WEBVTT 1 00:00:00.271 --> 00:00:01.492 The hot thing right now is DOTS. 2 00:00:01.852 --> 00:00:04.533 They're persistent agents from OpenAI. 3 00:00:05.053 --> 00:00:09.654 I was there at DevDay when they launched them, I interviewed Sam about them, and I've been using DOTS since before they came out. 4 00:00:09.755 --> 00:00:12.175 And now we're into week two of DOTS. 5 00:00:12.315 --> 00:00:13.856 So how's it going? 6 00:00:14.616 --> 00:00:24.160 If you read my vibe check, you know that I thought they were a little bit rough to begin with, but that the idea of a persistent proactive agent was obviously the future. 7 00:00:25.020 --> 00:00:29.261 Even before Dots launched, they'd become my main way of interacting with ChatGPT. 8 00:00:29.882 --> 00:00:32.022 And they're still that way after launch. 9 00:00:32.302 --> 00:00:39.305 I'm really excited about the little things they've been doing to make the experience less buggy and more complete. 10 00:00:39.485 --> 00:00:42.105 And I think it'll get even better over the next couple of weeks. 11 00:00:42.846 --> 00:00:46.867 What's been really interesting, though, is seeing the team at every. 12 00:00:46.887 --> 00:00:53.289 We've got about 30 people seeing them start to adopt it and some of the use cases that have started to emerge. 13 00:00:53.389 --> 00:00:53.489 So 14 00:00:54.057 --> 00:00:58.701 In case you're looking for some use cases, I actually had my dot collect some use cases to talk to you about. 15 00:01:00.062 --> 00:01:04.945 So one thing that dots are really useful for is keeping up with school emails. 16 00:01:05.106 --> 00:01:12.111 So Kieran Klassen, who works at Every, says that it scans all of the school emails he gets from his kid's school and tells him what he needs to know. 17 00:01:12.271 --> 00:01:20.237 Another thing that dots are really good for is catching access requests or like little things, little messages you might have missed in Slack and making sure that you get them. 18 00:01:20.317 --> 00:01:20.598 So... 19 00:01:21.198 --> 00:01:31.340 Ariel Shipper, our head of operations, her dot caught a few people asking for different accounts across the company and asked if she wanted it to just approve access. 20 00:01:31.520 --> 00:01:37.422 Dots are also really good for working when you're not at your computer or your hands are full. 21 00:01:37.662 --> 00:01:44.543 So our head of video, Randy Koonsman, was running around yesterday trying to get our launch video out for the Every Agent which launched. 22 00:01:44.864 --> 00:01:51.025 And he was doing work for that launch just on the phone with his dot, talking it through different video edits and different things he needed to get done. 23 00:01:51.345 --> 00:01:55.707 I also think this is something that came up in my interview with Sam and I feel this too. 24 00:01:56.207 --> 00:02:03.991 Dots are really good for filtering out all of the noise that you don't really want to see and only showing you stuff that really matters. 25 00:02:04.091 --> 00:02:08.213 So, for example, I was doing a lot of model testing over the weekend. 26 00:02:09.705 --> 00:02:18.307 Normally model testing requires me to be in the every Slack, which once I'm in Slack, it's just like this rabbit hole of I'm just going down doing tons and tons and tons of work. 27 00:02:18.928 --> 00:02:26.650 And what was amazing about using my dot doing model testing was I was able to post the results of all the experiments I was running in Slack. 28 00:02:27.230 --> 00:02:36.792 And if someone responded, that my dot would tell me about the response, which would help me stay up to date on what the team was thinking without getting me sucked into the whole rabbit hole of everything going on in Slack. 29 00:02:37.232 --> 00:02:40.073 And that helped keep my weekend pretty clear and calm. 30 00:02:40.813 --> 00:02:43.294 And I think that's a huge, huge thing for dots. 31 00:02:43.554 --> 00:02:44.554 So what's the overall vibe? 32 00:02:45.134 --> 00:02:49.475 If you don't have a dot, I highly recommend you get one, especially if you're in the Chachapiti ecosystem. 33 00:02:49.955 --> 00:02:53.836 If you're using Muse or Instinct or GrokBots, you should check it out. 34 00:02:57.997 --> 00:03:22.443 there's something really good about having something that's hooked into the entire rest of the ecosystem that you use for your work in your life there are still rough edges though the whole permission structure of dots is like completely a mystery to me and it's extremely annoying so for example I often have to give multiple approvals to my dot to do something basic and even when I give it approval it says it can't do it so they really need to figure that stuff out to make the experience awesome but I would expect if 35 00:03:22.631 --> 00:03:26.093 in a year, this is the default way that you're interacting with all your AIs. 36 00:03:26.893 --> 00:03:30.395 You're probably not going to have it be a cute character. 37 00:03:30.575 --> 00:03:32.436 That would be my other like little hot take. 38 00:03:32.476 --> 00:03:33.217 Mine is named Boo. 39 00:03:33.237 --> 00:03:33.957 He's super cute. 40 00:03:34.417 --> 00:03:42.361 But my experience using Open Claws and other named agents like this is that as soon as something better comes along, you throw away the cute little character you develop. 41 00:03:42.521 --> 00:03:47.304 So my guess is I'll be using a persistent agent in a year, but Boo will probably be dead. 42 00:03:48.098 --> 00:03:48.579 Rip boo. 43 00:03:48.839 --> 00:03:51.462 We've also actually been working on our own persistent agent. 44 00:03:51.482 --> 00:03:53.164 That's not a personal agent. 45 00:03:53.204 --> 00:03:54.205 It's a company agent. 46 00:03:54.365 --> 00:04:05.297 And that's actually come from this whole journey we've been on starting with open clause inside of every and everybody having a personal agent and realizing how valuable it was for the company to have a personal agent. 47 00:04:05.377 --> 00:04:09.802 So we built one for ourselves and then we just released it to all of our subscribers. 