[
  {
    "startMs": 1309,
    "text": "[music]"
  },
  {
    "startMs": 12400,
    "text": ">> Good afternoon."
  },
  {
    "startMs": 13960,
    "text": "I hope you're all having a lovely day"
  },
  {
    "startMs": 15440,
    "text": "here at AIE."
  },
  {
    "startMs": 16920,
    "text": "We've had great weather, though the UV"
  },
  {
    "startMs": 18200,
    "text": "has been like nine. So, hopefully you"
  },
  {
    "startMs": 19840,
    "text": "put your sunscreen on your being"
  },
  {
    "startMs": 21200,
    "text": "appropriate adults."
  },
  {
    "startMs": 22760,
    "text": "I'm here to talk to you about context"
  },
  {
    "startMs": 24040,
    "text": "engineering, and I have the good fortune"
  },
  {
    "startMs": 25480,
    "text": "of following AJ from LinkedIn because he"
  },
  {
    "startMs": 27200,
    "text": "talked a lot about the system that we"
  },
  {
    "startMs": 28520,
    "text": "actually design and sell to other"
  },
  {
    "startMs": 29960,
    "text": "solutions. And I'm going to give you a"
  },
  {
    "startMs": 31280,
    "text": "bunch of open source tools. So, if you"
  },
  {
    "startMs": 33280,
    "text": "watch that last talk just before me,"
  },
  {
    "startMs": 35280,
    "text": "you're going to get a bunch of tool"
  },
  {
    "startMs": 36000,
    "text": "chance you can go mess around yourself,"
  },
  {
    "startMs": 37400,
    "text": "and I'll teach you a bunch of techniques"
  },
  {
    "startMs": 38520,
    "text": "today. The goal, of course, is to fix"
  },
  {
    "startMs": 41360,
    "text": "your absolutely right."
  },
  {
    "startMs": 43000,
    "text": "I think they've taken that out of the"
  },
  {
    "startMs": 44200,
    "text": "prompts now, so it just says you're"
  },
  {
    "startMs": 45520,
    "text": "right or other things, but I'm sure"
  },
  {
    "startMs": 47320,
    "text": "you've all been there."
  },
  {
    "startMs": 49280,
    "text": "So, I'm Brandon."
  },
  {
    "startMs": 50520,
    "text": "I work at Unblocked. Uh yes, I have a"
  },
  {
    "startMs": 52520,
    "text": "coconut. We've been giving these away"
  },
  {
    "startMs": 53960,
    "text": "for fresh context, fresh fresh coconuts."
  },
  {
    "startMs": 57280,
    "text": "But, the thing that I want to talk to"
  },
  {
    "startMs": 58240,
    "text": "you about is with these models,"
  },
  {
    "startMs": 60080,
    "text": "especially with Meth O'Clock models, I"
  },
  {
    "startMs": 61680,
    "text": "think Fable 5's coming back today, so"
  },
  {
    "startMs": 63440,
    "text": "they say."
  },
  {
    "startMs": 64519,
    "text": "You can watch my Grain Call recording"
  },
  {
    "startMs": 66280,
    "text": "try to book this."
  },
  {
    "startMs": 67920,
    "text": "We'll ignore it."
  },
  {
    "startMs": 69440,
    "text": "But, what I want you to do is to think"
  },
  {
    "startMs": 71120,
    "text": "about the fact that with these tools,"
  },
  {
    "startMs": 73520,
    "text": "AI-generated code should feel like it"
  },
  {
    "startMs": 75640,
    "text": "was written by someone who's been on"
  },
  {
    "startMs": 77240,
    "text": "your team for years."
  },
  {
    "startMs": 81360,
    "text": "So, to get in the right headspace, for"
  },
  {
    "startMs": 83520,
    "text": "years you have to consider that you have"
  },
  {
    "startMs": 85200,
    "text": "been the context engine."
  },
  {
    "startMs": 87120,
    "text": "How did you do that?"
  },
  {
    "startMs": 89480,
    "text": "You built context by going to work"
  },
  {
    "startMs": 92160,
    "text": "and asking questions,"
  },
  {
    "startMs": 94120,
    "text": "shipping PRs and getting them rejected,"
  },
  {
    "startMs": 96120,
    "text": "going to meetings, and all this slowly"
  },
  {
    "startMs": 98040,
    "text": "over time built up the engine that is"
  },
  {
    "startMs": 99640,
    "text": "your brain."
  },
  {
    "startMs": 100760,
    "text": "You understand how it works here. You"
  },
  {
    "startMs": 102200,
    "text": "know how stuff gets shipped. You were on"
  },
  {
    "startMs": 104000,
    "text": "call that night when you took prod down"
  },
  {
    "startMs": 106720,
    "text": "and why that happened."
