WEBVTT NOTE This file was generated by Descript 00:00:02.996 --> 00:00:04.886 CJ: What happens in the desert stays in the desert. 00:00:05.160 --> 00:00:06.660 How, how was it this time around? 00:00:07.230 --> 00:00:07.930 Colin: It was good. 00:00:07.950 --> 00:00:11.770 It usually, yes, what, what happens at Burning Man stays at Burning Man, but 00:00:11.800 --> 00:00:16.080 it was not, I think what I did this year was not be what most people think 00:00:16.080 --> 00:00:20.340 of as, as Burning Man, but I think I mostly just needed to go camping. 00:00:20.500 --> 00:00:24.960 It didn't matter where, and so surprisingly got some 00:00:24.960 --> 00:00:26.360 of the best sleep I've had. 00:00:26.590 --> 00:00:27.820 Wasn't out all night. 00:00:27.900 --> 00:00:33.150 Woke up early, went to yoga and coffee and bloody Marys. 00:00:33.150 --> 00:00:36.680 And just like that, that was like rinse and repeat every day, 00:00:36.830 --> 00:00:38.700 which you could just do at home. 00:00:38.910 --> 00:00:42.370 But you know, being able to just wake up, not have to go to work. 00:00:42.380 --> 00:00:47.110 Cause it took, took some time off ride your bike, you know, people joke about, 00:00:47.785 --> 00:00:53.265 Like Burning Man being this like utopia city thing, but it is very unsustainable 00:00:53.335 --> 00:00:57.545 as, as that, but for a week being able to not just think about going to 00:00:57.545 --> 00:01:01.255 work, get on a bike, no cars, right. 00:01:01.295 --> 00:01:07.380 And go ride to the nearest little, this really cool live music yoga thing. 00:01:08.810 --> 00:01:11.430 And then just kind of see where the day takes you. 00:01:11.440 --> 00:01:13.560 Like you don't really get to do that very often. 00:01:13.630 --> 00:01:15.800 So it was a good little time. 00:01:15.870 --> 00:01:18.880 Not probably what most people think of as Burning Man, but it 00:01:18.880 --> 00:01:20.560 was, it was just what I needed. 00:01:20.996 --> 00:01:25.296 CJ: Great way to get away from screens and just like, let your mind reset 00:01:25.296 --> 00:01:29.846 to just like complete refresh reboot. 00:01:30.526 --> 00:01:33.916 So are you coming back feeling like really energized and excited? 00:01:33.916 --> 00:01:35.566 Or was it long enough? 00:01:35.566 --> 00:01:38.026 Do you want it to be like three weeks, four 00:01:38.091 --> 00:01:42.821 Colin: Yeah, I'm excited for my next vacation to be somewhere a 00:01:42.821 --> 00:01:44.991 little bit less survival focused. 00:01:45.051 --> 00:01:48.561 Because you don't really come back super refreshed when you're dealing 00:01:48.561 --> 00:01:49.941 with the desert and all of that. 00:01:50.061 --> 00:01:54.861 But yeah, no, it was good to be away from screens, to not have to have 00:01:54.861 --> 00:01:59.771 deadlines and Those kinds of things was feeling super good and strong. 00:01:59.771 --> 00:02:04.961 And then I like bruised a rib or maybe cracked a rib on the last 00:02:04.961 --> 00:02:08.391 day doing the lead again, the least exciting things possible. 00:02:08.411 --> 00:02:13.681 But yeah, so a little, little, little ouchy right now, but I 00:02:13.681 --> 00:02:17.261 think, and I think it's fine and it'll, I'll be back at it soon. 00:02:17.271 --> 00:02:19.361 CJ: Sounds like a solid year. 00:02:19.971 --> 00:02:21.911 Things are, yeah, things are busy here. 00:02:21.911 --> 00:02:24.301 We we just kicked off the school year. 00:02:24.301 --> 00:02:26.031 So the boys are back in school. 00:02:26.501 --> 00:02:30.481 They, we just had open house where we got to meet all of the new teachers. 00:02:30.991 --> 00:02:34.781 And go through kind of like all the processes and what's expected of them 00:02:34.781 --> 00:02:39.561 for this new year and learning about the new grading systems and whatever, 00:02:39.561 --> 00:02:40.711 but the kids are super pumped. 00:02:40.721 --> 00:02:44.091 We asked them at the end of the summer, like, you know, are you looking 00:02:44.111 --> 00:02:45.021 forward to going back to school? 00:02:45.021 --> 00:02:45.761 Are you dreading it? 00:02:45.871 --> 00:02:49.451 And they said of their friends, they're the most excited. 00:02:49.501 --> 00:02:53.636 They're just like so pumped to be back hanging out with their homies. 00:02:53.876 --> 00:02:54.866 Their words, not mine. 00:02:55.106 --> 00:02:57.446 And yeah, I think it's, it's, it's great. 00:02:57.476 --> 00:02:59.246 We love that being the reaction. 00:02:59.246 --> 00:02:59.606 So, 00:03:00.351 --> 00:03:00.611 Colin: Yeah. 00:03:00.766 --> 00:03:00.946 CJ: It's 00:03:00.976 --> 00:03:01.436 Colin: I think. 00:03:01.976 --> 00:03:05.656 You have a pretty good household of general learning and 00:03:06.896 --> 00:03:12.106 curiosity, so I think, yeah, the kids are going to be alright. 00:03:13.306 --> 00:03:13.816 CJ: Thank you. 00:03:13.836 --> 00:03:18.426 Yeah, we were, we're, we're trying hard to set them up for success and, you know, 00:03:18.436 --> 00:03:20.176 get them get them into position where. 00:03:21.131 --> 00:03:24.211 It would be very, very challenging for them to fumble the ball, 00:03:24.251 --> 00:03:25.351 but we'll see how it goes. 00:03:26.296 --> 00:03:30.656 Colin: Yeah, I mean, you model that, right, in terms of like, you're always 00:03:30.656 --> 00:03:36.246 trying new things, and From computer stuff to let's see if we can make 00:03:36.266 --> 00:03:38.436 maple syrup to, you know, whatever. 00:03:38.436 --> 00:03:38.706 Right. 00:03:38.706 --> 00:03:41.906 So I think that that's that natural curiosity. 00:03:41.936 --> 00:03:43.366 It rubs off on people. 00:03:43.376 --> 00:03:43.736 It's good. 00:03:44.491 --> 00:03:45.441 CJ: Yeah, that's true. 00:03:45.631 --> 00:03:45.891 Yeah. 00:03:45.901 --> 00:03:46.841 I never thought about it that way. 00:03:47.351 --> 00:03:51.691 Speaking of natural curiosity, have you seen, I know we actually, we talked 00:03:51.701 --> 00:03:53.701 about this a little bit, the cursor. 00:03:53.701 --> 00:03:54.871 ai editor. 00:03:56.456 --> 00:03:58.816 Are you, is that like your daily driver or what? 00:03:58.836 --> 00:03:59.056 Yeah. 00:03:59.056 --> 00:04:02.826 Like kind of, what are you using most, most of the time now to write code? 00:04:03.866 --> 00:04:06.406 Colin: So it is not my daily driver. 00:04:06.616 --> 00:04:09.836 I have not asked and I'm not sure if we're allowed to use it at work. 00:04:09.936 --> 00:04:14.016 So We already, we have copilot as part of GitHub enterprise. 