48 00:04:09.982 --> 00:04:24.717 So I brought Willie Williams on, our head of platform, who was instrumental in bringing the agent to life to talk about personal agents versus company agents, what it's like to build an agent at the frontier right now, and what that means for engineering teams going forward. 49 00:04:25.277 --> 00:04:25.778 I hope you enjoy. 50 00:04:30.182 --> 00:04:31.263 Thank you. 51 00:04:39.062 --> 00:04:40.002 Willy, welcome to the show. 52 00:04:40.402 --> 00:04:41.242 Oh, thank you for having me. 53 00:04:41.863 --> 00:04:43.943 So for people who don't know, you are the head of platform. 54 00:04:44.123 --> 00:04:50.865 You are one of the driving forces behind the Every Agent, which we, well, by the time this comes out, we'll have launched, it's launching tomorrow. 55 00:04:51.385 --> 00:04:56.226 And I want to talk to you about the whole agent phenomenon. 56 00:04:56.346 --> 00:05:00.807 Because I think we're in this world right now where personal agents are the hot thing. 57 00:05:03.729 --> 00:05:15.534 And there's like a little bit of Groundhog's Day for me, I assume for you, because we went through our personal agent phase like nine months ago or eight months ago with the original open claw phase. 58 00:05:15.554 --> 00:05:17.035 And I feel like we learned a lot from that. 59 00:05:17.215 --> 00:05:25.739 And then now I feel like there's this whole new era of people who are experiencing the same thing and maybe we'll learn the same lessons. 60 00:05:25.799 --> 00:05:27.280 And we, for us, we went from 61 00:05:27.940 --> 00:05:30.642 Open Claw, everyone in the company's got a personal agent too. 62 00:05:31.022 --> 00:05:35.445 We have a company agent and we liked the company agent so much that we actually just made a company agent for other people to use. 63 00:05:35.765 --> 00:05:42.129 But I want to just talk about that and try to talk about it as honestly as we can because it's possible we're wrong. 64 00:05:42.149 --> 00:05:42.970 Yeah. 65 00:05:43.110 --> 00:05:45.451 But I do think we've learned some things that are important. 66 00:05:45.471 --> 00:05:46.092 Yeah. 67 00:05:46.672 --> 00:05:51.095 Maybe let's start with Open Claw because we actually disagreed and I think you ended up being right. 68 00:05:52.396 --> 00:05:57.299 So tell us about the Open Claw situation and sort of your take and talk about why I was wrong. 69 00:05:57.868 --> 00:06:05.495 Well, I think that the funniest thing was actually at the time, it was called Claudius. 70 00:06:06.115 --> 00:06:06.515 Claudius? 71 00:06:06.776 --> 00:06:07.176 Claudius. 72 00:06:07.556 --> 00:06:10.959 It was like the earlier version before, because it was Clawbot, right? 73 00:06:10.979 --> 00:06:11.039 Oh. 74 00:06:11.119 --> 00:06:11.360 Right? 75 00:06:11.420 --> 00:06:12.260 And even before that. 76 00:06:12.721 --> 00:06:13.201 Interesting. 77 00:06:13.221 --> 00:06:15.123 It was called... Well, Moltbot. 78 00:06:15.143 --> 00:06:18.446 There was... We forget about the Moltbot. 79 00:06:18.466 --> 00:06:20.207 The two days of Moltbot. 80 00:06:20.888 --> 00:06:23.230 Big shout out to Ash, actually, who was like... 81 00:06:23.710 --> 00:06:25.471 who I think first messaged me about it. 82 00:06:25.511 --> 00:06:28.092 He was like, Ash is one of our contributing writers. 83 00:06:28.752 --> 00:06:31.534 And he was like, dude, this thing is awesome. 84 00:06:32.354 --> 00:06:38.517 And I remember because I, I, I knew like Peter's name from like the top token burners leaderboard. 85 00:06:38.537 --> 00:06:44.480 Like he was up there and just like, I was like, Oh man, this is all, this, this folks are burning a lot of tokens. 86 00:06:44.500 --> 00:06:44.700 Right. 87 00:06:45.160 --> 00:06:49.102 And they were just like, Oh yeah, they're making, he's making this thing called Claude, Claudus. 88 00:06:49.923 --> 00:06:50.143 And he, 89 00:06:51.338 --> 00:06:52.959 I was like, oh, this is kind of a weird thing. 90 00:06:52.999 --> 00:06:54.140 You know, kind of follow the project. 91 00:06:55.281 --> 00:06:58.364 And then Ash came back and was like, oh, yeah, dude, this is really cool. 92 00:06:59.424 --> 00:07:00.525 And so we started trying it. 93 00:07:00.685 --> 00:07:03.928 And then I think Brandon was the first one who was like, okay, I've got one. 94 00:07:04.428 --> 00:07:06.730 You know, I've got one. 95 00:07:06.950 --> 00:07:08.591 I'm starting to like wire it up, right? 96 00:07:08.912 --> 00:07:09.272 R.I.P. 97 00:07:09.352 --> 00:07:09.692 Zosia. 98 00:07:10.213 --> 00:07:10.513 R.I.P. 99 00:07:10.553 --> 00:07:10.893 Zosia. 100 00:07:11.634 --> 00:07:12.575 And I think, you know. 101 00:07:13.315 --> 00:07:18.717 And then it really kicked off in January because we went to Think Week. 102 00:07:18.917 --> 00:07:24.819 And for folks who don't know, it's like our semi-annual gathering of every where we sort of like think about the future, you know. 103 00:07:25.799 --> 00:07:29.300 And that was a place where we were finally all in person. 104 00:07:29.320 --> 00:07:31.601 We're like, oh, I can show you what's going on. 105 00:07:31.661 --> 00:07:32.441 And I think there's a lot. 106 00:07:32.481 --> 00:07:34.022 This is a reoccurring theme of like. 107 00:07:34.802 --> 00:07:40.129 showing people how to use AI is like a very transparent thing, but telling them is sort of like land splat. 108 00:07:40.729 --> 00:07:45.415 And so we had a bunch of people who were open claw pilled who were like, oh yeah, this is, this is cool. 109 00:07:45.435 --> 00:07:46.075 Like, this is what I do. 110 00:07:46.095 --> 00:07:47.017 I order my groceries. 111 00:07:47.057 --> 00:07:49.800 I've got it in my group chat with my, my family, you know? 112 00:07:50.140 --> 00:07:51.080 It really spread. 113 00:07:51.120 --> 00:07:55.181 We were all in Panama together, and I remember coming out of that. 114 00:07:55.201 --> 00:08:01.422 I think there was a model drop on the day we left, and the claws just started working way better. 115 00:08:02.142 --> 00:08:10.344 And then I made this internal place for them to talk, and they started talking to each other in our Slack, and it was like, whoa, this is crazy. 