[
  {
    "startMs": 1309,
    "text": "[music]"
  },
  {
    "startMs": 12719,
    "text": "Hi everyone, I'm Ben Kuss. I'm CTO of"
  },
  {
    "startMs": 15120,
    "text": "Box and today I'm going to be talking"
  },
  {
    "startMs": 16880,
    "text": "about uh building for change and"
  },
  {
    "startMs": 19680,
    "text": "specifically around uh AI agents and how"
  },
  {
    "startMs": 22480,
    "text": "to continue to adapt uh your"
  },
  {
    "startMs": 24880,
    "text": "infrastructure as we are all in the"
  },
  {
    "startMs": 26720,
    "text": "middle of this journey."
  },
  {
    "startMs": 28800,
    "text": "Um, so before I get too far, I will"
  },
  {
    "startMs": 30960,
    "text": "quickly sort of set a little bit of like"
  },
  {
    "startMs": 32880,
    "text": "who I am and uh sort of what I do. Um,"
  },
  {
    "startMs": 36239,
    "text": "for Box, I'm CTO and one of my jobs in"
  },
  {
    "startMs": 39040,
    "text": "my job for my whole career has been to"
  },
  {
    "startMs": 41040,
    "text": "build enterprise software. And so if"
  },
  {
    "startMs": 43440,
    "text": "today um you're from a consumer company"
  },
  {
    "startMs": 46320,
    "text": "or you're uh not involved in enterprise,"
  },
  {
    "startMs": 48640,
    "text": "I hope that a lot of it is still"
  },
  {
    "startMs": 49840,
    "text": "relevant. But in many cases um a lot of"
  },
  {
    "startMs": 52160,
    "text": "the lessons I've learned are enterprise"
  },
  {
    "startMs": 53840,
    "text": "uh specific."
  },
  {
    "startMs": 56160,
    "text": "So um I sort of will highlight um when"
  },
  {
    "startMs": 60399,
    "text": "I'm thinking and talking about"
  },
  {
    "startMs": 61840,
    "text": "infrastructure when I'm talking about uh"
  },
  {
    "startMs": 63840,
    "text": "the kind of challenges that we face um"
  },
  {
    "startMs": 65760,
    "text": "I'm typically talking about things that"
  },
  {
    "startMs": 67200,
    "text": "are sort of in a scale of like uh like"
  },
  {
    "startMs": 69280,
    "text": "for box we have over an exabyte of data"
  },
  {
    "startMs": 72880,
    "text": "not a gigabyte not a terabyte not a"
  },
  {
    "startMs": 74240,
    "text": "pabyte but an exabyte um and then"
  },
  {
    "startMs": 76400,
    "text": "oftentimes we're I'm thinking in the"
  },
  {
    "startMs": 77840,
    "text": "tens of of millions of users uh the"
  },
  {
    "startMs": 80880,
    "text": "hundreds of billions of things in our"
  },
  {
    "startMs": 82400,
    "text": "case files or content or unstructured"
  },
  {
    "startMs": 84240,
    "text": "content um and then the new stat that is"
  },
  {
    "startMs": 86799,
    "text": "sort of the the one that we talk about"
  },
  {
    "startMs": 88720,
    "text": "is uh tokens. Um so we are now in the"
  },
  {
    "startMs": 91600,
    "text": "ballpark of trillion tokens probably"
  },
  {
    "startMs": 93920,
    "text": "will be 10 trillion tokens sometime"
  },
  {
    "startMs": 95360,
    "text": "soon. Um and this of course is uh uh"
  },
  {
    "startMs": 98079,
    "text": "some of the new and interesting"
  },
  {
    "startMs": 99360,
    "text": "challenges that this kind of scale"
  },
  {
    "startMs": 100799,
    "text": "brings."
  },
  {
    "startMs": 103040,
    "text": "Um so in my career and um I think maybe"
  },
  {
    "startMs": 106720,
    "text": "many of us here um we've kind of lived"
  },
  {
    "startMs": 109119,
    "text": "through these this these technology"
  },
  {
    "startMs": 110640,
    "text": "changes. And so taking a quick step back"
  },
  {
    "startMs": 112960,
    "text": "um I started uh my career when the"
  },
  {
    "startMs": 115680,
    "text": "internet was sort of becoming a thing uh"
  },
  {
    "startMs": 118560,
    "text": "lived through mobile and sort of this"
  },
  {
    "startMs": 120320,
    "text": "idea of like you know carrying these"
  },
  {
    "startMs": 122159,
    "text": "different devices move to the cloud"
  },
  {
    "startMs": 123840,
    "text": "where you could kind of store and"
  },
  {
    "startMs": 124960,
    "text": "maintain all your data and then of"
  },
  {
    "startMs": 126320,
    "text": "course we're all in the middle of this"
  },
  {
    "startMs": 127200,
    "text": "AI change and I think when you see these"
  },
  {
    "startMs": 129280,
    "text": "kind of technology disruptions when"
  },
  {
    "startMs": 131360,
    "text": "you're sort of thinking about this idea"
  },
  {
    "startMs": 133200,
    "text": "of like all of these kind of have"
  },
  {
    "startMs": 134480,
    "text": "changed all of our lives um and then uh"
  },
  {
    "startMs": 137599,
    "text": "and you're thinking about it from the"
  },
  {
    "startMs": 138800,
    "text": "perspective of a technology leader or a"
  },
  {
    "startMs": 141840,
    "text": "startup or a engineer. Um, you kind of"
  },
  {
    "startMs": 144959,
    "text": "see that like these are where like major"
  },
  {
    "startMs": 147440,
    "text": "companies are born. Um, you see that"
  },
  {
    "startMs": 149440,
    "text": "like big companies adapt or die. Um,"
  },
  {
    "startMs": 152400,
    "text": "small companies are here to disrupt"
  },
  {
    "startMs": 153599,
    "text": "things. They're here to take bets."
  },
  {
    "startMs": 154720,
    "text": "They're here to grow. Um, I've had two"
  },
  {
    "startMs": 156560,
    "text": "startups. I've been acquired twice. Once"
  },
  {
    "startMs": 158239,
    "text": "in IBM, once in a box. Um, and um, so"
  },
  {
    "startMs": 161360,
    "text": "that we're in the middle of this kind of"
  },
  {
    "startMs": 162640,
    "text": "major opportunity."
