WEBVTT NOTE Sentence-level transcript of https://www.youtube.com/watch?v=Kz4QJmNrVXU NOTE One cue per sentence. Cue ids are the line anchors on /transcripts/Kz4QJmNrVXU.html. A cue ends where the next begins, or 2 s after its last word. s1 00:00:12.520 --> 00:00:14.960 Thank you so much for coming later in the week here. s2 00:00:14.960 --> 00:00:17.480 Uh I'm Matt. s3 00:00:17.480 --> 00:00:21.080 I'm the CEO and founder of Ref. s4 00:00:21.080 --> 00:00:28.880 The problem we work on at Ref is one you might be familiar with where individual engineers are going really fast with AI, but the team as a whole is not. s5 00:00:28.880 --> 00:00:32.000 And we're working to help close that gap. s6 00:00:32.000 --> 00:00:38.360 What I'm going to be talking about today is that is what happens when your whole team gets 10 times faster. s7 00:00:38.360 --> 00:00:40.960 All the things that don't go well. s8 00:00:40.960 --> 00:00:46.520 And my goal is to give you a blueprint for how to get get through those issues. s9 00:00:46.520 --> 00:00:51.680 Uh and have the whole team move faster together. s10 00:00:51.840 --> 00:00:58.160 The way I want to accomplish that is first talk about those issues, define some terms. s11 00:00:58.160 --> 00:01:00.040 Then we'll talk about how did we get here? s12 00:01:00.040 --> 00:01:01.600 Where are we at now? s13 00:01:01.600 --> 00:01:05.160 Um and we'll use that to triangulate on some solutions. s14 00:01:05.160 --> 00:01:12.600 We'll start high level and we'll work down to get more and more practical until we leave with a couple of things you can leave here and do immediately. s15 00:01:12.600 --> 00:01:13.920 Um how's that sound? s16 00:01:13.920 --> 00:01:14.920 I need thumbs up. s17 00:01:14.920 --> 00:01:15.960 All right, awesome. s18 00:01:15.960 --> 00:01:17.360 Great, that was great. s19 00:01:17.360 --> 00:01:22.400 Uh All right, so let's get into it. s20 00:01:22.400 --> 00:01:25.040 These are some of the problems you might be experiencing right now. s21 00:01:25.040 --> 00:01:34.480 Uh these are these are problems that affect individuals and are actually magnified at the team level once the entire team starts experiencing them. s22 00:01:34.480 --> 00:01:37.080 The first one, too many PRs to merge. s23 00:01:37.080 --> 00:01:41.760 This is the like classic first problem you hit when you start adopting AI as an engineer. s24 00:01:41.760 --> 00:01:43.080 You're like, "Great, I'm shipping stuff." s25 00:01:43.080 --> 00:01:47.400 And you push up those PRs and you're like, "There's no way I can merge all these." s26 00:01:47.400 --> 00:01:51.000 Uh and this is obviously magnified when the whole team does this. s27 00:01:51.000 --> 00:01:55.560 Merge conflicts, merge queue breaks down, things get bad. s28 00:01:55.680 --> 00:01:59.320 The second problem is that you're moving in many directions at once. s29 00:01:59.320 --> 00:02:02.120 This is also both individual and a team problem. s30 00:02:02.120 --> 00:02:05.120 Individually, this is where you have a bunch of agents doing different things. s31 00:02:05.120 --> 00:02:09.640 You're trying to remember who's doing what and your brain gets fried. s32 00:02:09.640 --> 00:02:15.280 At the organizational level, this is your engineers picking up things and running in a certain direction. s33 00:02:15.280 --> 00:02:17.440 Another engineer is running over here. s34 00:02:17.440 --> 00:02:25.320 Maybe some are bumping into each other and you're you're not moving cohesively with focus because you're just sprinting in all sorts of directions. s35 00:02:25.800 --> 00:02:29.680 The third problem you might be experiencing is declaring agent bankruptcy. s36 00:02:29.680 --> 00:02:34.280 This is a a common pattern I see engineers get into where you know, you're cranking. s37 00:02:34.280 --> 00:02:35.920 You have like your 