00:04:14.016 --> 00:04:18.236 So I imagine the answer is probably not, we're not allowed to use it. 00:04:18.346 --> 00:04:22.526 And just because I think like there's a lot of protections built into GitHub 00:04:22.726 --> 00:04:26.596 enterprise and copilot with like enterprise code and things like that. 00:04:26.596 --> 00:04:31.926 So for cursor is on my personal machine and VS code and 00:04:31.926 --> 00:04:33.636 copilot is on my work machine. 00:04:34.871 --> 00:04:35.181 CJ: Got it. 00:04:35.221 --> 00:04:35.621 Okay. 00:04:37.031 --> 00:04:37.231 Yeah. 00:04:37.231 --> 00:04:42.081 It seems like, well, given that cursor is a fork of VS code, it feels 00:04:42.081 --> 00:04:43.591 like a pretty safe thing to add. 00:04:43.591 --> 00:04:45.491 It's kind of just like an extension on top of it. 00:04:45.491 --> 00:04:51.681 It seems But yeah, like that extension I'm sure has a certain limitations in 00:04:51.681 --> 00:04:56.471 terms of, or as a, as a giant company, you would probably want to like enforce 00:04:56.471 --> 00:05:00.491 certain restrictions around just adding extensions sort of willy nilly. 00:05:00.591 --> 00:05:05.591 Colin: Yeah, I think Copilot has some rules around like what can be trained 00:05:05.591 --> 00:05:07.001 and what can't and things like that. 00:05:07.001 --> 00:05:11.611 That's enforced at the enterprise level So the org controls a little bit more. 00:05:11.631 --> 00:05:16.581 I don't think cursor is quite there yet Then maybe they are and so you 00:05:16.581 --> 00:05:20.701 do have to have another Payment plan with cursor right you have to do a 00:05:20.741 --> 00:05:25.711 subscription with them and all of that So are you using it as your daily driver? 00:05:26.471 --> 00:05:29.971 CJ: so I am not yet, but I have. 00:05:30.121 --> 00:05:33.391 I, I started to get FOMO. 00:05:33.881 --> 00:05:37.751 I was like, I'm in Vim over here and I've got co pilot working fine. 00:05:38.301 --> 00:05:42.761 But when I, I started, I started to see like a couple of screencasts 00:05:42.761 --> 00:05:46.371 where people were building like entire features where it's like adding and 00:05:46.371 --> 00:05:49.761 or editing like multiple files in a repo instead of just kind of like 00:05:49.781 --> 00:05:51.161 auto completing as you're typing. 00:05:51.591 --> 00:05:54.331 And I was like, I want that, that, that seems. 00:05:55.416 --> 00:05:58.566 Like, it seems like another step change in productivity. 00:05:59.096 --> 00:06:06.716 And so I started like kind of re setting and refreshing and like going 00:06:06.716 --> 00:06:11.576 out on this exploratory research expedition to try to figure out what 00:06:11.596 --> 00:06:17.416 is going to be the daily driver for the end of 2024 and going into 2025. 00:06:18.736 --> 00:06:22.116 And so I downloaded installed cursor was playing around with it. 00:06:22.176 --> 00:06:25.406 Kraftwerk is not a giant corporate organization. 00:06:25.446 --> 00:06:29.226 And so we're much more comfortable kind of like just seeing what tools 00:06:29.266 --> 00:06:31.126 work and make us most effective. 00:06:31.126 --> 00:06:36.236 And so installed it index, the code base, and was kind of like messing 00:06:36.256 --> 00:06:38.826 around with some basic features, which I thought were pretty. 00:06:39.071 --> 00:06:44.201 Pretty decent, but like any of this like chat GPT or AI stuff, like some, 00:06:44.391 --> 00:06:48.261 some, some of the output was wrong or like just needed to be nudged like 00:06:48.261 --> 00:06:49.561 several times in the right direction. 00:06:49.561 --> 00:06:51.511 I'm like, is this actually making us faster? 00:06:52.121 --> 00:06:57.531 So I tried cursor for like a week and a half and then went and I found this. 00:06:58.306 --> 00:06:59.926 Plugin, it was on Hacker News, Avante. 00:06:59.926 --> 00:07:05.346 Neovim and we'll drop a, drop links to it in the resources, but I started 00:07:05.346 --> 00:07:09.686 exploring like, what would it look like to have the same kind of these 00:07:10.086 --> 00:07:12.056 or like similar features in Neovim. 00:07:12.456 --> 00:07:16.946 And so the things beyond like type to complete that I was excited about was 00:07:16.956 --> 00:07:20.246 like highlight a block of code and. 00:07:21.681 --> 00:07:23.001 Have that refactored for you. 00:07:23.111 --> 00:07:27.781 So just say like, change this block of code to do X. 00:07:28.451 --> 00:07:30.941 Another thing was like implement this whole feature. 00:07:31.371 --> 00:07:33.371 And then the last one was like chat with my code. 00:07:33.731 --> 00:07:39.541 And so I was finding that I was copying and pasting like many, like entire 00:07:39.541 --> 00:07:43.341 code files and pasting them into either chat, GPT or Claude, and then 00:07:43.391 --> 00:07:45.701 asking questions like, how would you. 00:07:45.896 --> 00:07:49.926 Like find the memory leak in this Ruby thing, or how would you 00:07:49.926 --> 00:07:54.036 optimize this query or, you know, things like that, that needed the 00:07:54.036 --> 00:07:55.586 context of several different files. 00:07:55.946 --> 00:08:02.791 And so one of the nice things about cursor was that you can at mention files. 00:08:03.061 --> 00:08:06.511 And it'll like include that in the context in order to like, give 00:08:06.511 --> 00:08:10.471 you either like chat with it or give you edit recommendations. 00:08:10.891 --> 00:08:14.641 So I wanted that inside of Avante and it turns out that Avante doesn't 00:08:14.651 --> 00:08:16.991 have the app mentioning thing yet. 00:08:17.631 --> 00:08:19.311 And also it was. 00:08:20.151 --> 00:08:23.481 In order to get it set up and working as expected, I had to like change 00:08:23.481 --> 00:08:26.381 the mind, Vim plugin manager and like a whole bunch of other stuff. 00:08:26.381 --> 00:08:30.411 So yeah, so I'm just, I'm still very much in like the exploratory phase for 00:08:30.411 --> 00:08:33.851 what the next daily driver is going to be including VS code, including cursor, 00:08:33.851 --> 00:08:35.001 including a whole bunch of stuff. 00:08:35.331 --> 00:08:36.671 So I don't know. 00:08:36.671 --> 00:08:39.211 Yeah. 00:08:39.221 --> 00:08:47.361 Colin: had mixed success with it knowing like recent files. 00:08:47.411 --> 00:08:49.831 This is not, this is more co pilot. 00:08:50.341 --> 00:08:52.011 Like sometimes it's like, oh, you want to do this? 00:08:52.021 --> 00:08:52.841 I'm like, that's awesome. 