  },
  {
    "startMs": 164800,
    "text": "Um but for um a long tu uh no matter"
  },
  {
    "startMs": 168319,
    "text": "what you kind of come from and what area"
  },
  {
    "startMs": 170560,
    "text": "you're at I uh t typically if you were"
  },
  {
    "startMs": 172800,
    "text": "to ask my advice um and about kind of"
  },
  {
    "startMs": 175200,
    "text": "what makes it you successful as an"
  },
  {
    "startMs": 177599,
    "text": "company as an engineering organization"
  },
  {
    "startMs": 179360,
    "text": "as a technology startup or as a like a"
  },
  {
    "startMs": 182640,
    "text": "company who has a a technology division"
  },
  {
    "startMs": 185599,
    "text": "um I would say no matter what there's"
  },
  {
    "startMs": 187599,
    "text": "kind of three things the first is you"
  },
  {
    "startMs": 190000,
    "text": "need to build scalable reliable uh"
  },
  {
    "startMs": 192000,
    "text": "platforms and select the technology that"
  },
  {
    "startMs": 194400,
    "text": "you care about. Meaning that like"
  },
  {
    "startMs": 196400,
    "text": "there's a lot of ways to do things, but"
  },
  {
    "startMs": 198000,
    "text": "get good at something. Get good at that"
  },
  {
    "startMs": 199760,
    "text": "technology, at that system, at that"
  },
  {
    "startMs": 201040,
    "text": "stack, and then and then keep going with"
  },
  {
    "startMs": 202720,
    "text": "that. Um, and then you leverage this"
  },
  {
    "startMs": 205200,
    "text": "technology so that you do more for your"
  },
  {
    "startMs": 207040,
    "text": "customers. Um, you build your better"
  },
  {
    "startMs": 209200,
    "text": "product, you develop the capabilities,"
  },
  {
    "startMs": 210959,
    "text": "and then you optimize it, make it"
  },
  {
    "startMs": 212640,
    "text": "better, make it faster, make it cheaper,"
  },
  {
    "startMs": 215360,
    "text": "make it uh more capable. And this was"
  },
  {
    "startMs": 218000,
    "text": "sort of the generic enterprise advice,"
  },
  {
    "startMs": 220959,
    "text": "uh, the generic energy advice that many"
  },
  {
    "startMs": 222799,
    "text": "many people would follow. And I think"
  },
  {
    "startMs": 224400,
    "text": "this works really well"
  },
  {
    "startMs": 227120,
    "text": "except now"
  },
  {
    "startMs": 230239,
    "text": "I don't know if this is good advice. It"
  },
  {
    "startMs": 232319,
    "text": "has been across all these major"
  },
  {
    "startMs": 234080,
    "text": "disruptive changes over time. Unclear."
  },
  {
    "startMs": 236640,
    "text": "In fact, I don't think it's good advice"
  },
  {
    "startMs": 238000,
    "text": "right now because there's a funny thing"
  },
  {
    "startMs": 240560,
    "text": "happening right now which that didn't"
  },
  {
    "startMs": 242159,
    "text": "happen in those previous trends which is"
  },
  {
    "startMs": 243680,
    "text": "that the rate of change is dramatically"
  },
  {
    "startMs": 245840,
    "text": "higher. Um I and you might say look"
  },
  {
    "startMs": 248879,
    "text": "technology always is changing like you"
  },
  {
    "startMs": 250640,
    "text": "know there's other trends things change"
  },
  {
    "startMs": 251920,
    "text": "a lot but not that much um the internet"
  },
  {
    "startMs": 255040,
    "text": "is still based on HTTP the uh uh mobile"
  },
  {
    "startMs": 258400,
    "text": "devices are still um iOS and Android"
  },
  {
    "startMs": 260560,
    "text": "based and so on and so but nowadays um"
  },
  {
    "startMs": 263120,
    "text": "other than the fact that like uh"
  },
  {
    "startMs": 264720,
    "text": "generative AI exists most things that"
  },
  {
    "startMs": 266639,
    "text": "power it are changing and changing"
  },
  {
    "startMs": 268080,
    "text": "dramatically."
  },
  {
    "startMs": 270800,
    "text": "So last year I was here at this uh at"
  },
  {
    "startMs": 273280,
    "text": "the AI engineering world fair and I gave"
  },
  {
    "startMs": 274800,
    "text": "a speech and I said uh after spending a"
  },
  {
    "startMs": 278720,
    "text": "lot of time on this and thinking through"
  },
  {
    "startMs": 279919,
    "text": "this I think there's a key to this which"
  },
  {
    "startMs": 282560,
    "text": "is a gentic uh uh graph-based approach."