  },
  {
    "startMs": 108480,
    "text": "The problem is"
  },
  {
    "startMs": 110360,
    "text": "that these agents have this exact same"
  },
  {
    "startMs": 111960,
    "text": "problem. Every time you create a new"
  },
  {
    "startMs": 113640,
    "text": "terminal session with an agent in it,"
  },
  {
    "startMs": 115560,
    "text": "it's very intelligent, but it doesn't"
  },
  {
    "startMs": 117080,
    "text": "have any context on how your company"
  },
  {
    "startMs": 118800,
    "text": "operates. So, it needs to get that"
  },
  {
    "startMs": 120680,
    "text": "somehow."
  },
  {
    "startMs": 122120,
    "text": "The problem is as you move these agents"
  },
  {
    "startMs": 123600,
    "text": "up in scale,"
  },
  {
    "startMs": 125360,
    "text": "that cost compounds if you get it"
  },
  {
    "startMs": 127800,
    "text": "incorrect at the beginning. The leverage"
  },
  {
    "startMs": 130080,
    "text": "of context and content"
  },
  {
    "startMs": 132880,
    "text": "We're just going to fix this cuz I think"
  },
  {
    "startMs": 133959,
    "text": "people want to take some photos."
  },
  {
    "startMs": 137720,
    "text": "Perfect."
  },
  {
    "startMs": 139280,
    "text": "That context issue will compound. So, at"
  },
  {
    "startMs": 142040,
    "text": "the far left, we all remember the"
  },
  {
    "startMs": 144800,
    "text": "age-old time of 2 years ago where we had"
  },
  {
    "startMs": 146720,
    "text": "tab complete models that were pretty"
  },
  {
    "startMs": 147880,
    "text": "cool. What happened is it popped up and"
  },
  {
    "startMs": 149880,
    "text": "said, \"Hey, do you want to tab this?\""
  },
  {
    "startMs": 151120,
    "text": "And quickly in your head with your"
  },
  {
    "startMs": 152520,
    "text": "context engine, you go, \"No, that's"
  },
  {
    "startMs": 153600,
    "text": "bad.\" Or you went, \"Oh, sweet.\" You hit"
  },
  {
    "startMs": 155320,
    "text": "tab. Nice."
  },
  {
    "startMs": 156959,
    "text": "As we move along the agentic kind of"
  },
  {
    "startMs": 158560,
    "text": "adoption curve, what happens is you are"
  },
  {
    "startMs": 161040,
    "text": "moving into more situations in which you"
  },
  {
    "startMs": 163320,
    "text": "have agents running without a human in"
  },
  {
    "startMs": 165360,
    "text": "the loop, or at least you wish you"
  },
  {
    "startMs": 166959,
    "text": "didn't have to be in the loop."
  },
  {
    "startMs": 168880,
    "text": "What they need is some way to be able to"
  },
  {
    "startMs": 170760,
    "text": "ask the questions they need when they"
  },
  {
    "startMs": 172280,
    "text": "hit walls in order to write code or"
  },
  {
    "startMs": 174560,
    "text": "solve or basically fix the issue and"
  },
  {
    "startMs": 176360,
    "text": "ultimately output code that's mergeable"
  },
  {
    "startMs": 178880,
    "text": "into your code base, especially with"
  },
  {
    "startMs": 180440,
    "text": "many people here who actually work in"
  },
  {
    "startMs": 182480,
    "text": "brownfield code bases that have been"
  },
  {
    "startMs": 183959,
    "text": "around for a long time that run real"
  },
  {
    "startMs": 185959,
    "text": "revenue across them, not just greenfield"
  },
  {
    "startMs": 188080,
    "text": "fun projects."
  },
  {
    "startMs": 190760,
    "text": "So, that cost of bad context compounding"
  },
  {
    "startMs": 192840,
    "text": "at the beginning is cheap. If you think"
  },
  {
    "startMs": 194600,
    "text": "like shift left, finding a defect or a"
  },
  {
    "startMs": 196840,
    "text": "bug, you want to find it as early as"
  },
  {
    "startMs": 198800,
    "text": "possible. It's the same with context."
  },
  {
    "startMs": 201720,
    "text": "Cuz as you move across, you get into"
  },
  {
    "startMs": 203120,
    "text": "doom loops. You usually ask your to do"
  },
  {
    "startMs": 205360,
    "text": "something. It's like, \"Hey, I did it.\""
  },
  {
    "startMs": 206920,
    "text": "And you're like, \"No, man.\" And then you"
  },
  {
    "startMs": 208480,
    "text": "correct and correct and correct. That's"
  },
  {
    "startMs": 210440,
    "text": "wasted search tokens. It's also wasted"
  },
  {
    "startMs": 212720,
    "text": "rework time."