00:08:52.841 --> 00:08:56.196 But then like another minute later and it's like, Back to being dumb. 00:08:56.506 --> 00:08:59.146 And so I don't know if there's like literally like, Oh, it's keeping 00:08:59.146 --> 00:09:01.386 the file you're on in context. 00:09:01.416 --> 00:09:06.606 And then maybe the last file by default cursor felt like it had a bigger 00:09:06.606 --> 00:09:11.226 context window, but also was indexed on the actual code, like in copilot. 00:09:11.226 --> 00:09:15.586 I might try to use a symbol from another file and it just doesn't know about it. 00:09:16.081 --> 00:09:20.081 Which is also like most IDEs should just have that like it's not you don't 00:09:20.081 --> 00:09:24.661 need AI for that if it's especially if it's like a static language or something 00:09:24.681 --> 00:09:28.781 but I think we mentioned it but like what I have found is that Cursor and 00:09:28.781 --> 00:09:34.926 Rails has been fairly insane like for me just and this is Out of the box. 00:09:34.996 --> 00:09:40.236 I'm sure that there's going to be more and maybe cursors like just one step change 00:09:40.236 --> 00:09:44.296 towards this, like whatever the next thing is, but be anything that has lots of 00:09:44.296 --> 00:09:47.276 conventions, it seems to do really well. 00:09:47.496 --> 00:09:50.086 When you get into JavaScript land, I'm sure it does fine. 00:09:50.491 --> 00:09:54.481 But when you can put files anywhere and you can do things any way, it's a little 00:09:54.481 --> 00:09:58.061 bit less, there's less patterns to follow. 00:09:58.061 --> 00:10:01.811 And like, when you think of these things as predicting the next best thing, 00:10:02.781 --> 00:10:06.461 it makes a lot of sense that Rails conventions, you know, all of that. 00:10:06.471 --> 00:10:12.471 And who are we kidding, most apps are doing the same thing in different 00:10:12.471 --> 00:10:15.561 steps, different ways, different buckets of content and different, you know, 00:10:15.561 --> 00:10:17.241 buckets of views and things like that. 00:10:17.241 --> 00:10:20.901 So Yeah, it is funny when everyone's like, Oh, this is going to revolutionize. 00:10:20.901 --> 00:10:24.501 It's like, we, we don't really do like the services that apps and 00:10:24.501 --> 00:10:26.681 things enable are usually really cool. 00:10:27.021 --> 00:10:29.561 But like the tech itself is not rocket science. 00:10:29.591 --> 00:10:32.591 Like we are not doing crazy things over here. 00:10:32.591 --> 00:10:34.011 We're putting dibs on the pages. 00:10:34.021 --> 00:10:36.401 So like, yeah, it can handle. 00:10:36.466 --> 00:10:40.086 Doing dibs on pages like does, does that mean it's going to replace us? 00:10:40.116 --> 00:10:40.886 I don't think so. 00:10:41.416 --> 00:10:44.666 We still got to kind of figure out what we want those dibs to do and what 00:10:44.666 --> 00:10:49.556 the, we want them to represent and the services it's, it's app enabled 00:10:49.556 --> 00:10:51.106 services for a lot of things, right? 00:10:51.106 --> 00:10:54.646 Like what you guys are doing is literally a service that's enabled by tech. 00:10:56.966 --> 00:10:57.456 CJ: It's funny. 00:10:57.456 --> 00:10:59.646 Cause like probably. 00:11:00.196 --> 00:11:04.006 Most apps out there are going to have like some form of authentication, right? 00:11:04.316 --> 00:11:09.186 And like the way that you implement authentication is going to, if you like 00:11:09.186 --> 00:11:13.376 squint your eyes a little bit, look 90 percent the same for every single 00:11:13.376 --> 00:11:14.746 rails application out there, right? 00:11:14.786 --> 00:11:16.156 Especially if it's using device. 00:11:16.536 --> 00:11:18.996 And so if you go to auto complete, like. 00:11:19.596 --> 00:11:23.326 Some method in authentication controller that follows the same patterns and 00:11:23.746 --> 00:11:26.326 uses the same naming conventions, like the chances that it's going to 00:11:26.326 --> 00:11:30.616 end up just spitting out what someone else already did, you know, 9, 000 00:11:30.616 --> 00:11:32.216 other times, like is pretty high. 00:11:32.426 --> 00:11:38.666 And so yeah, I think the convention, the convention based like the strong 00:11:38.666 --> 00:11:40.596 conventions inside of rails, definitely. 00:11:40.991 --> 00:11:46.661 Is a huge, huge reason why I think these tools will be successful when paired with 00:11:46.661 --> 00:11:48.511 rails and yeah, fully, fully aligned. 00:11:48.521 --> 00:11:51.551 Like this is it's going to make it so much faster just because it's 00:11:51.551 --> 00:11:54.001 so much more predictable the way that you, you build rails apps. 00:11:54.041 --> 00:11:56.931 I do wonder the same thing about Python. 00:11:57.141 --> 00:12:01.121 I feel like because there's so many ways to do the same thing in 00:12:01.121 --> 00:12:03.711 Ruby, you know, the innumerable. 00:12:04.581 --> 00:12:08.831 Module in Ruby has just like a bajillion methods and all of them could be 00:12:08.831 --> 00:12:11.411 implemented with each or, you know, it's like a while loop or something. 00:12:11.411 --> 00:12:16.131 And so like, because there's so many ways to do it with Ruby, maybe that makes 00:12:16.131 --> 00:12:20.061 it a little tougher versus Python where there's like, okay, you can do this in two 00:12:20.061 --> 00:12:24.561 ways, pick, pick, or you can use a four loop or you can use a list comprehension. 00:12:24.561 --> 00:12:24.993 And that's it. 00:12:24.993 --> 00:12:26.611 Like, that's kind of like all you get for, 00:12:27.156 --> 00:12:27.486 Colin: Yeah. 00:12:27.541 --> 00:12:29.401 CJ: In, in some languages, like the the. 00:12:29.776 --> 00:12:33.596 The, just like the surface area of the methods in the standard library is so much 00:12:33.596 --> 00:12:37.556 smaller that I wonder if like autocomplete will work better for general things 00:12:37.556 --> 00:12:38.996 that are outside of web development. 00:12:39.046 --> 00:12:43.196 But yeah, like when working within rails, following the rails patterns, 00:12:43.196 --> 00:12:44.616 you know, you open up a controller. 00:12:44.901 --> 00:12:46.211 There's seven actions. 00:12:46.621 --> 00:12:50.071 They're always like putting the instance variable, the same spot, the same way, 00:12:50.081 --> 00:12:51.921 saving it, checking if it's saved. 