  },
  {
    "startMs": 285919,
    "text": "The idea was um uh you have a large"
  },
  {
    "startMs": 288560,
    "text": "language model and these nodes and then"
  },
  {
    "startMs": 290000,
    "text": "you sort of put them together and you"
  },
  {
    "startMs": 291280,
    "text": "have the sort of the AI traverse this"
  },
  {
    "startMs": 293360,
    "text": "graph that you set up you build the"
  },
  {
    "startMs": 294880,
    "text": "graph. This is the key approach that and"
  },
  {
    "startMs": 297440,
    "text": "if you use this this is going to really"
  },
  {
    "startMs": 299280,
    "text": "help you sort of build agents because"
  },
  {
    "startMs": 301759,
    "text": "what is anything that we do in life it's"
  },
  {
    "startMs": 303520,
    "text": "a agent it's a workflow um uh it's a"
  },
  {
    "startMs": 306479,
    "text": "it's a way and then if you have an"
  },
  {
    "startMs": 308000,
    "text": "intelligent uh uh a agent it can"
  },
  {
    "startMs": 310160,
    "text": "basically traverse this is the key"
  },
  {
    "startMs": 312080,
    "text": "approach and I believe that at the time"
  },
  {
    "startMs": 314080,
    "text": "and a lot of people did and I still love"
  },
  {
    "startMs": 316400,
    "text": "this approach but nowadays it's sort of"
  },
  {
    "startMs": 320320,
    "text": "a little bit out of date in fact I"
  },
  {
    "startMs": 322080,
    "text": "remember um uh a guy came up to me after"
  },
  {
    "startMs": 324479,
    "text": "my speech last time and he was like the"
  },
  {
    "startMs": 326960,
    "text": "problem you talked about the the answer"
  },
  {
    "startMs": 329440,
    "text": "is just this exactly you were speaking"
  },
  {
    "startMs": 331360,
    "text": "to me thank you so much and and I was"
  },
  {
    "startMs": 332960,
    "text": "happy I I you know I gave a good speech"
  },
  {
    "startMs": 334400,
    "text": "and gave somebody some good advice um"
  },
  {
    "startMs": 336479,
    "text": "and then I remember when we made a"
  },
  {
    "startMs": 337919,
    "text": "change I was like I wonder what happened"
  },
  {
    "startMs": 339120,
    "text": "to that guy I wonder if he's here uh"
  },
  {
    "startMs": 340800,
    "text": "it's uh uh so um but the problem is is"
  },
  {
    "startMs": 344160,
    "text": "that not that it was wrong that that was"
  },
  {
    "startMs": 345759,
    "text": "the best approach but a new way emerged"
  },
  {
    "startMs": 349919,
    "text": "um in fact I started to like look"
  },
  {
    "startMs": 352000,
    "text": "through like all of the last year's uh"
  },
  {
    "startMs": 355039,
    "text": "events and actually go to other"
  },
  {
    "startMs": 356400,
    "text": "conferences like what what did people"
  },
  {
    "startMs": 357520,
    "text": "talk about a year ago and most of them"
  },
  {
    "startMs": 360080,
    "text": "were again nothing much wrong all good"
  },
  {
    "startMs": 362800,
    "text": "speeches all all good ideas but most of"
  },
  {
    "startMs": 365680,
    "text": "them have now a better way there um and"
  },
  {
    "startMs": 369199,
    "text": "uh so um and this is be sort of the gist"
  },
  {
    "startMs": 373440,
    "text": "of of the challenge so if you look at"
  },
  {
    "startMs": 376319,
    "text": "our journey of the technologies the kind"
  },
  {
    "startMs": 378240,
    "text": "of things that we care about um I'll"
  },
  {
    "startMs": 380000,
    "text": "just kind of rapid file here like so"
  },
  {
    "startMs": 382000,
    "text": "let's say that you want to utilize AI"
  },
  {
    "startMs": 383199,
    "text": "models and let's just look the last"
  },
  {
    "startMs": 384479,
    "text": "couple years a long a while ago probably"
  },
  {
    "startMs": 386720,
    "text": "distant memory now like people would say"
  },
  {
    "startMs": 388080,
    "text": "train your own my models or maybe fine"
  },
  {
    "startMs": 389759,
    "text": "tune them like nah that doesn't that's"
  },
  {
    "startMs": 391520,
    "text": "too why bother just use a frontier model"
  },
  {
    "startMs": 393840,
    "text": "use something from openi use something"
  },
  {
    "startMs": 394880,
    "text": "from anthropic use something from from"
  },
  {
    "startMs": 396560,
    "text": "Gemini um and then that's great but it's"
  },
  {
    "startMs": 399199,
    "text": "kind of expensive okay great let's just"
  },
  {
    "startMs": 400560,
    "text": "use openweight models they're pretty"
  },
  {
    "startMs": 401759,
    "text": "close you can use them you can host them"
  },
  {
    "startMs": 403039,
    "text": "yourself you can get some good GPUs um"
  },
  {
    "startMs": 405600,
    "text": "but then uh some companies will come to"
  },
  {
    "startMs": 407360,
    "text": "you and they'll be like look we just did"
  },
  {
    "startMs": 408560,
    "text": "this big deal with open orthropic like"
  },
  {
    "startMs": 410160,
    "text": "can we use our own key or bring our own"
  },
  {
    "startMs": 411520,
    "text": "model like sure you can do that too but"
  },
  {
    "startMs": 413280,
    "text": "then nowadays is probably the best"
  },
  {
    "startMs": 415280,
    "text": "approach is to do an adaptive model"
  },
  {
    "startMs": 416720,
    "text": "selection where you basically are"
  },
  {
    "startMs": 418319,
    "text": "picking the big and smaller models what"
  },
  {
    "startMs": 419919,
    "text": "which model does well and this is kind"
  },
  {
    "startMs": 421440,
    "text": "of the cool new thing maybe I could give"
  },
  {
    "startMs": 423120,
    "text": "a talk on that um or let's say you're"
  },
  {
    "startMs": 425599,
    "text": "building agents like we lot lot of"
  },
  {
    "startMs": 427840,
    "text": "things here like um I mean the word"
  },
  {
    "startMs": 429360,
    "text": "agent hasn't really been around for that"
  },
  {
    "startMs": 430800,
    "text": "long but in that time it used to be like"
  },
  {
    "startMs": 432720,
    "text": "a singleshot uh LLM response call that"
  },
  {
    "startMs": 435280,
    "text": "an agent if you feel like it then you"
  },
  {
    "startMs": 436639,
    "text": "have to say no okay we're chain of"
  },
  {
    "startMs": 438080,
    "text": "thought reasoning now we're going to"
  },
  {
    "startMs": 439440,
    "text": "make a graph-based agent system like I"
  },
  {
    "startMs": 441599,
    "text": "presented last year uh no then it turns"
  },
  {
    "startMs": 443520,
    "text": "that why are you bothering to make"