  },
  {
    "startMs": 214040,
    "text": "And that is not acceptable with the"
  },
  {
    "startMs": 216480,
    "text": "tokenomics we have coming."
  },
  {
    "startMs": 218440,
    "text": "And then as you move into parallel"
  },
  {
    "startMs": 219480,
    "text": "agents, etc., you start hitting a review"
  },
  {
    "startMs": 221040,
    "text": "tax. So, these AI code reviewers we're"
  },
  {
    "startMs": 223239,
    "text": "trying to use, but again, key context is"
  },
  {
    "startMs": 225400,
    "text": "important there so that those code"
  },
  {
    "startMs": 227480,
    "text": "reviews are able to basically"
  },
  {
    "startMs": 230040,
    "text": "understand how the operations of the"
  },
  {
    "startMs": 231600,
    "text": "business are so it knows the business"
  },
  {
    "startMs": 233120,
    "text": "logic and more."
  },
  {
    "startMs": 234760,
    "text": "And then finally, if your hope is to"
  },
  {
    "startMs": 236720,
    "text": "move all the way out of the loop, you're"
  },
  {
    "startMs": 238360,
    "text": "like background agents, get it done,"
  },
  {
    "startMs": 240200,
    "text": "make no mistakes, you really need to"
  },
  {
    "startMs": 242240,
    "text": "make sure that you have a context engine"
  },
  {
    "startMs": 243720,
    "text": "so those agents can query it and get all"
  },
  {
    "startMs": 245520,
    "text": "the answers they need so they can keep"
  },
  {
    "startMs": 247000,
    "text": "operating in an effective way."
  },
  {
    "startMs": 250880,
    "text": "There are some common approaches that"
  },
  {
    "startMs": 252600,
    "text": "don't work. They're basically like a"
  },
  {
    "startMs": 254360,
    "text": "local maxima."
  },
  {
    "startMs": 255920,
    "text": "Two of the ones we see the most with our"
  },
  {
    "startMs": 257359,
    "text": "hundreds of enterprise clients and"
  },
  {
    "startMs": 258880,
    "text": "mid-market size businesses"
  },
  {
    "startMs": 260799,
    "text": "is the curated context trap. If you've"
  },
  {
    "startMs": 263280,
    "text": "ever sat down and taken a virtual file"
  },
  {
    "startMs": 264919,
    "text": "system or maybe a local file system, you"
  },
  {
    "startMs": 267080,
    "text": "put some markdown files in it and you're"
  },
  {
    "startMs": 268520,
    "text": "like, here's all the context of this"
  },
  {
    "startMs": 269680,
    "text": "project, it's how it works. You then"
  },
  {
    "startMs": 271480,
    "text": "allow your agent to grep over that and"
  },
  {
    "startMs": 273480,
    "text": "it gets a bunch of good data and then it"
  },
  {
    "startMs": 274919,
    "text": "will perform better."
  },
  {
    "startMs": 276400,
    "text": "The issue is first, now you have to"
  },
  {
    "startMs": 278880,
    "text": "distribute that so maybe you throw it up"
  },
  {
    "startMs": 280160,
    "text": "in a GitHub and your team can grab it."
  },
  {
    "startMs": 282080,
    "text": "But then the next is that repo is going"
  },
  {
    "startMs": 284520,
    "text": "to rot just like all the other docs you"
  },
  {
    "startMs": 286320,
    "text": "wrote down and then who at your org is"
  },
  {
    "startMs": 288800,
    "text": "the omnipotent one who has the taste to"
  },
  {
    "startMs": 291040,
    "text": "curate this file or repo for literally"
  },
  {
    "startMs": 293760,
    "text": "everyone in the org. So you start to hit"
  },
  {
    "startMs": 295600,
    "text": "these issues."
  },
  {
    "startMs": 297040,
    "text": "The next is the MCP plateau."
  },
  {
    "startMs": 299520,
    "text": "This one is pretty clear. We have MCPs,"
  },
  {
    "startMs": 302400,
    "text": "they're great. You can give it to your"
  },
  {
    "startMs": 303880,
    "text": "agent and now it can basically get"
  },
  {
    "startMs": 305320,
    "text": "information from another source system."