12 terminals open. s38 00:02:35.920 --> 00:02:38.720 You're like at the end of the day you're like, "Yeah, I did a lot of work." s39 00:02:38.720 --> 00:02:40.680 You step away from your laptop. s40 00:02:40.680 --> 00:02:43.320 Uh spend time with your friends and family. s41 00:02:43.320 --> 00:02:47.080 Next morning you come back and it's like walking into just a room of strangers. s42 00:02:47.080 --> 00:02:48.120 Like, who are these people? s43 00:02:48.120 --> 00:02:49.440 What are they doing here? s44 00:02:49.440 --> 00:02:51.240 But like no problem. s45 00:02:51.240 --> 00:02:55.720 They're agents so you just like get rid of them and start over again. s46 00:02:55.720 --> 00:03:03.280 The problem though is you know, it feels like you're doing a lot of work but you're doing the same work and you're spending tokens twice. s47 00:03:03.280 --> 00:03:07.040 You're doing the same problems over over again. s48 00:03:07.080 --> 00:03:14.600 And if you think about that organizationally, that's your team not being efficient with both their time and their their token resources. s49 00:03:16.040 --> 00:03:19.880 Problem four though is actually the most important one. s50 00:03:19.880 --> 00:03:22.840 It's critical decisions being made by agents. s51 00:03:22.840 --> 00:03:32.120 Once you have agents doing a lot of your work, if if you as an engineer are letting an agent make a critical decision, you are seeding control of your code. s52 00:03:32.120 --> 00:03:34.360 You are no longer the owner of that code. s53 00:03:34.360 --> 00:03:35.920 The agent is. s54 00:03:35.920 --> 00:03:44.880 And if you imagine that at scale at your company, if the engineers across your team are you know, giving up ownership of the code, you no longer own the product. s55 00:03:46.480 --> 00:03:51.120 These are a bunch of problems you might be experiencing at different degrees at different days. s56 00:03:51.120 --> 00:03:56.280 Uh The way I like bundle it up is into this term velocity sickness. s57 00:03:56.280 --> 00:04:02.000 This is uh the stress caused by sudden output increases thanks to AI. s58 00:04:02.000 --> 00:04:05.320 Um it affects individuals or teams. s59 00:04:05.320 --> 00:04:09.080 Um and the result is output without impact. s60 00:04:09.080 --> 00:04:11.959 So, this is that feeling of like we're moving really fast. s61 00:04:11.959 --> 00:04:13.160 This should feel great. s62 00:04:13.160 --> 00:04:16.600 This should feel awesome, but for some reason it doesn't feel awesome. s63 00:04:16.600 --> 00:04:20.920 We're not having the like the things you expect to be happening are not happening. s64 00:04:20.920 --> 00:04:24.760 Um despite this feeling of being so productive. s65 00:04:26.080 --> 00:04:30.000 So, that's to define that term, but I want to tell uh a little story. s66 00:04:30.000 --> 00:04:33.600 Uh my talk is happening now. s67 00:04:37.240 --> 00:04:41.560 I want to tell a little story about somebody like to make this very real. s68 00:04:41.560 --> 00:04:49.760 Um This is a story about somebody uh who's actually not an engineer, but I think it parallels our engineering workflows a lot. s69 00:04:49.760 --> 00:04:54.960 I Part of my job is I get to go, find, and talk to people who are really pushing the forefront and try and learn from them. s70 00:04:54.960 --> 00:04:56.280 And I love that part of it. s71 00:04:56.280 --> 00:05:00.440 And so, this is somebody and I'm I'm talking through their identical workflows. s72 00:05:00.440 --> 00:05:13.120 Um and this is somebody who writes a newsletter, and their workflow is around how do I I have my ideas and then they're doing research and exploration and then uh like cohesion between those ideas s73 00:05:13.120 --> 00:05:14.840 and editorial to make sure it's in their voice. s74 00:05:14.840 --> 00:05:21.200 And they're they're walking me