  },
  {
    "startMs": 444560,
    "text": "graphs when you could actually have an"
  },
  {
    "startMs": 445680,
    "text": "agent just figure out what to do? Make a"
  },
  {
    "startMs": 446960,
    "text": "plan. That's the new approach. That's"
  },
  {
    "startMs": 448240,
    "text": "kind of the way that um Claude sort of"
  },
  {
    "startMs": 450240,
    "text": "uh laid the approach there. And you're"
  },
  {
    "startMs": 452319,
    "text": "like, okay. And then now it's like,"
  },
  {
    "startMs": 454400,
    "text": "well, you if you want to use dedicated"
  },
  {
    "startMs": 456000,
    "text": "sub agents, maybe, but then maybe why"
  },
  {
    "startMs": 457840,
    "text": "not just make a generic agent and have"
  },
  {
    "startMs": 459199,
    "text": "it recursively work and then give it"
  },
  {
    "startMs": 460960,
    "text": "skills. Skills are very generic. They're"
  },
  {
    "startMs": 462400,
    "text": "super helpful. Um and then maybe now"
  },
  {
    "startMs": 464560,
    "text": "it's maybe the idea is not just to do"
  },
  {
    "startMs": 466319,
    "text": "that, but to do it with a agent sandbox"
  },
  {
    "startMs": 468240,
    "text": "so the agent can write code and execute"
  },
  {
    "startMs": 469759,
    "text": "it because that's super useful because"
  },
  {
    "startMs": 471440,
    "text": "agents are great programmers. I'd have"
  },
  {
    "startMs": 472880,
    "text": "them sort of just live in their own"
  },
  {
    "startMs": 473919,
    "text": "computer. Um, and then arguably now"
  },
  {
    "startMs": 475919,
    "text": "that's the best approach or maybe even"
  },
  {
    "startMs": 477520,
    "text": "we're in the world now of like don't"
  },
  {
    "startMs": 479520,
    "text": "even bother with any of that. Just bring"
  },
  {
    "startMs": 480479,
    "text": "your own harness like like like let"
  },
  {
    "startMs": 482160,
    "text": "people select if they want to use one of"
  },
  {
    "startMs": 483520,
    "text": "these other systems and then you know"
  },
  {
    "startMs": 485599,
    "text": "not even just building agents but the"
  },
  {
    "startMs": 487039,
    "text": "technology around context retrieval"
  },
  {
    "startMs": 488479,
    "text": "things like um you know in the old world"
  },
  {
    "startMs": 490639,
    "text": "we were like BM25 and keyword search"
  },
  {
    "startMs": 492479,
    "text": "that's the way to do things but that's"
  },
  {
    "startMs": 493840,
    "text": "like distant memory. Uh obviously the"
  },
  {
    "startMs": 495840,
    "text": "future is is retrieve augmented"
  },
  {
    "startMs": 497280,
    "text": "generation embeddings approximate"
  },
  {
    "startMs": 499039,
    "text": "nearest neighbor that's how we're going"
  },
  {
    "startMs": 500560,
    "text": "to find data. Turns out that doesn't"
  },
  {
    "startMs": 502240,
    "text": "really scale well and it kind of almost"
  },
  {
    "startMs": 503520,
    "text": "mimics randomness as you keep going. So"
  },
  {
    "startMs": 504960,
    "text": "then maybe it's about graphs. It's"
  },
  {
    "startMs": 506879,
    "text": "difficult to get working well. It's"
  },
  {
    "startMs": 507919,
    "text": "probably not the best. Um so then it's"
  },
  {
    "startMs": 509520,
    "text": "about hybrid. You want a lexical and you"
  },
  {
    "startMs": 511280,
    "text": "want to do semantic search and rank fuse"
  },
  {
    "startMs": 513200,
    "text": "those together. Arguably not. Arguably"
  },
  {
    "startMs": 515919,
    "text": "agents are actually way better at"
  },
  {
    "startMs": 517680,
    "text": "finding data because they can find"
  },
  {
    "startMs": 519039,
    "text": "things and apply their intelligence to"
  },
  {
    "startMs": 520320,
    "text": "get to it. So each of these things I"
  },
  {
    "startMs": 522399,
    "text": "just mentioned is arguably the leading"
  },
  {
    "startMs": 524800,
    "text": "approach for that moment over time."
  },
  {
    "startMs": 528399,
    "text": "If you asked me to give a speech right"
  },
  {
    "startMs": 529760,
    "text": "now on any one of these, I would pick"
  },
  {
    "startMs": 530800,
    "text": "the last. I'd pick the adapted models"
  },
  {
    "startMs": 533120,
    "text": "with dedicated RM style agent and"
  },
  {
    "startMs": 535120,
    "text": "agentic search powered by hybrid."
  },
  {
    "startMs": 537839,
    "text": "But uh is this the end of this journey?"
  },
  {
    "startMs": 541360,
    "text": "This is not that long of that time here."
  },
  {
    "startMs": 544240,
    "text": "And so um my guess is the stuff that"
  },
  {
    "startMs": 547360,
    "text": "you're learning today likely won't last"
  },
  {
    "startMs": 550160,
    "text": "that long. Not that it's not wrong, not"
  },
  {
    "startMs": 552240,
    "text": "that it is not the best answer right"
  },
  {
    "startMs": 553920,
    "text": "now, but probably something's going to"
  },
  {
    "startMs": 556800,
    "text": "change. The big thing that changed last"
  },
  {
    "startMs": 558880,
    "text": "year was in my mind Opus 40 to Opus 45."
  },
  {
    "startMs": 563120,
    "text": "When you did that, suddenly you got to a"
  },
  {
    "startMs": 565200,
    "text": "model that could do instruction"
  },
  {
    "startMs": 566480,
    "text": "following and really high scale. This is"
  },
  {
    "startMs": 568560,
    "text": "kind of to me the beginning the epic of"
  },
  {
    "startMs": 570399,
    "text": "like the new agent models. Uh also uh"
  },
  {
    "startMs": 573839,
    "text": "hardware is getting better, faster,"
  },
  {
    "startMs": 575040,
    "text": "cheaper. Maybe we'll start to use more"
  },
  {
    "startMs": 576480,
    "text": "tokens. Like token usage is off the"
  },
  {
    "startMs": 577920,
    "text": "charts of course and is that good or"
  },
  {
    "startMs": 579279,
    "text": "bad? What's going to change there?"
  },
  {
    "startMs": 580800,
    "text": "Enterprises are adopting things"
  },
  {
    "startMs": 582080,
    "text": "differently whether or not a company has"
  },
  {
    "startMs": 584480,
    "text": "decided to go all in on one agent to"
  },
  {
    "startMs": 586320,
    "text": "rule them all sort of like claude or"
  },
  {
    "startMs": 588240,
    "text": "maybe codec style agent um or maybe they"
  },
  {
    "startMs": 590640,
    "text": "want to utilize uh agents from different"
  },
  {
    "startMs": 592720,
    "text": "platforms and different systems or both."