  },
  {
    "startMs": 307960,
    "text": "The problem is, of course, based on how"
  },
  {
    "startMs": 310600,
    "text": "you write the server description, the"
  },
  {
    "startMs": 312600,
    "text": "tool descriptions, your agent may never"
  },
  {
    "startMs": 314440,
    "text": "call it even though it should have. Or"
  },
  {
    "startMs": 316680,
    "text": "if it does, there's a known bias called"
  },
  {
    "startMs": 318600,
    "text": "the satisfaction of search bias. What"
  },
  {
    "startMs": 320880,
    "text": "that means is the agent, when it finds"
  },
  {
    "startMs": 322520,
    "text": "the first piece of information that it"
  },
  {
    "startMs": 324080,
    "text": "thinks is correct, it goes, \"Oh, I have"
  },
  {
    "startMs": 326200,
    "text": "what I need.\" and it proceeds."
  },
  {
    "startMs": 328400,
    "text": "In most organizations, there's a Slack"
  },
  {
    "startMs": 330200,
    "text": "conversation from last night that says"
  },
  {
    "startMs": 332240,
    "text": "you should be doing A instead of doing B"
  },
  {
    "startMs": 334720,
    "text": "and the agent will never find it if it"
  },
  {
    "startMs": 336240,
    "text": "found some architecture record first."
  },
  {
    "startMs": 338800,
    "text": "So it doesn't actually consider all of"
  },
  {
    "startMs": 340480,
    "text": "the context."
  },
  {
    "startMs": 342919,
    "text": "The problem here is access to"
  },
  {
    "startMs": 344680,
    "text": "information is not understanding."
  },
  {
    "startMs": 347080,
    "text": "So to deliver understanding to a model,"
  },
  {
    "startMs": 349200,
    "text": "you have to do other techniques."
  },
  {
    "startMs": 352400,
    "text": "What I'm basically trying to say is"
  },
  {
    "startMs": 354360,
    "text": "what your agent can't see is everything"
  },
  {
    "startMs": 355800,
    "text": "below the waterline. It can 100% get"
  },
  {
    "startMs": 358480,
    "text": "code that compiles, but that code that"
  },
  {
    "startMs": 360480,
    "text": "compiles is taking down prod and you"
  },
  {
    "startMs": 362080,
    "text": "have a P0 at 1:00 in the morning."
  },
  {
    "startMs": 364360,
    "text": "Because it missed the fact that you have"
  },
  {
    "startMs": 366000,
    "text": "a certain rollout procedure, you're"
  },
  {
    "startMs": 367920,
    "text": "supposed to turn off a feature flag,"
  },
  {
    "startMs": 369760,
    "text": "whatever it might be."
  },
  {
    "startMs": 372920,
    "text": "So, your team needs a context engine"
  },
  {
    "startMs": 374440,
    "text": "because what it should do is understand"
  },
  {
    "startMs": 376800,
    "text": "who you are and where you work in an"
  },
  {
    "startMs": 378200,
    "text": "organization. So, if I say to you, I"
  },
  {
    "startMs": 380760,
    "text": "want to get off stood up, it knows where"
  },
  {
    "startMs": 382840,
    "text": "I work, it knows where my get commits"
  },
  {
    "startMs": 384800,
    "text": "are, it knows who reviews those commits,"
  },
  {
    "startMs": 387200,
    "text": "and it understands that my context, it"
  },
  {
    "startMs": 389280,
    "text": "can focus me, and then use that as a"
  },
  {
    "startMs": 391560,
    "text": "trigger point to find the rest of the"
  },
  {
    "startMs": 392880,
    "text": "information."
  },
  {
    "startMs": 394080,
    "text": "It resolves conflicts, as mentioned, an"
  },
  {
    "startMs": 396160,
    "text": "old architecture diagram and last"
  },
  {
    "startMs": 397960,
    "text": "night's Slack convo with the CTO,"
  },
  {
    "startMs": 400240,
    "text": "which one is right? You need to use a"
  },
  {
    "startMs": 402040,
    "text": "bunch of techniques to discern determine"
  },
  {
    "startMs": 404200,
    "text": "that."
  },
  {
    "startMs": 405240,
    "text": "Respects permissions and governance, of"
  },
  {
    "startMs": 406720,
    "text": "course. MCP allows us to use OAuth and"
  },
  {
    "startMs": 409120,
    "text": "other scopes and SSO, but if someone"
  },
  {
    "startMs": 411480,
    "text": "asks a question over here who's not"
  },
  {
    "startMs": 412800,
    "text": "supposed to know about secret project A,"
  },
  {
    "startMs": 416040,
    "text": "you need to make sure that doesn't leak"
  },
  {
    "startMs": 417360,
    "text": "into the response."