00:12:51.931 --> 00:12:53.591 If so respond this way. 00:12:53.591 --> 00:12:57.211 If not respond this other way, set your flash message, you know, like kind of the 00:12:57.211 --> 00:13:01.441 stuff that you would get from a scaffold, which actually probably that's probably 00:13:01.441 --> 00:13:02.661 also another big part of it, right? 00:13:02.661 --> 00:13:06.201 Like the scaffold itself is generating code. 00:13:06.201 --> 00:13:08.881 That is the same for everybody with just like different names. 00:13:09.391 --> 00:13:09.821 So 00:13:10.331 --> 00:13:13.011 Colin: Well, and I'm not familiar and I've been wondering 00:13:13.011 --> 00:13:15.871 whether or not it makes sense. 00:13:15.891 --> 00:13:20.681 Like, are these models just trained on all data, right? 00:13:20.721 --> 00:13:21.901 Cloud and things like that. 00:13:22.351 --> 00:13:26.511 But could you have a model that is specifically trained on Ruby? 00:13:26.986 --> 00:13:33.266 And rails and more so than just seeing lots of copies of 00:13:33.506 --> 00:13:35.266 GitHub and projects, right? 00:13:35.266 --> 00:13:38.466 Like, like we talked about the campfire project that was supposed to be this like 00:13:38.766 --> 00:13:41.656 ideal rails out from 37 signals, right? 00:13:41.926 --> 00:13:44.531 If it was trained specifically on that, what would it? 00:13:44.711 --> 00:13:45.851 Nudge you towards, right? 00:13:45.851 --> 00:13:49.861 There might be 10 ways to do something in Ruby, but because we've trained 00:13:49.861 --> 00:13:53.711 it on these, like, ideal these ideal apps or these ideal methods, 00:13:53.721 --> 00:13:57.611 or this is more memory performant, or this is this is the preferred 00:13:57.661 --> 00:14:01.406 pagination or preferred monetization, authentication, whatever those things are. 00:14:01.686 --> 00:14:04.196 It'd be really interesting to have a model that's specifically 00:14:04.206 --> 00:14:08.506 trained for web development and for performance and things like that. 00:14:08.636 --> 00:14:11.486 I don't think that any of these are doing that even cursor. 00:14:11.986 --> 00:14:16.076 I saw, and I'll try to find it and share it, but some prompts examples where 00:14:16.076 --> 00:14:20.306 people have these things where it's like, I want you to be as terse as possible. 00:14:20.546 --> 00:14:22.986 I don't want you to explain what you're doing to me. 00:14:22.986 --> 00:14:24.426 I want you to just give me the answer. 00:14:24.776 --> 00:14:28.806 And like making it so that even the prompting is is more custom 00:14:28.806 --> 00:14:33.086 and more and more quick in cursor, you can define a lot of settings. 00:14:33.086 --> 00:14:36.246 And so people are starting to share their cursor settings page, which 00:14:36.246 --> 00:14:39.076 I think is making it a little bit viral to where people's like, this 00:14:39.076 --> 00:14:44.006 is what my cursor setup is very much similar to how people have VS code 00:14:44.006 --> 00:14:45.396 extensions and things like that. 00:14:45.406 --> 00:14:50.666 So Yeah, it'd be very interesting to see, like, I think my knowledge 00:14:50.666 --> 00:14:55.646 of LLMs is just outside, like, just at the door of like, is it enough 00:14:55.686 --> 00:14:57.506 to not to be trained on everything? 00:14:57.846 --> 00:14:59.556 Or does it make it more unique? 00:14:59.866 --> 00:15:04.876 Because how I guess, has it changed how you work with documentation at all? 00:15:06.651 --> 00:15:07.561 CJ: I think. 00:15:07.986 --> 00:15:09.366 Colin: This is a loaded question. 00:15:09.476 --> 00:15:09.766 But 00:15:09.796 --> 00:15:11.126 CJ: yeah, loaded question. 00:15:11.156 --> 00:15:22.306 I think I, hmm, when it comes to like standard library or like core library 00:15:22.356 --> 00:15:27.626 documentation for rails, I will just like ask the LLM questions, but if it 00:15:27.626 --> 00:15:30.906 comes to like a question about a third party, then I will look at the docs for 00:15:30.906 --> 00:15:34.116 the third party, but I fully expect that. 00:15:34.336 --> 00:15:38.676 Docs will follow Stripe's pattern of like ask the docs or something 00:15:38.676 --> 00:15:39.876 like that, which will come soon. 00:15:40.356 --> 00:15:48.656 But I also wonder, like, you, when you load your your repo, it could 00:15:48.666 --> 00:15:53.596 theoretically go out and index and, Get all the docs from like, you know, 00:15:53.596 --> 00:15:58.156 the GitHub repo for, yeah, for all your dependencies and then just be like, 00:15:58.176 --> 00:15:59.986 okay, now you're fully in your IDE. 00:15:59.986 --> 00:16:04.216 And if you have a question you're like, you're doing ask the docs, but 00:16:04.236 --> 00:16:07.476 like the docs are behind a rag model. 00:16:07.476 --> 00:16:10.056 That's like, I dunno, just like right there in your IDE. 00:16:10.496 --> 00:16:11.016 I do. 00:16:11.086 --> 00:16:11.366 Yeah. 00:16:11.366 --> 00:16:13.906 I'm at the same point where I'm like right on the edge of like, 00:16:14.456 --> 00:16:18.256 how could we train the models. 00:16:18.651 --> 00:16:21.571 A little better so that they match the stack that we're using. 00:16:21.571 --> 00:16:24.421 So that it's like even, even more advanced. 00:16:24.471 --> 00:16:27.231 And yeah, sometimes it'll recommend actually very often, 00:16:27.751 --> 00:16:31.721 chat GPT will recommend stuff for rails that is out of date. 00:16:32.101 --> 00:16:34.411 And I'll be like, no, no, no, it doesn't work that way anymore. 00:16:34.411 --> 00:16:37.251 Remember like now we're in Ruby three and it looks like this. 00:16:37.551 --> 00:16:38.971 And then it's like, oh yeah, you're right. 00:16:38.981 --> 00:16:40.791 Like now, now it does work like that. 00:16:40.791 --> 00:16:44.051 And so, yeah, there's, there's like probably tons of outdated 00:16:44.251 --> 00:16:46.981 apps that are following outdated patterns that it's trained on. 00:16:46.981 --> 00:16:47.881 So like, how do you. 00:16:48.916 --> 00:16:49.166 Yeah. 00:16:49.166 --> 00:16:50.746 How do you make it work? 00:16:50.796 --> 00:16:52.366 For yeah, modern rails development. 00:16:52.926 --> 00:16:57.356 I don't know the answer, but I did finally go back and listen to that full podcast 00:16:57.366 --> 00:16:59.176 with Lex Friedman and Peter levels. 00:16:59.696 --> 00:17:04.246 And one of the most interesting parts of it was Peter talking about how he 00:17:04.246 --> 00:17:10.716 how every single person who uploaded photos to photo AI, he was training 00:17:10.716 --> 00:17:14.696 a model specific to that person with their photos that they uploaded. 