  },
  {
    "startMs": 595040,
    "text": "This is going to affect your lives um in"
  },
  {
    "startMs": 596800,
    "text": "addition to things like just the new"
  },
  {
    "startMs": 598080,
    "text": "techniques new interesting uh approaches"
  },
  {
    "startMs": 600480,
    "text": "new technology to power these things. So"
  },
  {
    "startMs": 603440,
    "text": "the fact that everybody in is working so"
  },
  {
    "startMs": 606080,
    "text": "hard on this trillions of dollars"
  },
  {
    "startMs": 607680,
    "text": "investment is actually leading to a lot"
  },
  {
    "startMs": 609279,
    "text": "of this change and again it's happening"
  },
  {
    "startMs": 610800,
    "text": "way faster than than I've ever seen for"
  },
  {
    "startMs": 613279,
    "text": "sure."
  },
  {
    "startMs": 615120,
    "text": "Um, now if you look back, um, it's not"
  },
  {
    "startMs": 618480,
    "text": "this way with everything else. Like if"
  },
  {
    "startMs": 620640,
    "text": "you go see some of these other"
  },
  {
    "startMs": 621839,
    "text": "discussions, like I've given a speech on"
  },
  {
    "startMs": 623279,
    "text": "some of these topics. I looked at them"
  },
  {
    "startMs": 624959,
    "text": "come some of a few years old. They're"
  },
  {
    "startMs": 626240,
    "text": "pretty good. I still think they're very"
  },
  {
    "startMs": 627680,
    "text": "relevant. You want to talk about large"
  },
  {
    "startMs": 628560,
    "text": "scale databases about uh identity access"
  },
  {
    "startMs": 631519,
    "text": "controls, how to scale engineering"
  },
  {
    "startMs": 632800,
    "text": "teams, how to do multi cloud storage,"
  },
  {
    "startMs": 635519,
    "text": "probably um these are still relevant"
  },
  {
    "startMs": 637920,
    "text": "things today. These do not change as"
  },
  {
    "startMs": 639120,
    "text": "fast despite being high-scale uh"
  },
  {
    "startMs": 641440,
    "text": "interesting powerful uh technologies."
  },
  {
    "startMs": 645040,
    "text": "So the previous wisdom of saying"
  },
  {
    "startMs": 647360,
    "text": "optimize for specific technologies"
  },
  {
    "startMs": 650079,
    "text": "go deep switch rarely right this what"
  },
  {
    "startMs": 652800,
    "text": "this the reason you do that is because"
  },
  {
    "startMs": 654320,
    "text": "switching is hard migrations suck"
  },
  {
    "startMs": 656640,
    "text": "whenever you migrate you break something"
  },
  {
    "startMs": 658079,
    "text": "every time no matter what um it's always"
  },
  {
    "startMs": 660640,
    "text": "harder than you think even [music] if"
  },
  {
    "startMs": 661760,
    "text": "you know that um and uh and the"
  },
  {
    "startMs": 663680,
    "text": "switching cost is basically high so"
  },
  {
    "startMs": 664959,
    "text": "basically don't do it for most things"
  },
  {
    "startMs": 666880,
    "text": "you're kind of uh just because"
  },
  {
    "startMs": 668959,
    "text": "something's better out there that's not"
  },
  {
    "startMs": 670079,
    "text": "the answer for most infrastructure so"
  },
  {
    "startMs": 672880,
    "text": "typically If you say half life of an"
  },
  {
    "startMs": 675040,
    "text": "agent infrastructure,"
  },
  {
    "startMs": 677360,
    "text": "3 to 5 years, reevali, see what's out"
  },
  {
    "startMs": 679680,
    "text": "there. We've been using databases like"
  },
  {
    "startMs": 681440,
    "text": "my SQL databases for a long time. It's"
  },
  {
    "startMs": 683040,
    "text": "still pretty good and probably need to"
  },
  {
    "startMs": 684399,
    "text": "replace it soon. But um but now with AI"
  },
  {
    "startMs": 688880,
    "text": "technologies, arguably the halflife is"
  },
  {
    "startMs": 691760,
    "text": "measured in months, meaning a few months"
  },
  {
    "startMs": 693920,
    "text": "after you've adopted what might be the"
  },
  {
    "startMs": 695839,
    "text": "best possible thing, there's a"
  },
  {
    "startMs": 697920,
    "text": "significant chance that you're going to"
  },
  {
    "startMs": 699279,
    "text": "have to replace it coming soon. And this"
  },
  {
    "startMs": 701440,
    "text": "is I think shocking."
  },
  {
    "startMs": 704079,
    "text": "Maybe I see from some of your reactions"
  },
  {
    "startMs": 705440,
    "text": "that like you're kind of like have"
  },
  {
    "startMs": 707120,
    "text": "experienced this a little bit. Um but"
  },
  {
    "startMs": 708880,
    "text": "this is a different aspect of the way"
  },
  {
    "startMs": 711040,
    "text": "that you build technology."
  },
  {
    "startMs": 713760,
    "text": "So if you're an engineer, this really"
  },
  {
    "startMs": 716720,
    "text": "sucks because the thing that you just"
  },
  {
    "startMs": 718399,
    "text": "learned and that you're making is now"
  },
  {
    "startMs": 720320,
    "text": "probably going to be out of date soon."
  },
  {
    "startMs": 722160,
    "text": "Uh no engineer I know likes this. Um uh"
  },
  {
    "startMs": 725680,
    "text": "as a startup, you bet on something."
  },
  {
    "startMs": 727360,
    "text": "You're like, we're going to go all in."