116 00:08:10.364 --> 00:08:15.085 Yeah, it was like a funny thing where we were like, okay, the first... 117 00:08:15.625 --> 00:08:18.107 The first blocker is, oh, it's just a little bit too technical. 118 00:08:18.527 --> 00:08:23.311 And so we went through a couple iterations of, like, how do we do it for the team or the less technical members of the team? 119 00:08:23.831 --> 00:08:27.394 And there was, like, you know, we've got a template, so we're spinning up info for that. 120 00:08:28.114 --> 00:08:35.920 And then we made a little service so where anyone could, like, enter, like, kind of walk through people through the setup. 121 00:08:36.260 --> 00:08:36.841 And then we'd get there. 122 00:08:37.141 --> 00:08:41.724 But we were also, like, looking at it and we were like, oh, man, security is hard, you know, for all of these things. 123 00:08:42.545 --> 00:08:44.867 But, you know, we sort of, like, I think we, like, powered through. 124 00:08:45.367 --> 00:08:46.448 a lot of that. 125 00:08:47.009 --> 00:08:54.876 And I mean, I have this vivid memory of just like so many bots in our Slack. 126 00:08:55.196 --> 00:08:56.377 Like so many bots. 127 00:08:56.417 --> 00:08:58.058 Like absolute chaos. 128 00:08:58.299 --> 00:08:59.740 Because you also have more than one. 129 00:08:59.760 --> 00:09:05.705 Brendan being like just having so many bots running around doing various things. 130 00:09:06.306 --> 00:09:09.829 And I think the big takeaway was one, it was so... 131 00:09:11.757 --> 00:09:14.640 valuable for people to see other people using AI. 132 00:09:14.720 --> 00:09:23.789 And up until this point, it was like, oh, I built this gigantic Claude or Codex command center on my local machine. 133 00:09:23.829 --> 00:09:24.931 I've got it doing everything. 134 00:09:25.351 --> 00:09:28.815 But the only thing that people would see was the output of it. 135 00:09:29.415 --> 00:09:33.638 And so once we had OpenCloset Slack, it was like, oh, I didn't know I could use it that way. 136 00:09:34.018 --> 00:09:35.379 Oh, that's actually pretty cool. 137 00:09:35.399 --> 00:09:38.261 And I didn't know that you prompted in this way. 138 00:09:38.361 --> 00:09:44.184 Like there's all these little things people do when they're using their AI that you never got to see. 139 00:09:44.264 --> 00:09:47.827 And it was like, oh, I'm like learning every time I look in Slack. 140 00:09:48.507 --> 00:09:49.468 Yeah. 141 00:09:49.768 --> 00:09:55.992 And the other thing I think that was really big was, you know, I think we liked our personal bots. 142 00:09:56.212 --> 00:09:57.613 And I think this is another thing that I think 143 00:09:58.784 --> 00:10:23.711 we're seeing now again is that you sort of form you're like oh i'm a i'm attached well mine was last right i'm like attached to last right and um and what it does and it's like a customization so for folks who are like tinkers folks who are builders who like playing with these things it was like a great these setups are great you know for folks who aren't folks who are like a little bit more in terms of like operator thing it was just like oh this is 144 00:10:24.331 --> 00:10:25.912 There's a bunch of security worries. 145 00:10:26.092 --> 00:10:27.332 It's super complex. 146 00:10:27.372 --> 00:10:28.373 How do I get it set up? 147 00:10:28.433 --> 00:10:30.033 It keeps breaking. 148 00:10:30.393 --> 00:10:34.395 And then if you don't invest in it, it kind of fades. 149 00:10:35.395 --> 00:10:36.576 It doesn't keep going with the rest of it. 150 00:10:36.596 --> 00:10:52.202 And I think that for us was the striking point where it was sort of like, we want, instead of everyone investing in their own agents and making them better day-to-day over time, it's like, well, why don't we take all that energy and put it into one agent that the whole company benefits from? 151 00:10:52.512 --> 00:10:55.254 If you're enjoying this show, you should subscribe to Every. 152 00:10:55.474 --> 00:10:58.196 Every is the only subscription you need to stay at the edge of AI. 153 00:10:58.736 --> 00:11:07.723 And for a limited time, we're giving all of the viewers and listeners of this show 14 days for free to get all the education and equipment you need to stay at the edge. 154 00:11:08.003 --> 00:11:09.664 If that sounds like you, you should subscribe. 155 00:11:09.684 --> 00:11:10.585 The links in the description. 156 00:11:10.923 --> 00:11:14.706 I think also what happened is everyone had open claws. 157 00:11:15.767 --> 00:11:22.111 And for sure, like one of the coolest things was your claw became a little bit of an extension of you. 158 00:11:22.151 --> 00:11:28.136 Like it sort of, I remember one of the first people who, well Brandon was like a big driving force here. 159 00:11:28.176 --> 00:11:35.261 And then the second person who like really took it to a new level was Austin, who was at the time our head of growth. 160 00:11:35.441 --> 00:11:39.024 And he had, he had like a growth claw that you could ask all growth questions to. 161 00:11:39.044 --> 00:11:40.125 And that was like really interesting. 162 00:11:41.511 --> 00:11:55.639 But I think at least my memory of what happened is then we started, we were like, oh, we're going to build a platform for anyone to make their own, their own clause, even if you're non-technical, plus one, RIP plus one. 163 00:11:56.039 --> 00:11:58.380 But I don't think at that point it was really company agent. 164 00:11:58.500 --> 00:12:01.282 It was just like make a claw, like one click claw maker. 165 00:12:02.503 --> 00:12:06.625 And we tried that and we were having a lot of internal debates about 166 00:12:07.336 --> 00:12:11.099 Should it be one agent for everyone or an agent per person? 167 00:12:11.119 --> 00:12:12.500 You were very one agent for everyone. 168 00:12:12.540 --> 00:12:14.021 I was very like agent per person. 169 00:12:14.321 --> 00:12:24.850 I think you had a lot of hesitancies about the idea of building up our own platform infrastructure to support this like claw as a service thing. 170 00:12:24.870 --> 00:12:25.670 And we were like, fuck it. 171 00:12:25.690 --> 00:12:26.831 Like, let's do it anyway. 