  },
  {
    "startMs": 419280,
    "text": "And then finally, deliver the right"
  },
  {
    "startMs": 420640,
    "text": "context at the right time to the model"
  },
  {
    "startMs": 422320,
    "text": "in a token optimized way."
  },
  {
    "startMs": 424400,
    "text": "We have multiple surface areas because"
  },
  {
    "startMs": 425880,
    "text": "human engineers still talk to Unblocked"
  },
  {
    "startMs": 427400,
    "text": "all the time to get information they"
  },
  {
    "startMs": 428600,
    "text": "need in Slack or otherwise, but then you"
  },
  {
    "startMs": 431120,
    "text": "want token optimized responses if you're"
  },
  {
    "startMs": 433080,
    "text": "just speaking machine to machine in"
  },
  {
    "startMs": 434680,
    "text": "order to not waste a bunch of bold"
  },
  {
    "startMs": 436080,
    "text": "classes on your token spend."
  },
  {
    "startMs": 439760,
    "text": "This is how an engine works. I'm going"
  },
  {
    "startMs": 441440,
    "text": "to be brief on this, but basically on"
  },
  {
    "startMs": 443440,
    "text": "the left-hand side,"
  },
  {
    "startMs": 445200,
    "text": "you see all the data sources that are"
  },
  {
    "startMs": 446640,
    "text": "coming in."
  },
  {
    "startMs": 447919,
    "text": "For us, we focus on engineering teams"
  },
  {
    "startMs": 449480,
    "text": "and that's who uses us, as well as the"
  },
  {
    "startMs": 451280,
    "text": "technically light teams around it, like"
  },
  {
    "startMs": 453280,
    "text": "support, sales, and otherwise."
  },
  {
    "startMs": 455520,
    "text": "You ingest all that data, you get"
  },
  {
    "startMs": 457320,
    "text": "real-time data from tools like your"
  },
  {
    "startMs": 458840,
    "text": "instant management tool chain."
  },
  {
    "startMs": 460680,
    "text": "It comes into the engine, where that"
  },
  {
    "startMs": 462400,
    "text": "engine is, it thinks at the bottom. I'll"
  },
  {
    "startMs": 464280,
    "text": "expand on that slide in a moment. But"
  },
  {
    "startMs": 466360,
    "text": "basically it uses these six key"
  },
  {
    "startMs": 468120,
    "text": "characteristics. And then on the right,"
  },
  {
    "startMs": 470320,
    "text": "you output the context to the exact"
  },
  {
    "startMs": 472520,
    "text": "workflow in the manner that it is"
  },
  {
    "startMs": 473960,
    "text": "needed."
  },
  {
    "startMs": 476200,
    "text": "Those six key points, as mentioned,"
  },
  {
    "startMs": 478360,
    "text": "unified system context, you have to go"
  },
  {
    "startMs": 479960,
    "text": "across the whole thing. At large orgs,"
  },
  {
    "startMs": 482560,
    "text": "companies like LinkedIn scale, Workday,"
  },
  {
    "startMs": 485200,
    "text": "General Motors, whatever, they need this"
  },
  {
    "startMs": 487600,
    "text": "type of data. They need to understand"
  },
  {
    "startMs": 489240,
    "text": "everything that's happening. And Threek"
  },
  {
    "startMs": 491360,
    "text": "this morning actually talking about"
  },
  {
    "startMs": 492919,
    "text": "Fable coming out potentially later"
  },
  {
    "startMs": 494640,
    "text": "today,"
  },
  {
    "startMs": 495600,
    "text": "he mentioned that you need to actually"
  },
  {
    "startMs": 497360,
    "text": "provide a map and then let Fable"
  },
  {
    "startMs": 499560,
    "text": "discover the territory. The way to help"
  },
  {
    "startMs": 501919,
    "text": "confine that is making sure that these"
  },
  {
    "startMs": 504120,
    "text": "models have access to all of the"
  },
  {
    "startMs": 505880,
    "text": "context, because they will find your"
  },
  {
    "startMs": 508000,
    "text": "unknown unknowns."
  },
  {
    "startMs": 509840,
    "text": "There are definitely things going on in"
  },
  {
    "startMs": 511280,
    "text": "your company that you're just unaware"
  },
  {
    "startMs": 512599,
    "text": "of, but would be really helpful for the"
  },
  {
    "startMs": 514479,
    "text": "task you're trying to do."