through their system for how they they manage this whole pipeline um and really scale out their their efficacy. s75 00:05:21.200 --> 00:05:23.880 And this is like very very much not slop. s76 00:05:23.880 --> 00:05:29.520 This is somebody who has is using agents to amplify their own voice uh in a way that's like very impressive. s77 00:05:29.520 --> 00:05:30.840 I'm like, "Wow, that's that is so cool. s78 00:05:30.840 --> 00:05:33.800 I'm I'm like I'm soaking it in." s79 00:05:33.800 --> 00:05:39.360 Uh And then they're like, "Yeah, I'm basically writing a book every week." s80 00:05:39.360 --> 00:05:40.760 And I'm like, "Oh, okay. s81 00:05:40.760 --> 00:05:44.640 Is like is your audience reading a book every week?" s82 00:05:44.640 --> 00:05:46.040 And they're like, "No, they're probably not." s83 00:05:46.040 --> 00:05:46.200 Right? s84 00:05:46.200 --> 00:05:46.880 They're not. s85 00:05:46.880 --> 00:05:53.680 They're they're This is a person who's writing a lot, but those pages that they're writing are going unread. s86 00:05:53.680 --> 00:06:04.520 Um I think that's part of this experience of velocity sickness is we're we're building things that are not mattering for the people we want them to matter for, the people we're trying to reach. s87 00:06:04.520 --> 00:06:09.760 Uh So, what we Instead of unread pages, um what we want is we want to write words that matter. s88 00:06:09.760 --> 00:06:12.000 We want to write words that connect with people. s89 00:06:12.000 --> 00:06:20.000 Um and the parallel to us as like software builders and product builders is is that we want to write we want to build products that connect with people. s90 00:06:20.000 --> 00:06:25.560 We want We want to build products that change the way people live and work and make their lives better. s91 00:06:25.560 --> 00:06:29.440 And we have more ability to do that than ever with AI. s92 00:06:29.440 --> 00:06:35.560 But we have this like feeling of velocity sickness where we're It feels like we should be doing that, but it's not quite landing. s93 00:06:35.800 --> 00:06:38.160 So, let's talk about how we got here. s94 00:06:38.160 --> 00:06:44.440 Um This is what the software engineering process used to look like before AI. s95 00:06:44.440 --> 00:06:46.200 We'd do some planning up front. s96 00:06:46.200 --> 00:06:47.040 We'd sit down and build. s97 00:06:47.040 --> 00:06:47.680 We'd implement. s98 00:06:47.680 --> 00:06:48.320 It'd be It'd be iterative. s99 00:06:48.320 --> 00:06:48.880 We'd be exploring. s100 00:06:48.880 --> 00:06:52.800 But largely we're sitting down building in isolation, implementing something. s101 00:06:52.800 --> 00:06:55.800 And at the end we sort of polish it up and ship it out the door. s102 00:06:55.800 --> 00:06:56.840 And this was great. s103 00:06:56.840 --> 00:06:57.600 We all knew how to do this. s104 00:06:57.600 --> 00:06:59.160 We had a lot of systems for this. s105 00:06:59.160 --> 00:07:00.360 Uh And it was good. s106 00:07:00.360 --> 00:07:01.919 And then AI came along. s107 00:07:01.919 --> 00:07:06.720 Um But at this time our our tools were built for this. s108 00:07:06.720 --> 00:07:11.320 Like all all our history of coding tools were built for this style of work. s109 00:07:11.320 --> 00:07:22.160 Um our IDE, our workhorse, um it was built for implementation and polish to be done by an individual, to be heads down building as a software engineer writing code. s110 00:07:22.160 --> 00:07:27.240 Um That's what That's a tool built for how we used to work. s111 00:07:27.240 --> 00:07:30.280 Um so, let's look at how how we work now. s112 00:07:30.280 --> 00:07:32.640 Our work looks a lot more like this. s113 00:07:32.640 --> 00:07:34.640 Where you do some planning up front. s114 00:07:34.640 --> 00:07:37.200 You start thinking about what am I trying to do here? s115 00:07:37.200 --> 00:07:40.160 You take this idea in your head and start to flesh it out. s116 00:07:40.160 --> 00:07:45.320 Um At some point an agent takes that idea and like implements it. s117 00:07:45.320 --> 00:07:48.960 This is like arguably should not even be on this slide cuz it's not our human work anymore. s118 00:07:48.960 --> 00:07:50.400 It's done by the agent. s119 00:07:50.400 --> 00:07:58.000 Um and then in the end we do some polishing where we take back that thing the agent has made for us and we like hold it in our hands and we say, is this s120 00:07:58.000 --> 00:07:59.440 is this what I wanted? s121 00:07:59.440 --> 00:08:02.400 Um This is a very different shape of work. s122 00:08:02.400 --> 00:08:03.440 Right? s123 00:08:03.440 --> 00:08:07.200 We're no longer doing this like heads down building. s124 00:08:07.200 --> 00:08:12.320 These are the two uh creative and collaborative parts of our work as engineers. s125 00:08:12.320 --> 00:08:15.880 This is where we like express our craft as an engineer. s126 00:08:15.880 --> 00:08:18.520 And the one that's most different is the the planning stage. s127 00:08:18.520 --> 00:08:21.200 So, I want to zoom in on that just a little bit. s128 00:08:21.200 --> 00:08:28.800 Uh that's the like exploratory, creative, collaborative part where we're we're thinking about like I have this vague idea. s129 00:08:28.800 --> 00:08:30.000 I have this complex system. s130 00:08:30.000 --> 00:08:36.000 I need to understand the contours of this system, apply this idea, and like really understand it. s131 00:08:36.000 --> 00:08:40.440 And I need to pull out what's relevant and express my taste as an engineer. s132 00:08:40.440 --> 00:08:42.960 As to like, where do I want this system to go? s133 00:08:42.960 --> 00:08:48.160 Um This is the kind of work that uh that's creative and and needs to be done together. s134 00:08:48.160 --> 00:08:52.880 I think the the shape of engineering teams and product teams is changing with AI. s135 00:08:52.880 --> 00:09:00.880 But we're always going to have this thing where we have a group of people responsible for managing a complex system and it's sort of deciding the future of that system. s136 00:09:00.880 --> 00:09:03.040 And that's this planning work. s137 00:09:03.040 --> 00:09:09.240 And so, we're we're starting to see some of the hints of like, okay, we had tools built for a certain type of work. s138 00:09:09.240 --> 00:09:11.240 Our work looks different now. s139 00:09:11.240 --> 00:09:14.760 Um Let's start to see think about that a little bit more. s140 00:09:14.760 --> 00:09:17.839 So, uh we used to have the IDE as our main tool. s141 00:09:17.839 --> 00:09:19.880 It was used for implementation. s142 00:09:19.880 --> 00:09:21.839 But now we're doing this different kind of work. s143 00:09:21.839 --> 00:09:26.480 So, at the very high level, we can start to think about solutions like, okay, we have a tool built for this type of work. s144 00:09:26.480 --> 00:09:27.400 We have a new type of work. s145 00:09:27.400 --> 00:09:31.000 Let's think about you know, what this new layer of our work is. s146 00:09:31.000 --> 00:09:32.760 This This is the decision layer. s147 00:09:32.760 --> 00:09:43.360 This is where we're thinking through what are the key decisions, doing that like craft of engineering, um and expressing our taste as engineers, ultimately a different thing than implementation. s148 00:09:43.360 --> 00:09:46.800 And it it's a different gear as an engineer. s149 00:09:46.800 --> 00:09:49.320 The skill now is what gear am I in? s150 00:09:49.320 --> 00:09:55.640 Am I using the appropriate tools for the gear that I'm what I'm trying to accomplish right now. s151 00:09:56.120 --> 00:10:00.880 So let's think about what a tool built for the decision layer would look like. s152 00:10:01.600 --> 00:10:04.800 It'd be