172 00:12:26.951 --> 00:12:28.272 And we were like, okay. 173 00:12:28.392 --> 00:12:28.733 Okay. 174 00:12:30.714 --> 00:12:34.737 which I appreciate, but also like in retrospect was like not the best idea. 175 00:12:35.498 --> 00:12:45.807 And like we so we launched that in beta and I think basically found that it was there was a lot of demand, especially at that time, there was a lot of demand for it, but 176 00:12:47.407 --> 00:12:50.328 OpenClaw itself, it was very hard to build a platform on. 177 00:12:50.508 --> 00:12:53.449 It was moving so fast, and it was completely vibe-coded. 178 00:12:54.290 --> 00:13:03.153 It's just hard to make the entire technical stack and then have an open-source product that's not really designed to be a platform as the thing you're serving. 179 00:13:04.554 --> 00:13:07.115 That was messy, and... 180 00:13:08.589 --> 00:13:10.950 Through that process, I think all of our claws basically died. 181 00:13:11.210 --> 00:13:12.150 Yeah. 182 00:13:12.470 --> 00:13:13.710 Because we just stopped. 183 00:13:14.651 --> 00:13:16.871 I remember my claw is named R2C2. 184 00:13:17.851 --> 00:13:23.433 And I felt like, oh yeah, I have this really attached relationship to R2C2. 185 00:13:23.533 --> 00:13:28.514 And then at some point, someone knocked the Mac Mini power off. 186 00:13:29.194 --> 00:13:30.334 And then it disappeared. 187 00:13:31.135 --> 00:13:31.955 But I didn't restart it. 188 00:13:32.015 --> 00:13:32.475 Yeah. 189 00:13:32.515 --> 00:13:33.875 Because at that point, I was using codecs. 190 00:13:33.895 --> 00:13:34.936 And I was like, this is fine. 191 00:13:34.956 --> 00:13:35.676 I don't need that. 192 00:13:37.117 --> 00:13:40.700 And that that was like a really interesting thing for me to realize like oh. 193 00:13:42.002 --> 00:13:44.504 OK there's all the people in the org who don't really. 194 00:13:44.524 --> 00:13:53.974 We don't really want to maintain their own cause and then the people who do want to do it like eventually you kind of get tired of it if there's a better option. 195 00:13:54.875 --> 00:14:05.819 And what we ended up doing is we, I think we were always starting to use like Cloud or Codex and then there are a couple of these company agents that we tried that like really sort of changed how we thought about this. 196 00:14:06.059 --> 00:14:09.760 So Victor being one, which was just really, really well designed. 197 00:14:09.780 --> 00:14:17.463 And I think at that point we started to realize, oh, like having a single company agent makes a huge difference. 198 00:14:18.350 --> 00:14:18.890 Yeah, there's a lot. 199 00:14:19.071 --> 00:14:22.253 I mean, it even just boils down to sort of like naming, right? 200 00:14:22.313 --> 00:14:29.399 Remembering which agent had which expertise was annoying and hard. 201 00:14:29.419 --> 00:14:33.743 I remember like Austin's agent, Montaigne, right? 202 00:14:33.843 --> 00:14:37.967 It sort of became the gravitational center because... 203 00:14:39.008 --> 00:14:49.014 He built it up to be so good at the growth job that it was like, well, I could add, I'm already using it for myself, minor growth jobs. 204 00:14:49.294 --> 00:14:53.577 I could invest in my agent or I could just add on to Montana and everyone else. 205 00:14:55.898 --> 00:14:56.839 Sort of gains from that. 206 00:14:57.459 --> 00:14:59.480 And I think there were a lot of pressures there. 207 00:14:59.620 --> 00:15:01.682 But of course, counteracting this again was like, 208 00:15:02.322 --> 00:15:03.722 I think we really liked our personal agents. 209 00:15:03.742 --> 00:15:05.723 You know, we're like, we're trying to keep it there. 210 00:15:05.803 --> 00:15:16.566 But, I mean, the other thing I think is underappreciated is that when you're building sort of like AI products now, the surface area for like products and agents is really large. 211 00:15:16.906 --> 00:15:18.987 You know, before it was like, oh, I've got a SaaS app. 212 00:15:19.027 --> 00:15:25.689 I can sort of like map out all the user flows and you go and build it and you sort of know how users will interact with your product. 213 00:15:26.629 --> 00:15:32.834 In the AI world, it's just... And with Plus One, what we noticed was, oh, we didn't even realize that you wanted to go do this thing. 214 00:15:32.854 --> 00:15:33.955 And how do you support it? 215 00:15:33.995 --> 00:15:34.595 So there was always... 216 00:15:34.615 --> 00:15:37.117 I talked a lot about complexity. 217 00:15:38.258 --> 00:15:45.163 Building and managing complexity for builders versus operators is a hard job. 218 00:15:45.223 --> 00:15:47.245 Builders want to be exposed to the complexity. 219 00:15:47.545 --> 00:15:52.308 Because that's everything you want to do is going in there and tinkering and turning knobs and whatnot. 220 00:15:52.729 --> 00:15:53.910 But if you're trying to run a business... 221 00:15:55.916 --> 00:15:57.577 You're not really... You don't want to do that. 222 00:15:57.677 --> 00:15:59.717 Mostly, you just want something that just works for you. 223 00:15:59.997 --> 00:16:05.379 And I think over time, we realized was, for us, we wanted something that just worked. 224 00:16:05.479 --> 00:16:09.340 We didn't want to deal with all of the little knobs and bells and whistles. 225 00:16:10.100 --> 00:16:12.321 And for the jobs that we... 226 00:16:12.961 --> 00:16:17.725 were using an agent for, like a lot of project management operations and whatnot. 227 00:16:18.045 --> 00:16:19.806 It was like better to have something solid. 228 00:16:19.966 --> 00:16:23.509 Yeah, and to have it be sort of shared. 229 00:16:23.649 --> 00:16:29.253 So when you use it, you're pretty confident that it's the right thing to be using for your company. 230 00:16:29.333 --> 00:16:29.473 Yeah. 231 00:16:29.614 --> 00:16:32.976 And someone has thought about it, you know, thought about the skill or the feature or whatever. 232 00:16:32.996 --> 00:16:33.917 Yeah. 233 00:16:34.599 --> 00:16:37.220 And so now we're in this place where we're launching the every agent. 