00:17:14.736 --> 00:17:18.136 And I'm like, Oh, I wonder if that's how, you know, if you wanted to do. 00:17:18.396 --> 00:17:21.726 Something where like you're, you take a base model, that's an open source base 00:17:21.726 --> 00:17:28.686 model about code and then train it on some like rails app repos that you pick. 00:17:28.726 --> 00:17:32.566 Maybe you just go find like 10, you know, these are the ideal rails repos. 00:17:32.576 --> 00:17:33.866 Maybe you buy them from 00:17:34.091 --> 00:17:35.101 Colin: And the docs, 00:17:35.436 --> 00:17:35.706 CJ: Yeah. 00:17:35.706 --> 00:17:36.416 And the docs. 00:17:36.446 --> 00:17:36.786 Yeah, 00:17:36.871 --> 00:17:40.151 Colin: docs, the, and like, cause I don't want Python, right? 00:17:40.151 --> 00:17:42.371 I only want Ruby and Rails. 00:17:42.371 --> 00:17:46.061 So like go super specific and hopefully it cuts down on 00:17:46.061 --> 00:17:47.821 hallucinations and things like that. 00:17:48.416 --> 00:17:48.916 CJ: totally. 00:17:49.156 --> 00:17:51.956 How are you thinking about it with regard to writing the docs? 00:17:53.486 --> 00:17:56.596 Colin: Yeah, so we can talk about this because I've been playing with a 00:17:56.596 --> 00:18:00.126 vendor and we're not using them right now, and we didn't build it ourselves. 00:18:00.126 --> 00:18:01.186 So I feel like we're gonna talk about it. 00:18:01.306 --> 00:18:08.616 But we have I have trained a model on our docs on our open API spec on our 00:18:08.616 --> 00:18:12.716 Zen desk help center articles that are public and on our GitHub issues. 00:18:13.246 --> 00:18:15.161 And it's Shockingly good. 00:18:15.231 --> 00:18:20.721 Like it, it, it, it's not pulling in all these other things. 00:18:20.721 --> 00:18:24.561 I think with discord specifically, we ran into this thing where most of the code 00:18:24.561 --> 00:18:26.901 samples out there are for discord JS. 00:18:27.471 --> 00:18:33.861 And so if you don't prompt that you want to You know, do like, I want to build a 00:18:33.861 --> 00:18:37.081 bot and I want to do it in pure like HTTP. 00:18:37.131 --> 00:18:40.561 I just want to use fetch and I don't want to use any libraries. 00:18:40.561 --> 00:18:43.001 You have to constantly prompt it to stop using discord. 00:18:43.001 --> 00:18:46.451 js because there's just so much content out there. 00:18:46.461 --> 00:18:48.341 They've done really good job of doing guides. 00:18:48.751 --> 00:18:51.571 Our docs don't have a lot of code samples cause they're meant to be. 00:18:52.046 --> 00:18:55.596 like the more implementation docs that other developers will use 00:18:55.596 --> 00:18:57.026 for their libraries and things. 00:18:57.456 --> 00:19:01.966 And so it was cool to see, like, based on just the docs and the, 00:19:02.156 --> 00:19:06.436 the spec itself, like it was able to understand and, and, and know 00:19:06.436 --> 00:19:07.916 what, what, what I was trying to do. 00:19:08.306 --> 00:19:13.316 And I like that idea of, like, it, it, it's still, you still need content. 00:19:13.936 --> 00:19:14.786 to train it on. 00:19:15.256 --> 00:19:17.466 And I knew it was getting it from the docs. 00:19:17.466 --> 00:19:19.716 So like writing the docs is very important still. 00:19:20.206 --> 00:19:25.006 Otherwise there wouldn't be any food for it to consume and then come back and say, 00:19:25.006 --> 00:19:27.276 Oh, I've learned how this thing works. 00:19:27.276 --> 00:19:27.446 Right. 00:19:27.446 --> 00:19:29.586 Cause just the open API spec is just the end points. 00:19:29.936 --> 00:19:33.216 So it's not going to be able, like even with descriptions and field names and 00:19:33.216 --> 00:19:37.766 stuff, it's not going to be able to like figure out the pros around that. 00:19:38.226 --> 00:19:40.576 At least not that I'm aware of, right? 00:19:40.576 --> 00:19:45.496 It's like, if this has been trained on all rest APIs and specifically ours, then 00:19:45.686 --> 00:19:48.986 it could be pretty interesting because you're like, oh, this is OAuth, right? 00:19:48.996 --> 00:19:51.056 At the end of the day, this is how OAuth works. 00:19:51.086 --> 00:19:54.286 And then you could even probably detect if someone is doing something 00:19:54.296 --> 00:19:57.196 a little bit different, like, oh, they have a special flavor of this 00:19:57.226 --> 00:20:01.726 OAuth that, you know, kind of breaks tradition and this is how you use it. 00:20:03.096 --> 00:20:05.786 So I've been trying to think about like, whether or not it makes sense 00:20:06.286 --> 00:20:11.396 To even put a search box or like an AI prompt box on the homepage of our docs. 00:20:11.846 --> 00:20:14.106 We have an onsite this week and I'm going to throw that out 00:20:14.106 --> 00:20:15.646 as a, just a fun brainstorm. 00:20:16.116 --> 00:20:19.286 But we also have a hack week later this year, so maybe, maybe I'll 00:20:19.476 --> 00:20:22.106 save it for that and just build it. 00:20:22.676 --> 00:20:23.906 CJ: That sounds like a fun project. 00:20:24.426 --> 00:20:29.256 It, it sounds like, well, one of the things that is falling out of 00:20:29.276 --> 00:20:32.376 like reading the docs too, is just like discovery of new features. 00:20:32.966 --> 00:20:39.236 And this is, I don't know, I guess like it's coming up for me when I'm thinking 00:20:39.236 --> 00:20:43.956 about all of these different tools and editors, like, how do you even know? 00:20:44.366 --> 00:20:49.396 What is possible when the features that are coming out for these 00:20:49.566 --> 00:20:51.746 editor tools are coming out so fast. 00:20:51.756 --> 00:20:58.026 And so like one problem I remember having with building buckets is that the stripe. 00:20:58.396 --> 00:21:01.516 API and the features that they would support would change 00:21:01.526 --> 00:21:02.656 and I wouldn't know about it. 00:21:03.076 --> 00:21:06.466 And I'm like, I don't know if there's a way to like subscribe to changes 00:21:06.476 --> 00:21:10.886 for certain pages or subscribe to, you know, I want to be alerted so that 00:21:10.896 --> 00:21:16.736 I know that my problem is fixed or like the product gap is, is filled in 00:21:16.736 --> 00:21:20.306 order to like retry or like, you know, be open to trying something again. 00:21:20.726 --> 00:21:26.261 And so what I was surprised by is like the last time I used VS code with and 00:21:26.621 --> 00:21:30.761 then they had like chat with your code and the paint brushes type things. 