a tool built for docs and not chat. s153 00:10:04.800 --> 00:10:06.720 And there's that's like a short sentence. s154 00:10:06.720 --> 00:10:08.520 There's a lot to unpack here though. s155 00:10:08.520 --> 00:10:11.120 So I'm going to spend a lot of time on this slide. s156 00:10:11.120 --> 00:10:17.720 The problem with chats is that they are the relic of building for implementation. s157 00:10:17.720 --> 00:10:23.720 So they're they're default isolated and ephemeral and and brain off. s158 00:10:23.720 --> 00:10:27.080 They're they're made to build things and get stuff done. s159 00:10:27.080 --> 00:10:30.680 And that's not really the same type of work we're doing at the decision layer. s160 00:10:30.680 --> 00:10:33.320 We're doing this creative exploratory work. s161 00:10:33.320 --> 00:10:45.200 So being in this isolated environment where I'm working with an agent, maybe I start with my vague idea and I'm exploring it and asking questions, but decisions are being made in that like that chat that are are not shared with my team s162 00:10:45.200 --> 00:10:54.880 that are going to disappear as as long I'm going to result in some code being output where those important decisions are not being made clear and shared with the team. s163 00:10:54.880 --> 00:10:58.960 And you're also in this mode where the agent saying Okay, this is what I want to do. s164 00:10:58.960 --> 00:10:59.920 Is is that okay? s165 00:10:59.920 --> 00:11:00.560 Let's go. s166 00:11:00.560 --> 00:11:07.760 Or or sometimes it'll say it'll ask you a question and it'll be like you know, this is the recommended option and then you're like, great. s167 00:11:07.760 --> 00:11:08.600 I don't even think about this. s168 00:11:08.600 --> 00:11:11.720 I'll just hit that one and we keep going. s169 00:11:12.600 --> 00:11:13.800 Docs start to solve this problem. s170 00:11:13.800 --> 00:11:23.200 So when you're if you center your work around working in docs they're meant to be bring forward the key decisions. s171 00:11:23.200 --> 00:11:25.280 This is this is what our work is now. s172 00:11:25.280 --> 00:11:34.800 Our work now is figure out what decisions matter and then make those decisions and then get out of the way while the agents fill in the rest. s173 00:11:35.600 --> 00:11:40.040 Working in Docs is the the classic way we would create alignment. s174 00:11:40.120 --> 00:11:50.560 If you were If you think back to being a manager or a lead on a team before AI, if your team was having struggling with alignment, you would not tell them like, "Let's go all work in Slack DMs. s175 00:11:50.560 --> 00:11:52.000 Let's like go direct message each other." s176 00:11:52.000 --> 00:12:02.520 You'd say, "Let's like bring forward key decisions, align on them, spend time on these, like find a way that we can really spend time getting our decisions right." s177 00:12:03.840 --> 00:12:08.880 So, there are two things you might be thinking looking at this that are not what I'm talking about here. s178 00:12:08.880 --> 00:12:10.840 The first one is plan mode. s179 00:12:10.840 --> 00:12:12.560 That's a great tool. s180 00:12:12.560 --> 00:12:15.760 The other one is like full-on factory spectrum and development. s181 00:12:15.760 --> 00:12:17.160 That's a great tool, too. s182 00:12:17.160 --> 00:12:19.880 I'm talking about something in the middle. s183 00:12:19.880 --> 00:12:30.800 Plan mode is great, but it's largely a a rich chat message where the agent is saying, "Hey, here's a like better visualization of what I'm trying to express to you." s184 00:12:30.800 --> 00:12:32.280 Yeah, that's great. s185 00:12:32.280 --> 00:12:35.040 But it's still in this isolated ephemeral environment. s186 00:12:35.040 --> 00:12:42.520 And what I'm suggesting is something more more durable, more shared, more long-lived that you and your team are spending