  },
  {
    "startMs": 728240,
    "text": "We're going to go on the technology this"
  },
  {
    "startMs": 729279,
    "text": "approach and then we're going to"
  },
  {
    "startMs": 731200,
    "text": "basically uh uh disrupt somebody which"
  },
  {
    "startMs": 733760,
    "text": "probably will but then you see and see"
  },
  {
    "startMs": 735680,
    "text": "now like the first phase of AI companies"
  },
  {
    "startMs": 737440,
    "text": "are starting to get disrupted by the"
  },
  {
    "startMs": 738959,
    "text": "next phase. If you're a technology"
  },
  {
    "startMs": 741200,
    "text": "buyer, you're a leader of a company, you"
  },
  {
    "startMs": 742959,
    "text": "buy technology, you you you select uh"
  },
  {
    "startMs": 745839,
    "text": "open source models, you select vendors,"
  },
  {
    "startMs": 748480,
    "text": "there's a significant chance that"
  },
  {
    "startMs": 749839,
    "text": "whatever you just bought is not going to"
  },
  {
    "startMs": 751839,
    "text": "be the approach you're going to invest."
  },
  {
    "startMs": 753519,
    "text": "That's you know, good luck doing a"
  },
  {
    "startMs": 755519,
    "text": "three-year deal like on on things about"
  },
  {
    "startMs": 757279,
    "text": "about this kind of stuff. Or if you're"
  },
  {
    "startMs": 758639,
    "text": "VC, uh maybe the coolest best thing that"
  },
  {
    "startMs": 761440,
    "text": "everybody agrees is the greatest"
  },
  {
    "startMs": 762560,
    "text": "opportunity is no longer going to be the"
  },
  {
    "startMs": 764320,
    "text": "opportunity soon um because everything's"
  },
  {
    "startMs": 767120,
    "text": "changing."
  },
  {
    "startMs": 769040,
    "text": "So here's my advice. Get good at"
  },
  {
    "startMs": 771440,
    "text": "changing."
  },
  {
    "startMs": 775680,
    "text": "It's almost silly to say because you"
  },
  {
    "startMs": 777360,
    "text": "know obviously technology changes."
  },
  {
    "startMs": 778800,
    "text": "Obviously it's something that is um you"
  },
  {
    "startMs": 781519,
    "text": "know built in. Of course we're all going"
  },
  {
    "startMs": 782720,
    "text": "to change. We've done this for a long"
  },
  {
    "startMs": 783680,
    "text": "time. Um it's hard. I think it's really"
  },
  {
    "startMs": 786160,
    "text": "hard and the faster that you do it the"
  },
  {
    "startMs": 788320,
    "text": "harder it is. Uh when I was going"
  },
  {
    "startMs": 790959,
    "text": "through that uh like oh yeah we switched"
  },
  {
    "startMs": 792399,
    "text": "from the graph based agent to a um to"
  },
  {
    "startMs": 794000,
    "text": "the more looping style deep style agent."
  },
  {
    "startMs": 796720,
    "text": "I remember very well the conversation"
  },
  {
    "startMs": 798160,
    "text": "with the engineer. He just he's like, \"I"
  },
  {
    "startMs": 799839,
    "text": "did it. I got a tic search working and"
  },
  {
    "startMs": 801519,
    "text": "this approach does deep research. It"
  },
  {
    "startMs": 802800,
    "text": "does all this stuff just like you"
  },
  {
    "startMs": 803839,
    "text": "asked.\" Okay, we're going to switch"
  },
  {
    "startMs": 806880,
    "text": "rebuild it again in this new technology."
  },
  {
    "startMs": 808720,
    "text": "And he's like, \"Wait, what?\" Like, \"It's"
  },
  {
    "startMs": 810880,
    "text": "working. You did what you're talking.\""
  },
  {
    "startMs": 812000,
    "text": "Yeah, but it's not as capable as we"
  },
  {
    "startMs": 813040,
    "text": "wanted it to be. Like, what do you mean?"
  },
  {
    "startMs": 814320,
    "text": "You didn't tell me that before. Like,"
  },
  {
    "startMs": 815440,
    "text": "and then and then so convince him like,"
  },
  {
    "startMs": 816639,
    "text": "\"Okay, this is a new approach.\" And"
  },
  {
    "startMs": 818399,
    "text": "then, you know, he does it and it's"
  },
  {
    "startMs": 820399,
    "text": "good. Two months later, uh, we we're"
  },
  {
    "startMs": 822639,
    "text": "we're actually shipping the product uh"
  },
  {
    "startMs": 824800,
    "text": "on on Tuesday. And then I was like,"
  },
  {
    "startMs": 826320,
    "text": "\"Okay, uh, guys, on Wednesday we're"
  },
  {
    "startMs": 828079,
    "text": "going to rebuild it again on the new"
  },
  {
    "startMs": 829680,
    "text": "approach.\" And they're like, \"What are"
  },
  {
    "startMs": 830959,
    "text": "you talking about?\" Like, like, um, then"
  },
  {
    "startMs": 833040,
    "text": "they'll say, like, it's it's almost hard"
  },
  {
    "startMs": 834399,
    "text": "on everybody like, \"Wait, wait, give me"
  },
  {
    "startMs": 835680,
    "text": "more time. I'll I'll make the new way"
  },
  {
    "startMs": 836959,
    "text": "the old way do it better.\" Um, uh, like,"
  },
  {
    "startMs": 839279,
    "text": "and then also they're skeptical. Like,"
  },
  {
    "startMs": 841600,
    "text": "now you say that, but like this is going"
  },
  {
    "startMs": 843120,
    "text": "to change again, right? Like, who are"
  },
  {
    "startMs": 844240,
    "text": "you to like make these choices? And the"
  },
  {
    "startMs": 846320,
    "text": "answer is, yeah, I'm pretty sure it's"
  },
  {
    "startMs": 847680,
    "text": "going to change again. So, this is, I"
  },
  {
    "startMs": 850079,
    "text": "think, a leadership problem. It's a"
  },
  {
    "startMs": 851519,
    "text": "technology problem. It's a morale"
  },
  {
    "startMs": 853600,
    "text": "problem. It's a team problem. It's a"
  },
  {
    "startMs": 855519,
    "text": "company problem and if you're not"
  },
  {
    "startMs": 857199,
    "text": "careful, it is actually can destroy you."
  },
  {
    "startMs": 859440,
    "text": "It can destroy a lot of things because"
  },
  {
    "startMs": 861120,
    "text": "people lose faith, they lose morale."