00:21:30.761 --> 00:21:33.301 But I feel like there's actually like full patterns that have 00:21:33.711 --> 00:21:35.381 started to fall out of continue. 00:21:35.381 --> 00:21:36.031 dev. 00:21:36.061 --> 00:21:37.911 And now they're built into copilot. 00:21:37.931 --> 00:21:42.671 So you can do like command I and command K and command L to like chat with or 00:21:42.711 --> 00:21:45.831 edit or like suggest refactorings. 00:21:46.391 --> 00:21:50.751 And, One of the problems I'm seeing is like this, it's just coming out so fast 00:21:51.211 --> 00:21:56.241 that I don't even know like which of these editors is the best right now, or 00:21:56.241 --> 00:22:00.131 like which features I want, because I don't even know like what's possible. 00:22:00.441 --> 00:22:04.671 And so I think there's a, there's a huge opportunity right now for dev rel, 00:22:05.101 --> 00:22:10.876 specifically in the like Editor and tool space to show people how to leverage these 00:22:10.886 --> 00:22:13.916 things to get like crazy fast at building. 00:22:14.366 --> 00:22:18.076 And there was a tick tock I watched recently where someone was talking about, 00:22:18.076 --> 00:22:21.956 like, find, find a couple of friends who are like playing around with AI and 00:22:21.956 --> 00:22:23.936 just pair with them on random stuff. 00:22:23.936 --> 00:22:25.926 Like how are they organizing their email? 00:22:25.936 --> 00:22:30.346 How are they, you know, writing up a doc or, you know, using Excel 00:22:30.356 --> 00:22:33.146 or doing SQL queries or whatever. 00:22:33.146 --> 00:22:37.336 Like, just try to like, Try to pair and cross pollinate with the 00:22:37.336 --> 00:22:40.066 different tools that people are using because they're coming out so fast. 00:22:40.066 --> 00:22:42.956 And there's so many of them that it's impossible to stay on top of. 00:22:43.486 --> 00:22:47.896 So other than the, the like features that I've suggested for editing, 00:22:47.906 --> 00:22:50.876 like, are there other things that you're using or like other ways that 00:22:50.876 --> 00:22:54.166 you're using this stuff, whether it's like writing docs or code, like, 00:22:55.556 --> 00:22:56.156 Colin: Yeah. 00:22:56.396 --> 00:23:00.316 I, so I have been looking for, and I'll put this out there if anyone 00:23:00.316 --> 00:23:05.636 knows of one, but I might, I was going to go build it and then I discovered 00:23:05.866 --> 00:23:07.286 the API doesn't exist anymore. 00:23:07.286 --> 00:23:10.076 So I put my docs into Grammarly. 00:23:10.766 --> 00:23:13.616 For some reason, Grammarly does not play nice with VS code. 00:23:13.806 --> 00:23:15.646 And I even had the desktop app. 00:23:16.596 --> 00:23:20.746 Grammarly, I mean, has been doing AI, they, before LLM, right? 00:23:20.746 --> 00:23:21.536 They've been doing. 00:23:22.101 --> 00:23:26.651 Grammar and all this stuff, but what I really do want and I think a lot of 00:23:26.651 --> 00:23:31.391 engineers could benefit from having like a true grammarly VS code extension and 00:23:31.621 --> 00:23:32.961 they killed their developer platform. 00:23:33.021 --> 00:23:33.921 Like it's just gone. 00:23:34.581 --> 00:23:36.841 You go to there and the API has been shut down. 00:23:36.851 --> 00:23:38.091 They do not let anyone. 00:23:38.101 --> 00:23:42.491 And I'm assuming, I don't know if it was similar to like when Twitter like 00:23:42.911 --> 00:23:49.121 crapped all over their developer platform or what, but like it felt like they got 00:23:49.121 --> 00:23:51.421 rid of their developer advocates and. 00:23:51.961 --> 00:23:54.351 And maybe you're only just doing partnerships or something. 00:23:54.821 --> 00:23:59.986 But there are not very many good models for that and or plugins. 00:24:00.296 --> 00:24:02.986 Like there's like a latex one that kind of it's like, Hey, 00:24:02.986 --> 00:24:05.056 this word is a double extra word. 00:24:05.406 --> 00:24:08.886 But Grammarly is really good at being like trying to teach you around passive 00:24:08.906 --> 00:24:10.626 and active voice and things like that. 00:24:11.136 --> 00:24:13.906 So I've been using it, but I've been literally writing my docs and then 00:24:13.906 --> 00:24:18.316 copy and pasting them into Grammarly because even the overlay, like Mac 00:24:18.326 --> 00:24:22.636 app isn't catching my stuff like it does when I put it into their UI. 00:24:23.136 --> 00:24:24.756 So that that's a big one. 00:24:24.756 --> 00:24:27.626 Like I'm, I'm finding that it is making me a better writer. 00:24:27.676 --> 00:24:31.896 There's a, there's a big drama going on right now with NaNoWriMo, which 00:24:31.896 --> 00:24:36.296 is the November national national writing month where they were 00:24:36.336 --> 00:24:40.586 basically saying that AI generation is okay for you to write your novel. 00:24:41.136 --> 00:24:42.796 And they're also pushing their own. 00:24:43.176 --> 00:24:44.606 LLM tool. 00:24:45.266 --> 00:24:47.696 And so a lot of the tweets were really funny where they're like, 00:24:47.696 --> 00:24:50.876 Oh my God, I'm, it's not even November and I'm done with my novel. 00:24:50.876 --> 00:24:51.916 I can't wait to read it. 00:24:55.156 --> 00:24:55.466 Right. 00:24:55.466 --> 00:25:00.986 So it's, I, I don't, when you think about what we're generating with these things, 00:25:01.066 --> 00:25:04.726 I like to still think like we could generate, Everything on the, that we could 00:25:04.726 --> 00:25:07.566 possibly think of, but like why, right? 00:25:07.566 --> 00:25:10.236 If no one's going to read it or if it's not going to bring value or if it's not 00:25:10.236 --> 00:25:14.116 worth our time, like then it's probably also not worth the energy and, and all 00:25:14.116 --> 00:25:17.166 of the other, you know, expense that comes with generating these things. 00:25:17.756 --> 00:25:19.971 For me, I still want to write it. 00:25:20.231 --> 00:25:24.681 And then I want to maybe take a pass and use it as a learning opportunity. 00:25:24.681 --> 00:25:27.461 And now I'm going to stop writing that way if it's like, 00:25:27.491 --> 00:25:28.901 Oh, stop using passive voice. 00:25:28.911 --> 00:25:29.761 Use active here. 00:25:30.251 --> 00:25:32.911 Sometimes in technical docs, it's not always obvious. 00:25:32.991 --> 00:25:34.891 Like, it's not like writing an essay. 00:25:34.951 --> 00:25:37.931 It's like you keep using this word over here and you, it 00:25:37.931 --> 00:25:38.911 isn't what you think it means. 