time on. s187 00:12:42.520 --> 00:12:52.080 Similarly, the spectrum and development where we just define the behaviors, we operate at the like product level, is a little far away from the engineering reality. s188 00:12:52.080 --> 00:13:01.640 The engineering reality is that I need to understand my system and have a tool that like helps me understand that system and lay out those key decisions in a in a technical sense. s189 00:13:01.640 --> 00:13:09.320 The way I like to conceptualize this is as the portal to the software system, where you are like Tony Stark and you're like, "Show me what matters." s190 00:13:09.320 --> 00:13:10.760 And you're like, "I'm working on this. s191 00:13:10.760 --> 00:13:12.240 Pull out the bits that are relevant." s192 00:13:12.240 --> 00:13:19.280 And AI is amazing at finding things that are related to other things, helping you find what's what's relevant and lay them out on the table in front of you. s193 00:13:19.280 --> 00:13:24.400 Organize the pieces in the way you want to represent how you want the system to grow. s194 00:13:25.839 --> 00:13:30.640 But the big conceptual flip here is actually that we're pulling out the state. s195 00:13:30.640 --> 00:13:40.160 So, when you're living and working in a a long-lived session with an agent, there's this implicit context being built up like over that work, and that's great. s196 00:13:40.160 --> 00:13:44.840 Um But there's you're also doing actions and it's it's not shared. s197 00:13:44.840 --> 00:13:51.160 What you want is to separate the the agent as the action and the doc as the state. s198 00:13:51.160 --> 00:14:01.440 And so, you can spawn new agents that have the same context or starting from the same place that are able to collaborate and work on the same same piece of context and state. s199 00:14:01.440 --> 00:14:08.880 You're ultimately doing context engineering in this doc, so that every agent is largely stateless and starts from this place um the same place. s200 00:14:08.880 --> 00:14:15.800 And what that gives you is your team and yourself can look into this and understand what actually is in here. s201 00:14:15.800 --> 00:14:22.440 What are the decisions that are being made um and have a a clear understanding of what the key decisions are. s202 00:14:23.320 --> 00:14:29.360 So, this is a a different way of working where your your core atom of your work is a doc rather than a chat. s203 00:14:29.360 --> 00:14:33.880 Um What happens when you actually start to implement this? s204 00:14:33.880 --> 00:14:39.720 Well, the first thing we see happen actually is that people start to plan and then not implement their plan. s205 00:14:39.720 --> 00:14:43.200 Uh and this is actually like a really good sign. s206 00:14:43.200 --> 00:14:45.400 Because what that means is they're thinking through ideas. s207 00:14:45.400 --> 00:14:46.280 They're saying, "I have this idea. s208 00:14:46.280 --> 00:14:47.839 Let me explore it. s209 00:14:47.839 --> 00:14:48.680 Let me start to flesh it out. s210 00:14:48.680 --> 00:14:50.240 Like I have this vague thought. s211 00:14:50.240 --> 00:14:50.960 Don't just give me some code. s212 00:14:50.960 --> 00:14:56.200 Don't go off and like build it for me, but let me help help me understand the idea that I'm talking about. s213 00:14:56.200 --> 00:15:01.480 Um And then they have a bunch of these and then some of them are getting built and some of them aren't. s214 00:15:01.480 --> 00:15:10.200 So, that means they're prioritizing the ideas that after they've explored them are the ones that are worth building and and going in this to the next step with. s215 00:15:10.200 --> 00:15:16.880 Um One way I like to frame this is that you're you're shifting from code velocity to idea velocity. s216 00:15:16.880 --> 00:15:23.000 So, going back to our our problem statement like how are we