  },
  {
    "startMs": 862560,
    "text": "It's a problem."
  },
  {
    "startMs": 864639,
    "text": "So, um if my advice is change um and be"
  },
  {
    "startMs": 868639,
    "text": "ready for change, how are you going to"
  },
  {
    "startMs": 870320,
    "text": "do it? Three things to give you."
  },
  {
    "startMs": 873440,
    "text": "One, um you just got to prepare people."
  },
  {
    "startMs": 875760,
    "text": "Well, this is a kind of a people"
  },
  {
    "startMs": 877120,
    "text": "challenge. So when you build your teams,"
  },
  {
    "startMs": 879440,
    "text": "when you talk to them, when you prepare"
  },
  {
    "startMs": 880639,
    "text": "them, if they're in AI world, you got to"
  },
  {
    "startMs": 882399,
    "text": "tell them like expect change. It's"
  },
  {
    "startMs": 884959,
    "text": "normal. It's not a problem. It's not"
  },
  {
    "startMs": 887040,
    "text": "that you did something wrong. This is is"
  },
  {
    "startMs": 888959,
    "text": "weirdly like um like helps people like I"
  },
  {
    "startMs": 891040,
    "text": "have a a technology review team and and"
  },
  {
    "startMs": 893199,
    "text": "then and then they're like we can't like"
  },
  {
    "startMs": 894880,
    "text": "change. We don't know. We're not sure."
  },
  {
    "startMs": 895920,
    "text": "We can't tell you that in two years from"
  },
  {
    "startMs": 897279,
    "text": "now this is going to be best. Like"
  },
  {
    "startMs": 898560,
    "text": "that's okay. Uh we're gonna we're gonna"
  },
  {
    "startMs": 900639,
    "text": "build these things that change. So just"
  },
  {
    "startMs": 902320,
    "text": "go with it. You have to pick something."
  },
  {
    "startMs": 904240,
    "text": "Um, also whenever possible if you can"
  },
  {
    "startMs": 906240,
    "text": "build an abstraction so that it lets you"
  },
  {
    "startMs": 908000,
    "text": "swap out what's underneath. We have an"
  },
  {
    "startMs": 909440,
    "text": "agent extraction in box and you're able"
  },
  {
    "startMs": 910959,
    "text": "to go through and be like uh like select"
  },
  {
    "startMs": 913040,
    "text": "things underneath and the agent still"
  },
  {
    "startMs": 914480,
    "text": "works the same for the customers but it"
  },
  {
    "startMs": 916160,
    "text": "it's better underneath. Um, and the idea"
  },
  {
    "startMs": 919760,
    "text": "is that change is not a mistake and and"
  },
  {
    "startMs": 922160,
    "text": "I highlight like that's very hard for"
  },
  {
    "startMs": 924720,
    "text": "most people and and I and I would sort"
  },
  {
    "startMs": 926240,
    "text": "of just you just I tell them all the"
  },
  {
    "startMs": 927600,
    "text": "time change is not a mistake. You"
  },
  {
    "startMs": 929279,
    "text": "wouldn't nobody knew six months ago."
  },
  {
    "startMs": 931199,
    "text": "Nobody today will know six months from"
  },
  {
    "startMs": 932720,
    "text": "now. It's Seems very true. So uh at Box"
  },
  {
    "startMs": 936480,
    "text": "we are now in the habit of reviewing"
  },
  {
    "startMs": 937760,
    "text": "every six months no matter what. This is"
  },
  {
    "startMs": 939519,
    "text": "great technology. We love it. Review in"
  },
  {
    "startMs": 941360,
    "text": "six months like because uh which is just"
  },
  {
    "startMs": 943920,
    "text": "completely crazy for everything else"
  },
  {
    "startMs": 945519,
    "text": "that we're doing. Everything else is"
  },
  {
    "startMs": 946880,
    "text": "like three years. Um also even though"
  },
  {
    "startMs": 950480,
    "text": "change is critical you um you need to"
  },
  {
    "startMs": 953040,
    "text": "define what you mean when change. If you"
  },
  {
    "startMs": 954800,
    "text": "just change all the time there's a new"
  },
  {
    "startMs": 956720,
    "text": "paper it's awesome. You know our CEO"
  },
  {
    "startMs": 959279,
    "text": "Aaron is very active on all the newest"
  },
  {
    "startMs": 961120,
    "text": "things. He's like check this out. Like"
  },
  {
    "startMs": 962480,
    "text": "don't change just because of that. Like"
  },
  {
    "startMs": 964399,
    "text": "don't change just because it's a trend."
  },
  {
    "startMs": 965759,
    "text": "Change because you know it matters. And"
  },
  {
    "startMs": 968079,
    "text": "how do you know it matters? Probably"
  },
  {
    "startMs": 969759,
    "text": "pitch you on eval sets. If you're"
  },
  {
    "startMs": 971120,
    "text": "building agents, if you're building AI,"
  },
  {
    "startMs": 973040,
    "text": "make sure that you know what people"
  },
  {
    "startMs": 974240,
    "text": "have. You have the ability to give the"
  },
  {
    "startMs": 975519,
    "text": "same input, expect certain output. Grade"
  },
  {
    "startMs": 977360,
    "text": "that cost, speed, quality, capabilities."