234 00:16:37.240 --> 00:16:40.141 It's something that we use every day, like many, many times a day. 235 00:16:40.161 --> 00:16:41.401 The company would kind of like fall over. 236 00:16:41.421 --> 00:16:41.861 We didn't have it. 237 00:16:41.881 --> 00:16:43.562 And we have a bunch of beta customers. 238 00:16:44.622 --> 00:16:50.704 And we're also in this moment where, for example, I'm using my dot all the time from OpenAI. 239 00:16:50.824 --> 00:16:54.205 And there's Muse and Instinct on the personal agent side. 240 00:16:54.526 --> 00:16:58.327 One of the things, I'm curious for your bet here, one of the things that I 241 00:16:59.200 --> 00:17:07.444 feel is I think I will be using something like my dot in a year in Slack. 242 00:17:07.724 --> 00:17:14.167 Specifically because I use it inside of like it's basically just a portal to my codex. 243 00:17:15.127 --> 00:17:16.107 But I probably won't. 244 00:17:16.227 --> 00:17:18.809 Mine is named Boo and like it probably won't be named Boo. 245 00:17:18.989 --> 00:17:21.290 Like Boo is going to be dead in a year. 246 00:17:21.310 --> 00:17:23.651 Like lifespan six months maybe you know because I. 247 00:17:24.511 --> 00:17:32.617 The personal thing or the personality thing is cool, but particularly in a work context, I mostly just care like it doesn't do the thing. 248 00:17:33.818 --> 00:17:41.544 And the minute there's like, I'm like, oh, I have this personal connection, but the minute there's something that performs better, even if it doesn't have it, I'll just switch. 249 00:17:41.944 --> 00:17:45.787 I think there's much more room for that on the truly personal side. 250 00:17:47.848 --> 00:18:02.832 So I think that's my bet with dots is like you still actually need a company agent like the every agent and then that needs to play well with like everyone having their own like dot or dot like thing that is basically a portal to their command center. 251 00:18:03.132 --> 00:18:03.813 What do you think? 252 00:18:04.153 --> 00:18:05.033 Yeah, I think... 253 00:18:07.111 --> 00:18:12.514 I think we can see the line between professional and personal sort of shaping up very solidly. 254 00:18:13.294 --> 00:18:32.463 I think on the professional side, I do think it's sort of like a one agent per company or org kind of structure because there are just so many advantages again to it learning the workflows and what you do and everyone contributing just by working with it and making it better that it shapes itself to your company. 255 00:18:33.244 --> 00:18:36.225 And it's like the advantage of doing 256 00:18:36.345 --> 00:18:41.147 bringing a personal agent in, it's just gonna be, it doesn't have that context here, it's gonna be hard to teach. 257 00:18:41.907 --> 00:18:45.469 Like, it needs to have been there a year to really get up to speed. 258 00:18:45.489 --> 00:18:58.434 And so, I think we're gonna have that for the workplace, and I really don't see like, us, or the industry, repeating the chaos of like, every worker gets their coworker, and we're all gonna like double our company size, you know? 259 00:18:59.654 --> 00:19:07.696 But on the personal side, I actually think the, I was talking about those, the forces, the opposing force, the one of personalization, I think is actually huge. 260 00:19:08.297 --> 00:19:13.218 I feel like the thing that you love is that you have an agent that is just yours. 261 00:19:13.898 --> 00:19:18.619 And before the downsides there were like, it was open cloud, how do I do security? 262 00:19:19.059 --> 00:19:20.240 How do I manage it? 263 00:19:20.280 --> 00:19:21.120 How do I run it? 264 00:19:21.140 --> 00:19:24.881 There are all these questions that prevented it from gaining larger adoption. 265 00:19:25.321 --> 00:19:35.064 And now that you have bigger players coming in and more money, it's just like, oh, we're starting to get closer to like a better consumer personal agent experience. 266 00:19:35.244 --> 00:19:37.525 And it's going to be like incredibly attractive. 267 00:19:37.565 --> 00:19:38.585 Yeah. 268 00:19:38.805 --> 00:19:40.126 And so I think we're going to see a split. 269 00:19:40.146 --> 00:19:46.788 I think we're going to see like I go to work and I work with my company agent at the company because we all are working together towards this collective goal. 270 00:19:47.733 --> 00:19:55.024 And at home, it might just be me or me and my family that is using a personal agent to help us do very similar things, but it's like personalized to us. 271 00:19:55.906 --> 00:19:59.451 One thing that I'm really interested... We've been talking about this a lot. 272 00:19:59.491 --> 00:20:00.553 I think you have a very... 273 00:20:01.502 --> 00:20:07.968 good perspective on and I just want to hear more about it is you've had a long career in tech. 274 00:20:08.989 --> 00:20:14.895 Doing engineering management at like great startups, bigger companies, you've founded your own companies. 275 00:20:15.075 --> 00:20:17.057 You led the engineering for this effort. 276 00:20:18.804 --> 00:20:20.806 with a team of, I don't know, five to seven-ish people. 277 00:20:21.187 --> 00:20:26.852 How has engineering management changed, especially over the last couple months? 278 00:20:26.872 --> 00:20:31.677 And what have you learned in the process of bringing this to market that you didn't necessarily know? 279 00:20:32.238 --> 00:20:33.519 Ironically, engineering management... 280 00:20:35.971 --> 00:20:47.124 feels the most similar out of any of the jobs that I've done before in the present day, especially like now getting to come back to it and see what it's like in the age of AI. 281 00:20:48.346 --> 00:20:50.429 I think there are lots of advantages, like 282 00:20:52.094 --> 00:20:58.421 For an engineering manager, a lot of your leverage comes from the ability to see how much of the field you can see. 283 00:20:58.742 --> 00:21:02.586 And so in previous years, you would spend a lot of time trying to gather that information. 284 00:21:02.606 --> 00:21:05.289 What would be an example of something that's on the field that you need to see? 285 00:21:05.589 --> 00:21:07.632 Oh, you know, there are a lot of slack... 286 00:21:08.192 --> 00:21:17.638 channels, you know, like keeping up with Slack channels and emails and, you know, customer reports and how are you doing the integration of all of this into a single place? 