dealing with velocity sickness? s217 00:15:23.000 --> 00:15:26.680 The velocity sickness is we're shipping too much code that's not going anywhere. s218 00:15:26.680 --> 00:15:39.360 The solution is to shift that velocity to ideas so that rather than you know, getting stuck in prototype gravity where we we build something and we're so excited to just ship that thing s219 00:15:39.360 --> 00:15:48.960 and we're going down like one path of the idea maze, we can now like more more effectively explore that whole maze and find that gold that's around the corner. s220 00:15:48.960 --> 00:15:54.640 Um and and really impact the people we're trying to help. s221 00:15:55.720 --> 00:15:57.000 So, let's go back. s222 00:15:57.000 --> 00:16:00.640 Here are those problems, those velocity sickness problems I talked about the beginning. s223 00:16:00.640 --> 00:16:03.760 Let's see how this starts to address those. s224 00:16:03.960 --> 00:16:07.080 First, too many PRs. s225 00:16:07.080 --> 00:16:10.320 We've moved the the review point earlier into the process. s226 00:16:10.320 --> 00:16:13.760 So, we're aligning on the key decisions up front. s227 00:16:13.760 --> 00:16:18.960 That means the code review is easier because the hardest part of any code review is, you know, what actually matters here. s228 00:16:18.960 --> 00:16:22.160 That's the first step is like, "Hmm, what do I care about here?" s229 00:16:22.160 --> 00:16:26.600 If you move that earlier, we've aligned on that, the code review becomes much simpler. s230 00:16:27.040 --> 00:16:29.040 Number two, moving in too many directions. s231 00:16:29.040 --> 00:16:36.040 Again, we're aligning early as a team and individually, I'm like understanding what I'm working on across many agents. s232 00:16:36.040 --> 00:16:40.880 If I'm working on a large thing, I've I've understood it initially. s233 00:16:40.880 --> 00:16:43.240 So, as a team, we're sharing these plans and we're aligning. s234 00:16:43.240 --> 00:16:54.760 This is what we're trying to build and it's easy before someone has spent even a day in AI going deep on some idea building a prototype, we can talk about it early and make sure we're aligned on where we actually taking the system. s235 00:16:54.960 --> 00:16:58.960 Um declaring agent bankruptcy is just not a thing because you've made your agent stateless. s236 00:16:58.960 --> 00:17:01.800 So, the result of their work is in the stock. s237 00:17:01.800 --> 00:17:05.000 If you need to rebuild your human context, you just read the doc. s238 00:17:05.000 --> 00:17:10.680 Now, you understand the state of this project and you can pick up from there. s239 00:17:10.680 --> 00:17:13.920 And the most important one, humans own the decisions. s240 00:17:13.920 --> 00:17:15.720 Like, that's what we're solving for. s241 00:17:15.720 --> 00:17:18.720 Humans need to own the decisions. s242 00:17:18.800 --> 00:17:28.600 That's how we retain ownership of our software and our products and make it like a true expression of what we're trying to create in the world. s243 00:17:29.440 --> 00:17:30.840 There are also some other benefits. s244 00:17:30.840 --> 00:17:37.000 So, when you have this state extracted and you're working from a shared context, you get more parallel agents. s245 00:17:37.000 --> 00:17:39.280 It's easier to work with parallel agents. s246 00:17:39.280 --> 00:17:41.000 Uh you get this durable decision log. s247 00:17:41.000 --> 00:17:43.760 So, something uh that's really powerful. s248 00:17:43.760 --> 00:17:49.280 I think a lot of people are thinking about how do we capture all the decisions going into these sessions? s249 00:17:49.400 --> 00:17:58.560 A really great solution to that is let's pull out all the decisions up front and agree to them and put them in a place that's durable so that we don't have to have like some LLM summarizing