00:25:38.941 --> 00:25:42.731 Or you know, especially when you're talking about APIs and things like that. 00:25:43.251 --> 00:25:43.316 Yeah. 00:25:44.486 --> 00:25:48.306 The one that, though, it was a funny one, eh, it's actually not important, 00:25:48.316 --> 00:25:51.436 but Yeah, so copy is a big one. 00:25:51.766 --> 00:25:57.186 When you were talking about training it on specific things, we mostly are focused 00:25:57.186 --> 00:26:00.816 on generating code, but I would almost be interested to like have an IDE where 00:26:00.816 --> 00:26:02.216 you're like running your app in it. 00:26:02.756 --> 00:26:06.406 And it's watching all the requests and all of the execution. 00:26:06.426 --> 00:26:08.216 And it's like aware of your app itself. 00:26:08.646 --> 00:26:11.776 And it's like, hey, this code over here, like is not good. 00:26:12.376 --> 00:26:15.546 Like every time you do this thing, we're making five trips to the database. 00:26:16.091 --> 00:26:21.151 Like, the AI can understand that so there's almost this, like, once the code's 00:26:21.171 --> 00:26:27.921 running, this pager duty monitoring thing, where it's like, don't just optimize 00:26:27.921 --> 00:26:31.061 the code path, but also understand that, like, the executions are taking 00:26:31.061 --> 00:26:36.163 really long, and you're, you have no index on a database, right, that, on 00:26:36.163 --> 00:26:38.079 a table that you really should have. 00:26:38.079 --> 00:26:38.626 Mm hmm. 00:26:38.841 --> 00:26:41.891 CJ: It's like the scout APM performance monitoring metrics. 00:26:41.891 --> 00:26:45.701 But like, yeah, interpreted by your, that would be amazing. 00:26:45.701 --> 00:26:48.451 Like literally that's a giant problem we're trying to solve is like, how 00:26:48.451 --> 00:26:50.021 do we drive down memory consumption? 00:26:50.481 --> 00:26:53.461 And I am going in copying and pasting and like, why the hell is 00:26:53.461 --> 00:26:55.011 this making so many allocations? 00:26:56.181 --> 00:26:58.921 And yeah, like it's just looking at scout APM. 00:26:59.371 --> 00:27:02.461 And then going in the code, copying a bunch of code that scout told me is, 00:27:02.511 --> 00:27:06.281 is like, you know, the culprit pasting it in Claude, hoping that Claude gives 00:27:06.281 --> 00:27:10.391 me something that is like maybe a thread of a, an idea of where to go. 00:27:10.891 --> 00:27:11.661 So interesting. 00:27:12.096 --> 00:27:14.836 Colin: but it's almost like a little like by bug, like, like, 00:27:14.846 --> 00:27:17.066 just, just hang out and watch this. 00:27:17.146 --> 00:27:18.716 And then you tell me what you see, 00:27:19.241 --> 00:27:19.831 CJ: totally, 00:27:19.986 --> 00:27:21.506 Colin: this is the true AI assistant. 00:27:21.516 --> 00:27:25.026 Like, I don't actually don't care about writing the code as much as that, right? 00:27:25.036 --> 00:27:28.336 Like, reaching out and doing an integration, like send this to 00:27:28.336 --> 00:27:31.716 Stripe, send this to this, pull this in here, talk to the database. 00:27:32.211 --> 00:27:32.741 CJ: totally. 00:27:33.021 --> 00:27:37.011 So one of the, I was just like digging through settings today. 00:27:37.341 --> 00:27:38.831 This is a great example where it's like. 00:27:39.626 --> 00:27:42.646 They need developer advocates to go and like talk about how this stuff like 00:27:42.896 --> 00:27:44.826 is what's coming out and what's new. 00:27:44.826 --> 00:27:46.826 But I was digging through the cursor settings. 00:27:47.306 --> 00:27:52.756 And if you go into beta, they have something it's, it's in 00:27:52.756 --> 00:27:56.636 the beta tab, but it's marked alpha and it's called AI review. 00:27:57.056 --> 00:28:05.406 And it's supposed to use chat GPT to scan your current PR diff for books. 00:28:05.886 --> 00:28:06.016 Colin: Mm 00:28:06.126 --> 00:28:07.966 CJ: So it's going to like look and see if it can find a 00:28:07.966 --> 00:28:10.346 bug in your, in your PR like. 00:28:10.856 --> 00:28:11.446 That's great. 00:28:11.476 --> 00:28:12.676 That's a, that's a cool idea. 00:28:12.836 --> 00:28:16.376 So yeah, tell you before you commit something that you've got a bug. 00:28:16.836 --> 00:28:18.256 But yeah, I don't know. 00:28:18.336 --> 00:28:21.956 I coming back to your point about authoring and like wanting to have 00:28:21.956 --> 00:28:27.316 grammarly built into VS code, it might be interesting to have like a, 00:28:27.986 --> 00:28:33.046 an LSP or something like that's based on technical writing or based on kind 00:28:33.046 --> 00:28:34.436 of like different flavors of writing. 00:28:34.436 --> 00:28:36.776 And it goes through and like helps you update. 00:28:37.016 --> 00:28:40.376 Are you mostly authoring and Markdown or like, are you, you're right. 00:28:40.376 --> 00:28:40.626 Okay. 00:28:40.626 --> 00:28:40.936 Yeah. 00:28:41.181 --> 00:28:42.331 Colin: Yeah, mostly markdown. 00:28:42.711 --> 00:28:46.051 So yeah, I'll even send it with all the markdown tags and everything 00:28:46.061 --> 00:28:48.551 because I'm not going to go clean them out and then put them back in 00:28:48.551 --> 00:28:52.851 so I'll just say like, what is wrong with this and it does it pretty well, 00:28:52.851 --> 00:28:54.611 but I would love to have it in place. 00:28:54.651 --> 00:28:58.091 But if you've used the grammarly UI, I can also understand how hard that 00:28:58.091 --> 00:29:00.361 would be in VS code because they do. 00:29:00.731 --> 00:29:03.381 They like highlight everything in different colors and show you 00:29:03.381 --> 00:29:06.381 like, This many weird grammar things you should think about. 00:29:06.381 --> 00:29:07.101 This is passive. 00:29:07.101 --> 00:29:10.621 This is, and then they like make you upgrade to get the, like, we recommend 00:29:10.651 --> 00:29:13.591 different words and you don't get to know what they are unless you pay. 00:29:13.621 --> 00:29:17.401 So I can understand that they're trying to probably create a moat in this world 00:29:17.421 --> 00:29:23.301 of all the AI things that like you probably could pull off a DIY grammarly 00:29:23.301 --> 00:29:25.051 on your own and like an afternoon. 00:29:25.071 --> 00:29:25.471 So. 00:29:26.276 --> 00:29:26.606 CJ: Yeah. 00:29:27.046 --> 00:29:29.226 So for work, we're reading We've been doing these book clubs. 00:29:29.226 --> 00:29:30.006 I love these book clubs. 