  },
  {
    "startMs": 516919,
    "text": "That will move faster, but the targeted"
  },
  {
    "startMs": 518320,
    "text": "retrieval, you should be able to if you"
  },
  {
    "startMs": 519840,
    "text": "provide a link quickly, unfurl it, get"
  },
  {
    "startMs": 521800,
    "text": "that document back and move along. So,"
  },
  {
    "startMs": 524120,
    "text": "two tasks, deep research, go long,"
  },
  {
    "startMs": 526080,
    "text": "that's fine, but you also need speed"
  },
  {
    "startMs": 528040,
    "text": "when speed is required. Conflict"
  },
  {
    "startMs": 530040,
    "text": "resolution, we already talked about"
  },
  {
    "startMs": 531320,
    "text": "that, but one thing says do A, one thing"
  },
  {
    "startMs": 533680,
    "text": "says do B, who is right?"
  },
  {
    "startMs": 535720,
    "text": "Personalized relevance, who am I, where"
  },
  {
    "startMs": 537640,
    "text": "do I work, what am I working on?"
  },
  {
    "startMs": 539760,
    "text": "That token optimization, making sure the"
  },
  {
    "startMs": 541320,
    "text": "response is good and effective and"
  },
  {
    "startMs": 543440,
    "text": "doesn't bloat the window."
  },
  {
    "startMs": 545120,
    "text": "And then permission enforcement, of"
  },
  {
    "startMs": 546520,
    "text": "course, OAuth, you shouldn't see it, you"
  },
  {
    "startMs": 548520,
    "text": "shouldn't see it."
  },
  {
    "startMs": 550560,
    "text": "What we did with some tests is we"
  },
  {
    "startMs": 551960,
    "text": "actually ran the exact same prompt to"
  },
  {
    "startMs": 553960,
    "text": "the same model and one with context and"
  },
  {
    "startMs": 556280,
    "text": "one without. This is the wall clock time"
  },
  {
    "startMs": 558560,
    "text": "savings."
  },
  {
    "startMs": 559720,
    "text": "And then 2 hours, which is great. And"
  },
  {
    "startMs": 562240,
    "text": "then the tokens savings. So, it was a"
  },
  {
    "startMs": 564400,
    "text": "sizable task, it took about 21 million"
  },
  {
    "startMs": 567040,
    "text": "tokens without and then 18, or sorry,"
  },
  {
    "startMs": 569520,
    "text": "10.8 million tokens with it."
  },
  {
    "startMs": 572560,
    "text": "This is the type of experience that you"
  },
  {
    "startMs": 573920,
    "text": "typically see when you're using a"
  },
  {
    "startMs": 575400,
    "text": "context engine, cuz the majority of"
  },
  {
    "startMs": 577120,
    "text": "those wasted search tokens where it has"
  },
  {
    "startMs": 579200,
    "text": "to grab at the beginning of every"
  },
  {
    "startMs": 580720,
    "text": "session to understand and discover"
  },
  {
    "startMs": 582200,
    "text": "things are no longer there when it's"
  },
  {
    "startMs": 583920,
    "text": "hydrated with context. Hydrated."
  },
  {
    "startMs": 587360,
    "text": "And then, as you move forward, you get"
  },
  {
    "startMs": 589560,
    "text": "these types of outcomes."
  },
  {
    "startMs": 591200,
    "text": "50% fewer tokens, faster triage, and the"
  },
  {
    "startMs": 593760,
    "text": "answer quality is actually better"
  },
  {
    "startMs": 595360,
    "text": "because it knew what was going on inside"
  },
  {
    "startMs": 597240,
    "text": "of the business."
  },
  {
    "startMs": 599400,
    "text": "Now, this next part,"
  },
  {
    "startMs": 601120,
    "text": "you'll probably want to photo. If you"
  },
  {
    "startMs": 602640,
    "text": "don't know, you can actually take a"
  },
  {
    "startMs": 603800,
    "text": "picture of a QR code and then later in"
  },
  {
    "startMs": 605720,
    "text": "photos tap on it and then load the link"
  },
  {
    "startMs": 607680,
    "text": "so you don't need to float here cuz I'm"
  },
  {
    "startMs": 608720,
    "text": "going to give you three QR codes."
  },
  {
    "startMs": 611280,
    "text": "This first one is for the social comment"
  },
  {
    "startMs": 612840,
    "text": "network. I'll pop that up so you can"
  },
  {
    "startMs": 613920,
    "text": "take a photo."