s250 00:17:58.560 --> 00:18:01.040 it and maybe picking the wrong things later on. s251 00:18:01.040 --> 00:18:05.560 We want to bring that forward so we can save it and refer to it later. s252 00:18:05.560 --> 00:18:10.680 Um and then my favorite one um is actually more collaboration. s253 00:18:10.680 --> 00:18:17.840 Like this process, we want we're doing more work that is creative, which means we should be collaborating more as engineers. s254 00:18:17.840 --> 00:18:21.000 I think the future of engineering is multiplayer. s255 00:18:21.000 --> 00:18:25.160 It's going to be multi multiplayer by default sooner than we think. s256 00:18:25.160 --> 00:18:31.400 Um and it's more important than ever because we're in this moment with AI where everything's moving faster than ever. s257 00:18:31.400 --> 00:18:35.160 There's a pressure like everyone's company feels existential, small or big. s258 00:18:35.160 --> 00:18:36.560 You need to deliver. s259 00:18:36.560 --> 00:18:39.320 And the way we get through that as humans is by working together. s260 00:18:39.320 --> 00:18:42.520 And we need tools that help us do that. s261 00:18:43.120 --> 00:18:48.320 So, here are three concrete things you can leave this room and do right now. s262 00:18:48.320 --> 00:18:52.640 The first one is to think of your work in terms of planning and polish. s263 00:18:52.640 --> 00:18:55.200 So, recognizing that there are two gears I'm working in. s264 00:18:55.200 --> 00:18:57.480 There's no longer this one focus of implementation. s265 00:18:57.480 --> 00:19:01.680 There's uh plan and then polish. s266 00:19:01.680 --> 00:19:07.560 Um and notice when you're in a single session and you're doing both of these in one session. s267 00:19:07.560 --> 00:19:16.040 Notice when you drift from the planning phase into the polish phase and is your tool serving you for what you're trying to do at that moment? s268 00:19:16.040 --> 00:19:20.760 Uh number two is to start to treat your plan as a portal to the software system. s269 00:19:20.760 --> 00:19:27.520 So, really treating it as this powerful, malleable tool to say like, what matters to to me for what I'm working on right now? s270 00:19:27.520 --> 00:19:32.960 Um and asking it to show that to you so that you can make the best decisions possible. s271 00:19:33.400 --> 00:19:35.200 And number three is share a plan. s272 00:19:35.200 --> 00:19:40.440 Like, don't just you know, write the plan, give it to your agent, and have them implement it. s273 00:19:40.440 --> 00:19:41.840 Give it to someone on your team. s274 00:19:41.840 --> 00:19:44.960 This is like, I feel like very unnatural for a lot of people. s275 00:19:44.960 --> 00:19:47.960 We always we we think we know what's going on. s276 00:19:47.960 --> 00:19:49.760 But, it's always valuable. s277 00:19:49.760 --> 00:19:50.640 You have smart teammates. s278 00:19:50.640 --> 00:19:52.600 They have great context in their heads. s279 00:19:52.600 --> 00:19:53.440 You should tap into that. s280 00:19:53.440 --> 00:19:55.360 They will give you good feedback. s281 00:19:55.360 --> 00:19:58.760 Uh it's a really valuable thing to do. s282 00:19:59.360 --> 00:20:00.040 That's my talk. s283 00:20:00.040 --> 00:20:00.800 I'm Matt. s284 00:20:00.800 --> 00:20:02.120 Um I'm the CEO of Ref. s285 00:20:02.120 --> 00:20:06.000 If you want to talk about any of this stuff, this has all my like connection information. s286 00:20:06.000 --> 00:20:08.480 Ref is a tool built for the decision layer. s287 00:20:08.480 --> 00:20:12.000 Uh we work with all of your existing implementation tools. s288 00:20:12.000 --> 00:20:16.160 Um if you want to see a demo, come you can come by our booth or just see me out there. s289 00:20:16.160 --> 00:20:19.000 Uh and I'd love to talk about this stuff, so come say hi. s290 00:20:19.000 --> 00:20:21.000 Thanks.