287 00:21:17.658 --> 00:21:25.423 And like, I run a 10 feed, right, that does this where it consolidates that all into a single feed. 288 00:21:25.804 --> 00:21:33.929 And during the day, I can just kind of go through and it really keeps me, gives me a much better like vision into everything going on related to the project. 289 00:21:33.949 --> 00:21:36.250 It's like, okay, is marketing happening over here? 290 00:21:36.310 --> 00:21:36.571 Like... 291 00:21:37.311 --> 00:21:39.313 are we seeing custom reports over there? 292 00:21:39.674 --> 00:21:41.956 Which normally would have been meetings. 293 00:21:42.797 --> 00:21:43.338 That makes sense. 294 00:21:43.378 --> 00:21:43.798 Yeah. 295 00:21:43.958 --> 00:21:53.809 So for people listening who don't know, Tend is this open source app I made that you can create basically custom feeds for yourself where you hook up data sources. 296 00:21:53.829 --> 00:21:57.473 So in this case it's meetings or notion notes or anything like that. 297 00:21:57.573 --> 00:21:57.834 And then 298 00:21:58.374 --> 00:22:05.799 Your AI can just automatically turn those data sources into a feed that either has decisions you need to make or is like FYI, this thing happened in a meeting, here's what you want. 299 00:22:05.819 --> 00:22:09.241 And it sort of learns over time what you want so you can make a feed for yourself. 300 00:22:09.622 --> 00:22:11.583 So you have one for your engineering stuff. 301 00:22:11.623 --> 00:22:13.264 I have one for just like the full company. 302 00:22:13.524 --> 00:22:18.788 What are a couple of things that you've learned maybe even in the last week from like reading that you might not have known otherwise? 303 00:22:19.473 --> 00:22:23.314 Yeah, a lot on the marketing side of launching a product. 304 00:22:23.334 --> 00:22:32.777 So normally when you're building a product, you're sort of in it with your teammates and you're looking at screens and designs and user error bug reports and whatnot. 305 00:22:33.517 --> 00:22:35.518 And you kind of have a feel for everything going on. 306 00:22:36.487 --> 00:22:48.216 But launching a product, there's a lot of things and conversations and meetings that happen that are public, but they're just out in places where you don't necessarily have the time to check or the attention. 307 00:22:48.416 --> 00:22:58.164 And so it's like, how do you go and grab those things and just make sure that one, they're going the way you want them to, and two, that you know who to give guidance to. 308 00:22:58.184 --> 00:23:03.107 Sometimes the rumor mill of a company comes back, it's like, oh, did you hear we're doing this sort of thing? 309 00:23:03.127 --> 00:23:03.468 And it's like, 310 00:23:04.209 --> 00:23:10.973 well, no, we shouldn't be doing that, but then who do I talk to is like a huge question. 311 00:23:11.053 --> 00:23:13.274 And as a manager, that's your leverage, right? 312 00:23:13.354 --> 00:23:17.477 It's like I need to be able to go and talk to and set guidance and set vision for all this. 313 00:23:17.677 --> 00:23:25.782 And so having sort of this higher, like field marshal level view of a lot of things is just so helpful, even for a small team. 314 00:23:26.608 --> 00:23:37.895 Yeah, I think you've been the beneficiary or I don't know if beneficiary, maybe it's the opposite beneficiary on the receiving end of a few times where I'm like, hey, I like read this thing in 10 like that. 315 00:23:37.995 --> 00:23:42.398 This happened in a meeting, like what's going on here or like just just to put my thumb on it or whatever. 316 00:23:42.418 --> 00:23:49.062 It's I mean, as a manager, it's like it's like the coolest thing and it can be like weird for employees. 317 00:23:49.142 --> 00:23:50.663 But like, yeah, I'm. 318 00:23:51.323 --> 00:23:53.104 Very curious to see how that changes. 319 00:23:53.584 --> 00:24:01.208 Now would be a time as a manager, an engineering manager, where you could actually be ripping tokens while you're managing a fairly large team. 320 00:24:01.668 --> 00:24:03.029 What is your take on that? 321 00:24:03.089 --> 00:24:05.870 I think a lot of it comes down to people. 322 00:24:05.910 --> 00:24:10.873 I think the job of the manager is still a force multiplier on folks. 323 00:24:10.913 --> 00:24:15.435 And so it's like, where do I spend my epsilon extra piece of energy? 324 00:24:15.455 --> 00:24:16.295 And... 325 00:24:18.357 --> 00:24:22.520 it's really probably not on tokens for coding. 326 00:24:22.860 --> 00:24:23.100 Right? 327 00:24:23.160 --> 00:24:32.247 And this is the part that's actually surprisingly very much the same as going, it's like a people, still very much a people job and you want it to be a people job. 328 00:24:32.667 --> 00:24:33.267 Now the difference is 329 00:24:34.268 --> 00:24:36.829 your leverage has increased because their leverage has increased. 330 00:24:37.170 --> 00:24:42.032 And so the most valuable things for me recently have just been like, what are the bottlenecks? 331 00:24:42.512 --> 00:24:42.932 Right? 332 00:24:42.992 --> 00:24:44.313 And how do we go and remove the bottlenecks? 333 00:24:44.653 --> 00:24:49.536 And so it's funny as I sort of transition from like doing more technical work back into engineering management. 334 00:24:49.976 --> 00:24:52.497 Like you see my token usage just fall off a cliff, right? 335 00:24:52.517 --> 00:24:56.359 Because I'm using it and it's sort of gathering all this information. 336 00:24:56.379 --> 00:24:57.460 But when it comes to action, 337 00:24:58.300 --> 00:25:04.304 Like, there's no billion dollar, sorry, billion token move I can make. 338 00:25:04.864 --> 00:25:08.606 It's mostly, like, we still need to sit down and, like, have a conversation, you know. 339 00:25:08.907 --> 00:25:14.930 Now, everyone, it can, like, there's a whole bunch of routing functions in there that are now, you know, help with AI. 340 00:25:14.950 --> 00:25:16.451 Like, we can do summaries. 341 00:25:16.511 --> 00:25:19.773 Like, how do we summarize and pass information? 