  },
  {
    "startMs": 615160,
    "text": "But, this is an open source tool that"
  },
  {
    "startMs": 616520,
    "text": "we've got that actually, using all"
  },
  {
    "startMs": 618600,
    "text": "deterministic programming, goes over"
  },
  {
    "startMs": 620200,
    "text": "your GitHub and understands who works on"
  },
  {
    "startMs": 622200,
    "text": "your team. This is my real team. We"
  },
  {
    "startMs": 623960,
    "text": "called Rasheem the machine cuz he ships"
  },
  {
    "startMs": 626040,
    "text": "like crazy. But, on the right, you can"
  },
  {
    "startMs": 628080,
    "text": "see who he commits, where he commits,"
  },
  {
    "startMs": 629920,
    "text": "who's reviewing his work. And then in"
  },
  {
    "startMs": 631400,
    "text": "those tabs, you can find a distilled"
  },
  {
    "startMs": 633000,
    "text": "experts graph. You get full coverage of"
  },
  {
    "startMs": 635040,
    "text": "what's going on in your business. And if"
  },
  {
    "startMs": 636680,
    "text": "you optionally add one of the API keys"
  },
  {
    "startMs": 638840,
    "text": "for either OpenAI or um Anthropic, it'll"
  },
  {
    "startMs": 641680,
    "text": "um determine what your teams are by"
  },
  {
    "startMs": 643320,
    "text": "doing some labeling for you."
  },
  {
    "startMs": 645080,
    "text": "It's a really cool tool to understand"
  },
  {
    "startMs": 646400,
    "text": "where your team works and get that"
  },
  {
    "startMs": 647760,
    "text": "social network in there in order to"
  },
  {
    "startMs": 649360,
    "text": "focus the context engine if you're going"
  },
  {
    "startMs": 650800,
    "text": "to be building these tools yourself."
  },
  {
    "startMs": 653600,
    "text": "The next is called the repo rules agent."
  },
  {
    "startMs": 656080,
    "text": "This is a sample from our real code"
  },
  {
    "startMs": 658040,
    "text": "base. I'm going to pop that up anyway so"
  },
  {
    "startMs": 659600,
    "text": "you don't need to talk to the thing, but"
  },
  {
    "startMs": 661480,
    "text": "in short, what it does is discover all"
  },
  {
    "startMs": 664040,
    "text": "the places your team has written rules"
  },
  {
    "startMs": 665920,
    "text": "files, checks them all, and then tells"
  },
  {
    "startMs": 668400,
    "text": "you what severities you've given,"
  },
  {
    "startMs": 670360,
    "text": "what other things you've given. Should I"
  },
  {
    "startMs": 671480,
    "text": "just switch to this?"
  },
  {
    "startMs": 673600,
    "text": "It tells you what it Whoa, hey."
  },
  {
    "startMs": 675560,
    "text": "It's good to meet you all."
  },
  {
    "startMs": 677839,
    "text": "Basically, it will find all the rules"
  },
  {
    "startMs": 679320,
    "text": "that are inside of your repo and then"
  },
  {
    "startMs": 681280,
    "text": "tell you if you have duplicate issues or"
  },
  {
    "startMs": 683520,
    "text": "others problems and then you can grab"
  },
  {
    "startMs": 685320,
    "text": "over it as an index. So, that index can"
  },
  {
    "startMs": 687360,
    "text": "be called and you can dedupe and it'll"
  },
  {
    "startMs": 688920,
    "text": "help improve um your retrieval of"
  },
  {
    "startMs": 690720,
    "text": "context."
  },
  {
    "startMs": 692720,
    "text": "And then finally, on Monday we delivered"
  },
  {
    "startMs": 694520,
    "text": "this workshop, which was going beyond"
  },
  {
    "startMs": 696320,
    "text": "rag and taught how to build a relational"
  },
  {
    "startMs": 698760,
    "text": "context engine from scratch."
  },
  {
    "startMs": 701200,
    "text": "So, if you scan that, you'll get the"
  },
  {
    "startMs": 702440,
    "text": "full workbook. It has six PRs stacked"
  },
  {
    "startMs": 704840,
    "text": "that teach you how to walk through doing"
  },
  {
    "startMs": 706200,
    "text": "this. But in short, rag is an incredible"
  },
  {
    "startMs": 708560,
    "text": "technique and you want that. But the"
  },
  {
    "startMs": 710600,
    "text": "other half of the problem is what people"
  },
  {
    "startMs": 712520,
    "text": "actually ask is, \"What are the open PRs"
  },
  {
    "startMs": 715280,
    "text": "that I worked on in the last week with"
  },
  {
    "startMs": 717160,
    "text": "authentication?\""
  },
  {
    "startMs": 718720,
    "text": "Rag cannot answer that question alone."
