": "quickly. I think the way I try to stay"
  },
  {
    "startMs": 1371200,
    "text": "at least relatively sane, um one is like"
  },
  {
    "startMs": 1374120,
    "text": "actually carving time off and I think"
  },
  {
    "startMs": 1375680,
    "text": "the topic co-founders do a good job of"
  },
  {
    "startMs": 1377480,
    "text": "like saying like, look, like burnout if"
  },
  {
    "startMs": 1379920,
    "text": "you you out, like you're kind of done."
  },
  {
    "startMs": 1381320,
    "text": "I've seen it happen unfortunately to"
  },
  {
    "startMs": 1382640,
    "text": "people I'm really close to and then it"
  },
  {
    "startMs": 1384160,
    "text": "takes a long time to recover from that."
  },
  {
    "startMs": 1386280,
    "text": "Um so actually encouraging people like"
  },
  {
    "startMs": 1388040,
    "text": "there's no job that is so important that"
  },
  {
    "startMs": 1389560,
    "text": "you can't be offline for a couple of"
  },
  {
    "startMs": 1390960,
    "text": "days. Um so I think that's like a big"
  },
  {
    "startMs": 1393240,
    "text": "key like piece in there. So like let's"
  },
  {
    "startMs": 1396476,
    "text": ">> [applause]"
  },
  {
    "startMs": 1396720,
    "text": ">> strongly believe."
  },
  {
    "startMs": 1398280,
    "text": "Um"
  },
  {
    "startMs": 1399520,
    "text": "And if it is you're probably doing"
  },
  {
    "startMs": 1400560,
    "text": "something wrong and you talk to somebody"
  },
  {
    "startMs": 1402480,
    "text": "who could be a mentor to figure out how"
  },
  {
    "startMs": 1403760,
    "text": "you can unblock that. Um and then I"
  },
  {
    "startMs": 1406320,
    "text": "think the other one as well is like I"
  },
  {
    "startMs": 1408840,
    "text": "love sports and like uh"
  },
  {
    "startMs": 1411080,
    "text": "this is the notion of like you're never"
  },
  {
    "startMs": 1412240,
    "text": "as good as like your best game and"
  },
  {
    "startMs": 1413400,
    "text": "you're never as bad as your worst game."
  },
  {
    "startMs": 1414520,
    "text": "I think that's also really true. Like I"
  },
  {
    "startMs": 1416400,
    "text": "know like in AI there's like the you"
  },
  {
    "startMs": 1418080,
    "text": "know it's so over we're so back thing."
  },
  {
    "startMs": 1419920,
    "text": "Like that like if you internalize that"
  },
  {
    "startMs": 1422000,
    "text": "that cycle is always going to be at play"
  },
  {
    "startMs": 1424240,
    "text": "in some way, you realize like it's never"
  },
  {
    "startMs": 1426480,
    "text": "that bad. Like Ben Horowitz's book is"
  },
  {
    "startMs": 1429080,
    "text": "the hard thing about hard things has"
  },
  {
    "startMs": 1430840,
    "text": "this chapter on like we're effed it's"
  },
  {
    "startMs": 1433200,
    "text": "over and like that feeling as a startup"
  },
  {
    "startMs": 1434720,
    "text": "that probably many of you have had at"
  },
  {
    "startMs": 1435760,
    "text": "startups where you're like oh I can't"
  },
  {
    "startMs": 1437200,
    "text": "believe this thing happened like we're"
  },
  {
    "startMs": 1438160,
    "text": "never going to like recover from this. I"
  },
  {
    "startMs": 1439640,
    "text": "definitely we definitely had an"
  },
  {
    "startMs": 1440520,
    "text": "Instagram a couple of times. And then"
  },
  {
    "startMs": 1442200,
    "text": "you get through it and like it's that"
  },
  {
    "startMs": 1443960,
    "text": "like def like defines the company when"
  },
  {
    "startMs": 1445720,
    "text": "you can actually go through that. I try"
  },
  {
    "startMs": 1447280,
    "text": "to remind myself and the team here even"
  },
  {
    "startMs": 1449480,
    "text": "with an Entropic which is like like this"
  },
  {
    "startMs": 1451240,
    "text": "is a is a fast-moving but is also a long"
  },
  {
    "startMs": 1453800,
    "text": "game. And it's like we're never it's"
  },
  {
    "startMs": 1455360,
    "text": "never just about today's model launch"
  },
  {
    "startMs": 1458120,
    "text": "and reaction or this product launch and"
  },
  {
    "startMs": 1459840,
    "text": "everything else. Like you're playing and"
  },
  {
    "startMs": 1461280,
    "text": "you're building and you just have to"
  },
  {
    "startMs": 1462320,
    "text": "trust that you're building like the team"
  },
  {
    "startMs": 1463680,
    "text": "and culture that is going to get through"
  },
  {
    "startMs": 1465440,
    "text": "those things and have that sense of"
  },
  {
    "startMs": 1467120,
    "text": "perspective even if perspective is"
  },
  {
    "startMs": 1468600,
    "text": "saying like look 3 months ago we were in"
  },
  {
    "startMs": 1470520,
    "text": "a similar position. Maybe it's not a"
  },
  {
    "startMs": 1471920,
    "text": "year it's just a matter of months but"
  },
  {
    "startMs": 1473440,
    "text": "it's still like zooming out and not"
  },
  {
    "startMs": 1475720,
    "text": "taking things not letting your internal"
  },
  {
    "startMs": 1478200,
    "text": "sort of like sense of self and success"
  },
  {
    "startMs": 1481000,
    "text": "be so driven by the day-to-day."