342 00:25:19.793 --> 00:25:24.977 A lot of jobs that managers used to do and a lot of the things that you used to be responsible for in terms of, like, 343 00:25:25.717 --> 00:25:33.524 ops and the health of the team in the company are much less manual now. 344 00:25:33.664 --> 00:25:34.625 Have you automated that? 345 00:25:35.105 --> 00:25:38.868 I mean, mostly for the agent team, it's just like, we're good. 346 00:25:39.008 --> 00:25:43.252 It was just amazing to me having built a lot of dashboards and a lot of learning systems. 347 00:25:43.312 --> 00:25:49.677 It was like, oh no, because it's really definable and really testable, agents are great at it. 348 00:25:50.127 --> 00:25:50.267 Hmm. 349 00:25:50.927 --> 00:25:57.650 Like what's an example of, because I don't, to be honest, I don't look at a lot of like agent team health metrics that you're looking at. 350 00:25:57.670 --> 00:25:59.210 Like what's an example of some of those dashboards? 351 00:25:59.230 --> 00:26:03.832 Well, some of them are what you might call like performance dashboards, right? 352 00:26:03.852 --> 00:26:07.773 It's like how many errors are our API throwing, right? 353 00:26:07.793 --> 00:26:12.175 It's like a, okay, like there's a dashboard for that. 354 00:26:12.495 --> 00:26:15.356 Same thing on the product side, it's like how many users are using them. 355 00:26:15.656 --> 00:26:16.877 But the real magic now is, 356 00:26:17.837 --> 00:26:22.803 Before, it was like you had to set... Like, alerting systems were... 357 00:26:22.923 --> 00:26:27.649 Even just a few years ago, were still pretty, like, neolithic, right? 358 00:26:27.669 --> 00:26:29.191 It was sort of like, I'm going to set an alert. 359 00:26:29.211 --> 00:26:30.272 I'm going to set some bars. 360 00:26:30.352 --> 00:26:32.074 You know, it's usually above this or below this. 361 00:26:32.595 --> 00:26:36.719 And one place that just the magic is there is just, like, you can put an agent. 362 00:26:36.799 --> 00:26:37.580 It's like, look at this. 363 00:26:38.700 --> 00:26:41.502 every five minutes and see if there's an anomaly. 364 00:26:41.542 --> 00:26:44.244 If there's an anomaly, do intelligent routing. 365 00:26:44.604 --> 00:26:45.725 They'll alert me. 366 00:26:45.785 --> 00:26:57.833 You know, there's all these things like setting up on-call structures where you're like, okay, this person needs to be on-call and this person needs to be on-call and now you have an agent on-call as the first person all the time. 367 00:26:57.853 --> 00:26:59.654 You don't have to worry about vacation. 368 00:26:59.674 --> 00:27:00.955 You're not scheduling around. 369 00:27:01.696 --> 00:27:05.438 All these sort of below the line pieces of engineering management have gotten much better. 370 00:27:06.459 --> 00:27:07.200 That's interesting. 371 00:27:07.340 --> 00:27:08.120 I want to throw it back at you. 372 00:27:08.140 --> 00:27:08.621 I'm curious. 373 00:27:08.801 --> 00:27:10.522 We were sort of on opposite sides before. 374 00:27:11.103 --> 00:27:14.826 How do you think this personal agent dynamic is going to play out? 375 00:27:15.106 --> 00:27:19.089 I mean, my take on technology is you're never wrong. 376 00:27:19.129 --> 00:27:20.410 You're just too early usually. 377 00:27:20.750 --> 00:27:23.252 So I'm still holding out that I might be right. 378 00:27:24.893 --> 00:27:40.414 No, I think my current feeling is a big, and I think you probably disagree with this, but a big reason why we went from everyone has their own to a company is they were just too hard to maintain. 379 00:27:43.144 --> 00:27:47.406 And that we will start to go from, there's a company one which will continue to exist. 380 00:27:47.786 --> 00:27:51.108 And then that will start to subdivide into like teams which you already kind of have. 381 00:27:51.168 --> 00:27:59.633 Like we sort of also have a couple of these agents like Claudia or Bill or there's a couple other ones that are floating around that different teams maintain that are just like for that team. 382 00:28:00.959 --> 00:28:10.802 And then that will probably again bifurcate a little bit and and maybe that's maybe that's just an a dot an extension of your personal code. 383 00:28:10.822 --> 00:28:11.642 But maybe it's something else. 384 00:28:12.383 --> 00:28:19.145 So I do think we will probably get to the shadow org chart at some point. 385 00:28:20.366 --> 00:28:29.872 I thought it might be a more bottoms up like everyone has their own and then we kind of like coalesce into but now it seems like actually it's More likely to be you make something really great. 386 00:28:29.892 --> 00:28:42.982 That's at the top and then you kind of split that out every once in a while when you when you need The specialization yeah, I mean this sort of details to a little bit of what we've seen of like folks who are trying to do like a transformation right is like 387 00:28:43.990 --> 00:28:49.456 Usually if you're a line engineer, you're not the one trying to do AI transformation in your company. 388 00:28:49.476 --> 00:28:53.740 Usually it's the CEO or someone high up who's like, we really need to accelerate. 389 00:28:54.861 --> 00:28:56.042 We're struggling to do it. 390 00:28:56.082 --> 00:29:01.067 Let's drop an agent in here that everyone can use and see other people using. 391 00:29:02.489 --> 00:29:07.314 I'm also really intrigued by... 392 00:29:08.712 --> 00:29:13.874 like, if people are going to have, like, separate agents for each role. 393 00:29:13.894 --> 00:29:21.637 So, like, one of the things is, funny enough, in my, on the personal side, like, in GrokBot, I use GrokBot, and then it's, like, I will run, like, I have, like, a lot of coaches, right? 394 00:29:21.677 --> 00:29:23.898 And for whatever reason, that mental model is, like, 395 00:29:24.778 --> 00:29:25.719 works really well for me. 396 00:29:26.460 --> 00:29:27.742 Maybe because it maps to humans. 397 00:29:27.762 --> 00:29:30.705 So it's like, you have a coach for this, a coach for this, a coach for this. 398 00:29:30.725 --> 00:29:32.587 What are your different coaches? 