  },
  {
    "startMs": 979600,
    "text": "These are the things that you probably"
  },
  {
    "startMs": 981040,
    "text": "are going to to be wanting. So for us,"
  },
  {
    "startMs": 983279,
    "text": "it's easy. Does the new approach work"
  },
  {
    "startMs": 987040,
    "text": "better for our eval sets? What the"
  },
  {
    "startMs": 988880,
    "text": "customer cares about? If the answer is"
  },
  {
    "startMs": 990399,
    "text": "yes, strongly consider switching. If the"
  },
  {
    "startMs": 992000,
    "text": "answer is no, don't bother like or or"
  },
  {
    "startMs": 994160,
    "text": "keep working on a little bit of work to"
  },
  {
    "startMs": 995839,
    "text": "see if you can make sure that you you've"
  },
  {
    "startMs": 997120,
    "text": "fully explored it. Um and then so the"
  },
  {
    "startMs": 999920,
    "text": "idea is uh build a system that lets you"
  },
  {
    "startMs": 1002399,
    "text": "be able to change. And then the third uh"
  },
  {
    "startMs": 1005759,
    "text": "and final piece of advice here is um"
  },
  {
    "startMs": 1009040,
    "text": "almost certainly none of us can keep up"
  },
  {
    "startMs": 1011040,
    "text": "with everything. It is very hard. Um I"
  },
  {
    "startMs": 1014160,
    "text": "think I heard uh Andre Kaparthy uh he he"
  },
  {
    "startMs": 1016720,
    "text": "was like everything changes so fast I"
  },
  {
    "startMs": 1018639,
    "text": "can't keep up. and you're like you're"
  },
  {
    "startMs": 1020240,
    "text": "sort of quite famously good at keeping"
  },
  {
    "startMs": 1021839,
    "text": "up and so like what's the hope for"
  },
  {
    "startMs": 1023040,
    "text": "everybody else if if that's the case. Um"
  },
  {
    "startMs": 1025120,
    "text": "and so but then so what you do is you"
  },
  {
    "startMs": 1026959,
    "text": "rely on somebody else. You rely on a"
  },
  {
    "startMs": 1028880,
    "text": "technology, you rely on a vendor, you"
  },
  {
    "startMs": 1030720,
    "text": "rely on a platform. Um you when you"
  },
  {
    "startMs": 1033600,
    "text": "select it and um and then here I think"
  },
  {
    "startMs": 1036880,
    "text": "very use I mean like whenever you"
  },
  {
    "startMs": 1038559,
    "text": "whenever anybody's bought technology in"
  },
  {
    "startMs": 1039839,
    "text": "the past I would had advised them like"
  },
  {
    "startMs": 1042400,
    "text": "look at what they do now. Double check"
  },
  {
    "startMs": 1044400,
    "text": "the road map. Make sure it's good. Make"
  },
  {
    "startMs": 1045919,
    "text": "sure it's on the path you want but just"
  },
  {
    "startMs": 1047120,
    "text": "focus on what's available now. But I"
  },
  {
    "startMs": 1049280,
    "text": "think something else here is um should"
  },
  {
    "startMs": 1051440,
    "text": "do that of course that's most important"
  },
  {
    "startMs": 1053360,
    "text": "thing but like look back how have they"
  },
  {
    "startMs": 1056000,
    "text": "handled change what's their attitude"
  },
  {
    "startMs": 1057760,
    "text": "towards change how can what can you when"
  },
  {
    "startMs": 1059280,
    "text": "you talk to them when you read about"
  },
  {
    "startMs": 1061280,
    "text": "their stuff like what happened six"
  },
  {
    "startMs": 1063200,
    "text": "months ago what happened a year ago how"
  },
  {
    "startMs": 1065200,
    "text": "did they handle that transition many of"
  },
  {
    "startMs": 1067280,
    "text": "the vendors that I really like right now"
  },
  {
    "startMs": 1069120,
    "text": "have reinvented themselves three times"
  },
  {
    "startMs": 1070640,
    "text": "in the last year and I now trust that if"
  },
  {
    "startMs": 1073600,
    "text": "something else comes along they're very"
  },
  {
    "startMs": 1075039,
    "text": "good at this they understand agent"
  },
  {
    "startMs": 1076480,
    "text": "technologies they understand the the"
  },
  {
    "startMs": 1077520,
    "text": "eval sets they understand the the"
  },
  {
    "startMs": 1079200,
    "text": "observability systems and then you can"
  },
  {
    "startMs": 1080720,
    "text": "say ah okay good I hope that they keep"
  },
  {
    "startMs": 1082960,
    "text": "up and then I now my sort of thing I"
  },
  {
    "startMs": 1085679,
    "text": "need to do is just evaluate whether or"
  },
  {
    "startMs": 1087039,
    "text": "not that's a good platform"
  },
  {
    "startMs": 1089120,
    "text": "so um making sure that you have this"
  },
  {
    "startMs": 1091840,
    "text": "sort of platforms that do well is is is"
  },
  {
    "startMs": 1094160,
    "text": "critical um and um if anybody's"
  },
  {
    "startMs": 1096799,
    "text": "interested in unstructured content and"
  },
  {
    "startMs": 1098720,
    "text": "AI associated with it uh Box has a booth"
  },
  {
    "startMs": 1100880,
    "text": "downstairs happy to talk to you about"
  },
  {
    "startMs": 1102080,
    "text": "those kind of things"
  },
  {
    "startMs": 1104320,
    "text": "um and then um I I'll leave you with"
  },
  {
    "startMs": 1106640,
    "text": "this is um I actually I fully bet and I"
  },
  {
    "startMs": 1109039,
    "text": "believe that um a company that's born"
  },
  {
    "startMs": 1111760,
    "text": "this year was born last year um will or"
  },
  {
    "startMs": 1114720,
    "text": "maybe even a company a medium-sized"
  },
  {
    "startMs": 1116400,
    "text": "company or a big company will will they"
  },
  {
    "startMs": 1118480,
    "text": "they'll shoot very high the company that"
  },
  {
    "startMs": 1120799,
    "text": "will dominate tomorrow is is is now born"
  },
  {
    "startMs": 1123280,
    "text": "today. Uh but I kind of bet you that the"
  },
  {
    "startMs": 1127039,
    "text": "technology approach that they have right"
  },
  {
    "startMs": 1128640,
    "text": "now is probably going to change multiple"
  },
  {
    "startMs": 1130559,
    "text": "times before they do that. So"
  },
  {
    "startMs": 1132880,
    "text": "interestingly it's like the challenge"
  },
  {
    "startMs": 1134799,
    "text": "the advice the thought here is build for"
  },
  {
    "startMs": 1137039,
    "text": "change"
  },
  {
    "startMs": 1138559,
    "text": "adaptability arguably that's the moat"
  },
  {
    "startMs": 1141360,
    "text": "that you have"
  },
  {
    "startMs": 1143360,
    "text": "until that changes."
  },
  {
    "startMs": 1145679,
    "text": "Okay thank you everyone."
  },
  {
    "startMs": 1148228,
    "text": "[applause]"
  }
]