  },
  {
    "startMs": 721160,
    "text": "You need queries. So, this shows you how"
  },
  {
    "startMs": 723200,
    "text": "to do a"
  },
  {
    "startMs": 724880,
    "text": "schema-less basically look up that"
  },
  {
    "startMs": 726680,
    "text": "allows the agent to discover a schema"
  },
  {
    "startMs": 729000,
    "text": "and then write queries against it"
  },
  {
    "startMs": 730280,
    "text": "deterministically in order to get that"
  },
  {
    "startMs": 732120,
    "text": "type of relational data out."
  },
  {
    "startMs": 734560,
    "text": "Very useful technique."
  },
  {
    "startMs": 737720,
    "text": "Use cases of a context engine, of"
  },
  {
    "startMs": 739240,
    "text": "course, do go beyond code generation."
  },
  {
    "startMs": 741600,
    "text": "This is, you know, where we live a lot,"
  },
  {
    "startMs": 743200,
    "text": "a lot of our customers spend their time."
  },
  {
    "startMs": 745040,
    "text": "But it's amazing to see what happens"
  },
  {
    "startMs": 746640,
    "text": "when a bunch of other people around the"
  },
  {
    "startMs": 748720,
    "text": "business start picking up these tools,"
  },
  {
    "startMs": 751000,
    "text": "customer success people solving tickets"
  },
  {
    "startMs": 753400,
    "text": "right at the time that it comes in from"
  },
  {
    "startMs": 754760,
    "text": "a customer."
  },
  {
    "startMs": 756120,
    "text": "We've got sales people closing deals"
  },
  {
    "startMs": 758200,
    "text": "earlier in their quarter because they're"
  },
  {
    "startMs": 759560,
    "text": "able to just query the Unblocked context"
  },
  {
    "startMs": 762200,
    "text": "engine on the fly while in the field."
  },
  {
    "startMs": 764800,
    "text": "And so many more."
  },
  {
    "startMs": 768040,
    "text": "What you can also do is if you saw that"
  },
  {
    "startMs": 769640,
    "text": "curve chart earlier where I talked about"
  },
  {
    "startMs": 771160,
    "text": "the levels, we've built a fun little"
  },
  {
    "startMs": 772720,
    "text": "tool where basically an LLM will quiz"
  },
  {
    "startMs": 774520,
    "text": "you and ask you about what's going on"
  },
  {
    "startMs": 776200,
    "text": "and then it will map you to exactly"
  },
  {
    "startMs": 777880,
    "text": "where you are and then tell you some"
  },
  {
    "startMs": 779600,
    "text": "techniques about how to level up through"
  },
  {
    "startMs": 781560,
    "text": "that if you are looking to basically"
  },
  {
    "startMs": 783440,
    "text": "compound your capabilities and ship with"
  },
  {
    "startMs": 785640,
    "text": "AI tools at scale. It's"
  },
  {
    "startMs": 787400,
    "text": "readiness.unblocked.com."
  },
  {
    "startMs": 791280,
    "text": "The gap is not intelligence any longer."
  },
  {
    "startMs": 793520,
    "text": "It's context. We will continue to get"
  },
  {
    "startMs": 796080,
    "text": "incredible models like Mythos as it's"
  },
  {
    "startMs": 797800,
    "text": "been grown by Anthropic and I'm sure"
  },
  {
    "startMs": 799920,
    "text": "Soul once I'm allowed to see it. I will"
  },
  {
    "startMs": 802000,
    "text": "get it. Happy Canada Day, by the way."
  },
  {
    "startMs": 804839,
    "text": "But what's happening is it's about the"
  },
  {
    "startMs": 806120,
    "text": "context you surround these models with"
  },
  {
    "startMs": 808440,
    "text": "in order for them to be effective and"
  },
  {
    "startMs": 811000,
    "text": "token efficient inside of your"
  },
  {
    "startMs": 812240,
    "text": "organization."
  },
  {
    "startMs": 815760,
    "text": "So, I have a question slide, but I'm not"
  },
  {
    "startMs": 818280,
    "text": "sure I'm allowed."
  },
  {
    "startMs": 820800,
    "text": "Nope. So, what you'll do is come meet me"
  },
  {
    "startMs": 822960,
    "text": "at booth P16. You can look for the"
  },
  {
    "startMs": 825000,
    "text": "coconut."
  },
  {
    "startMs": 826520,
    "text": "It'll be great to hang out with all of"
  },
  {
    "startMs": 827520,
    "text": "you and get into details here if you"
  },
  {
    "startMs": 828800,
    "text": "need it. Thank you for your time."
  },
  {
    "startMs": 831859,
    "text": ">> [applause]"
  },
  {
    "startMs": 846269,
    "text": "[music]"
  }
]