  },
  {
    "startMs": 1483120,
    "text": ">> Yeah. It has anyone any coach or mentor"
  },
  {
    "startMs": 1486000,
    "text": "said something to you that you repeat to"
  },
  {
    "startMs": 1487800,
    "text": "yourself that gets you through the hard"
  },
  {
    "startMs": 1489920,
    "text": "tough times?"
  },
  {
    "startMs": 1490520,
    "text": ">> Um I think the biggest one was"
  },
  {
    "startMs": 1493400,
    "text": "this like sense of like if you're"
  },
  {
    "startMs": 1496160,
    "text": "feeling something it's really often the"
  },
  {
    "startMs": 1497880,
    "text": "case that other people on the team are"
  },
  {
    "startMs": 1499160,
    "text": "feeling it too. So this is like advice I"
  },
  {
    "startMs": 1501280,
    "text": "got from my my coach around just being"
  },
  {
    "startMs": 1503120,
    "text": "like like just verbalizing emotions like"
  },
  {
    "startMs": 1505360,
    "text": "even saying like hey I'm feeling really"
  },
  {
    "startMs": 1507000,
    "text": "stressed out about this or yeah I'm"
  },
  {
    "startMs": 1508400,
    "text": "really sad that we are shutting down"
  },
  {
    "startMs": 1510160,
    "text": "this labs initiative. I literally had"
  },
  {
    "startMs": 1511440,
    "text": "this meeting a couple months ago where I"
  },
  {
    "startMs": 1512640,
    "text": "was working really hard on something and"
  },
  {
    "startMs": 1514280,
    "text": "I kicked off the meeting like"
  },
  {
    "startMs": 1516080,
    "text": "I'll kick it off like I'm really sad"
  },
  {
    "startMs": 1517880,
    "text": "like I'm frustrated like I wish this"
  },
  {
    "startMs": 1519360,
    "text": "thing had worked out and I think that"
  },
  {
    "startMs": 1520840,
    "text": "holds the space for other people to be"
  },
  {
    "startMs": 1522120,
    "text": "like yeah I'm pissed off too or like I'm"
  },
  {
    "startMs": 1523760,
    "text": "sad too and like I think giving that"
  },
  {
    "startMs": 1525800,
    "text": "advice around like not"
  },
  {
    "startMs": 1527920,
    "text": "I think if you can get yourself to be"
  },
  {
    "startMs": 1530480,
    "text": "open and vulnerable it often like lets"
  },
  {
    "startMs": 1532000,
    "text": "other people verbalize that and then you"
  },
  {
    "startMs": 1533200,
    "text": "can from there you can be like great"
  },
  {
    "startMs": 1534240,
    "text": "what are we going to do about it like"
  },
  {
    "startMs": 1535480,
    "text": "you know it's much easier to start from"
  },
  {
    "startMs": 1536880,
    "text": "that place."
  },
  {
    "startMs": 1537720,
    "text": ">> Yeah we actually kicked off AIE with a"
  },
  {
    "startMs": 1539960,
    "text": "session from Carol Robbins who runs"
  },
  {
    "startMs": 1541680,
    "text": "touchy feely at Stanford"
  },
  {
    "startMs": 1543400,
    "text": "and I can't think of a better way to end"
  },
  {
    "startMs": 1545320,
    "text": "than encouraging people to talk about"
  },
  {
    "startMs": 1546680,
    "text": "their feelings manage their mental"
  },
  {
    "startMs": 1548120,
    "text": "health and keep shipping."
  },
  {
    "startMs": 1549480,
    "text": ">> Yeah. Thanks so much Mike."
  },
  {
    "startMs": 1550520,
    "text": ">> Thanks for having me."
  }
]