399 00:29:33.749 --> 00:29:37.513 I'm like, I'm a terrible swimmer, which is really funny because I like to surf. 400 00:29:39.095 --> 00:29:40.917 I'm like a really bad lap swimmer for a reason. 401 00:29:40.937 --> 00:29:43.540 I was like, okay, I need someone who's going to keep me on like, 402 00:29:44.120 --> 00:29:47.681 like, a lap-swimming cadence, you know? 403 00:29:47.781 --> 00:29:52.422 But, like, why not have that just be, like, part of a bigger bot that, like, does a bunch of different things? 404 00:29:52.462 --> 00:30:02.183 Yeah, I mean, this is where it's, like, an interesting sort of, like, why is it psychologically that I'm, like, I need this sort of, like, this separation for this one role, you know? 405 00:30:02.223 --> 00:30:03.564 It's like, do I have that? 406 00:30:03.624 --> 00:30:05.024 And then I have, like, a therapist. 407 00:30:05.064 --> 00:30:07.384 It's like, do they talk together, you know? 408 00:30:07.444 --> 00:30:08.124 Yeah, yeah, yeah. 409 00:30:08.204 --> 00:30:09.945 And it's funny because I can kind of line them up on... 410 00:30:11.945 --> 00:30:18.286 jobs to be done, but like, not in the way that I think that like, if I rationally was trying to explain it, what it would look like. 411 00:30:18.306 --> 00:30:20.227 You know, I have to deal with a storage unit. 412 00:30:20.287 --> 00:30:22.407 I don't have like a storage unit bot that's dealing with it. 413 00:30:22.707 --> 00:30:26.808 But like, I do for, you know, maybe it's like, is this a throwaway person? 414 00:30:27.088 --> 00:30:29.288 Do I expect to never like interact with them again? 415 00:30:29.708 --> 00:30:32.309 If it's transactional, it's like a thread in Codex. 416 00:30:32.829 --> 00:30:39.010 If it's like, I'm building a relationship, you know, one day I will FaceTime my Grokbot into my workouts, you know? 417 00:30:40.185 --> 00:30:41.006 That is interesting. 418 00:30:41.046 --> 00:30:46.231 Yeah, I think that there's something about it's almost saying to yourself, planting a flag. 419 00:30:46.251 --> 00:30:51.857 This is an area I care about and will care about for a long time that maybe you want a little bot for. 420 00:30:52.377 --> 00:30:56.281 Right now, my closest thing to that is like pinned tabs and codecs. 421 00:30:56.642 --> 00:30:59.424 But now those are totally outmoded because I just use dot. 422 00:30:59.464 --> 00:31:00.045 There's no like... 423 00:31:01.546 --> 00:31:02.948 You can't have multiple dots. 424 00:31:03.488 --> 00:31:05.850 And I also wonder how much context comes with it. 425 00:31:05.970 --> 00:31:13.216 I think the really interesting thing that's happening now is we're ratcheting up the amount of proactivity we allow. 426 00:31:14.538 --> 00:31:18.461 And so Instinct and Muse and Grappar are doing these things where they're starting to reach out. 427 00:31:19.342 --> 00:31:20.523 And I think when you have... 428 00:31:21.721 --> 00:31:29.945 AIs are reaching out, if it's all in Codex or Claude, and it's coming at you, and it's like, you have to remember what thread it was. 429 00:31:30.286 --> 00:31:33.487 But if it's like, Willie's swimming coach is reaching out. 430 00:31:33.587 --> 00:31:36.189 I'm like, okay, this is the swimming coach. 431 00:31:36.289 --> 00:31:42.492 I can instantly place you in the conversation of like, okay, I didn't swim enough, you know, yeah. 432 00:31:44.593 --> 00:32:12.027 Yeah, the productivity thing is a big deal and that's something that we are just baking into the agent that I'm super excited about because this is very related to tend to like there's this whole thing that's starting to happen which is it used to be that in order to make a feed of anything you had to be a gigantic corporation with like hundreds of millions of users and a lot of money and now everyone can have their own feed and companies like ours can you know for example in the every agent it has. 433 00:32:12.867 --> 00:32:18.448 frontier alerts that are just like, okay, this thing happened, here's what it means for you, which I'm super excited about. 434 00:32:18.468 --> 00:32:33.871 I think there's so many places to go in that sort of proactively telling you what you need to know at the right time that it used to be that AI was not able to do that in a way that was good, but now with Jeb and some of these other things like Decisions API, it's happening. 435 00:32:34.111 --> 00:32:34.291 Yeah. 436 00:32:34.591 --> 00:32:41.653 I mean, we very much have a feed-based world across everything, and it's a tension because... 437 00:32:42.613 --> 00:32:45.855 I spend a lot of time on X, like probably everyone, I assume. 438 00:32:46.336 --> 00:32:49.117 Everyone in the world spends time on X. 439 00:32:49.638 --> 00:33:01.125 And for me, it's a funny sort of explore-exploit problem, which is like I'm on there mostly to see the new ideas emerging and hear how people debate them. 440 00:33:01.705 --> 00:33:02.745 There's also a lot of trash. 441 00:33:03.125 --> 00:33:08.327 And my cognitive defenses say that they have to be up. 442 00:33:08.367 --> 00:33:15.749 I have to have my protections up when I'm reading it just because it's like, I don't want to be dragged into this flame war or this whatever. 443 00:33:16.569 --> 00:33:20.450 I'm trying to sift out the nuggets from this stream of drops. 444 00:33:20.890 --> 00:33:23.151 It would be great if I didn't have to be that thing. 445 00:33:23.551 --> 00:33:23.751 Right? 446 00:33:23.831 --> 00:33:27.332 If I had an agent that could take the brunt of that and then I get out the nuggets. 447 00:33:27.692 --> 00:33:34.975 That's actually something that's been helpful for me with my dot is like it does, especially, I mean, this is not a feed, but like for Slack. 448 00:33:35.155 --> 00:33:37.475 Like just telling me, here's a couple things that you probably missed. 449 00:33:37.736 --> 00:33:38.776 I think Slack's a feed too. 450 00:33:38.796 --> 00:33:39.696 So helpful. 451 00:33:39.796 --> 00:33:40.816 It's so helpful. 452 00:33:41.497 --> 00:33:41.677 Yeah. 453 00:33:42.417 --> 00:33:43.777 Willie, always a pleasure. 454 00:33:43.957 --> 00:33:44.538 Always a pleasure. 455 00:33:44.578 --> 00:33:45.138 Thanks for joining. 456 00:33:45.318 --> 00:33:45.538 Thanks.