00:29:30.016 --> 00:29:32.496 They're that's like how we read the unreasonable hospitality. 00:29:32.906 --> 00:29:35.196 And so the one we're going through now is called smart brevity. 00:29:35.826 --> 00:29:40.896 And it's by, I think it's, yeah, it's by the people who made Axios and Politico 00:29:41.306 --> 00:29:45.546 but the idea is about trying to like, be very clear with your communication. 00:29:45.996 --> 00:29:51.726 And several of the chapters are basically sales pitches for Axios HQ, 00:29:51.986 --> 00:29:55.406 which is their like smart brevity AI. 00:29:55.581 --> 00:30:02.981 Whatever thing that tries to re word chunks of text into their recommended 00:30:02.981 --> 00:30:08.341 framework and in order to be like more yeah, more concise and brief, which 00:30:08.341 --> 00:30:12.811 I think is definitely relevant for technical writing, especially when you're 00:30:12.811 --> 00:30:14.931 trying to like organize thoughts and be. 00:30:15.431 --> 00:30:18.541 Very direct and concise, especially for working with 00:30:18.581 --> 00:30:20.881 impatient, discerning developers. 00:30:21.191 --> 00:30:22.521 I, the book is great. 00:30:22.531 --> 00:30:27.021 It's got a bunch of tips and tricks, but yeah, I would say several of the chapters 00:30:27.021 --> 00:30:30.831 are just like pitching this smart brevity thing, but it definitely comes to mind 00:30:30.831 --> 00:30:35.141 that in order to communicate effectively, we're going to have this embedded. 00:30:35.711 --> 00:30:40.111 You know, like LLM tools embedded in every single input box that we encounter. 00:30:40.111 --> 00:30:45.111 And so how do we make it the best, smoothest experience ever? 00:30:45.111 --> 00:30:47.421 Whether you're writing a doc or you're writing, you know, a 00:30:47.421 --> 00:30:48.751 new feature for an application. 00:30:49.231 --> 00:30:53.511 And yeah, like the, the IDEs have all this built in to highlight 00:30:53.511 --> 00:30:58.011 stuff, different colors, or show you warnings or show you different things. 00:30:58.351 --> 00:31:03.161 Maybe the limitation for Grammarly was like, how do we make people pay for this? 00:31:03.171 --> 00:31:03.471 But 00:31:04.391 --> 00:31:07.361 Colin: I mean, they've been around for a while, so it probably was. 00:31:07.471 --> 00:31:09.251 I think we knew someone at Grandmama. 00:31:09.251 --> 00:31:11.451 I might have to reach out and find out what the deal was. 00:31:12.051 --> 00:31:14.501 I think we know a devrel that was there. 00:31:14.551 --> 00:31:18.531 The smart brevity thing reminds me of like the opposite of what we're 00:31:18.531 --> 00:31:20.681 seeing with the generated copy. 00:31:20.681 --> 00:31:26.761 Cause like apparently kids can't, and are only writing essays by using this now. 00:31:26.801 --> 00:31:30.791 And so I had a friend who was applying to YC and they sent, 00:31:30.851 --> 00:31:32.221 had me read over their answers. 00:31:32.261 --> 00:31:35.011 And I was like, this just feels icky. 00:31:35.181 --> 00:31:36.571 Like, did you use AI on this? 00:31:36.571 --> 00:31:39.861 Like, this doesn't feel like a human talking to a human. 00:31:39.871 --> 00:31:40.451 And. 00:31:40.656 --> 00:31:43.136 I think too many people are using it as that crutch. 00:31:43.136 --> 00:31:46.706 Like, the Grammarly one is really good at not doing that. 00:31:46.706 --> 00:31:50.076 And I actually would be curious to take some purely generated chat GPT 00:31:50.076 --> 00:31:52.886 and just drop it into Grammarly to see what Grammarly says about it. 00:31:53.426 --> 00:31:56.536 Because I think somewhere along the line people thought that putting in 00:31:56.536 --> 00:31:59.236 20 to 5 words was a good thing, right? 00:31:59.246 --> 00:32:02.636 Like you're in college and you need to pad that essay to get to your word count. 00:32:03.006 --> 00:32:07.806 That's one thing, but we just, I don't think people want to be talked to with 00:32:07.806 --> 00:32:13.036 these like very weird, strange just like slightly off word choice where like, 00:32:13.076 --> 00:32:16.756 yeah, maybe that was technically correct, but like, we don't talk like that. 00:32:17.316 --> 00:32:20.986 And when I was training the docs on it, it was pretty good. 00:32:21.486 --> 00:32:25.006 I don't know that it was necessarily answering in the style of our, the writing 00:32:25.006 --> 00:32:29.876 of our docs, but I wouldn't be surprised if it was because it felt like the docs, 00:32:29.966 --> 00:32:34.666 it didn't feel like this other thing came in and was like, you must proceed this 00:32:34.666 --> 00:32:38.256 way and use caution and code passionately. 00:32:38.276 --> 00:32:40.986 Like there's always these like weird words where they're like, like, Oh, 00:32:40.986 --> 00:32:42.176 you're gonna have so much fun doing this. 00:32:42.226 --> 00:32:46.046 I was like, no one is going to be like, Oh, I, I sent this email passionately. 00:32:46.046 --> 00:32:47.466 It's like, that was the wrong word there. 00:32:47.826 --> 00:32:48.216 CJ: Yeah. 00:32:48.576 --> 00:32:48.886 Colin: yeah. 00:32:50.476 --> 00:32:50.726 CJ: Yeah. 00:32:50.726 --> 00:32:51.396 Just, yeah. 00:32:51.396 --> 00:32:54.433 Figuring out the training data set. 00:32:54.483 --> 00:33:00.643 That's not just like the entire opus of the internet as your input is important. 00:33:00.683 --> 00:33:01.633 And so like, yeah. 00:33:01.633 --> 00:33:03.213 Training on other. 00:33:03.798 --> 00:33:07.528 Material that's similar to what you want to create is like super important. 00:33:07.618 --> 00:33:12.008 I was just Googling like, what are the open source models that you 00:33:12.008 --> 00:33:13.688 would use to rebuild Grammarly? 00:33:13.898 --> 00:33:17.898 And it looks like there are several from like Mistral and others that you 00:33:17.898 --> 00:33:20.708 could take and yeah, mess around with. 00:33:20.818 --> 00:33:21.858 So yeah, I don't know. 00:33:21.878 --> 00:33:23.078 I definitely. 00:33:23.368 --> 00:33:26.608 That feels like the next step in, in my learning at least is like, how do 00:33:26.608 --> 00:33:32.418 we take a model and then like, not just fine tune it, but like adjust it to make 00:33:32.418 --> 00:33:36.738 it do a very specific job, much better. 00:33:37.168 --> 00:33:38.878 So, I, let's 00:33:39.113 --> 00:33:42.473 Colin: think that's where everyone's trying to get general purpose AI, right? 00:33:42.483 --> 00:33:46.113 They want general intelligence, but turns out that's hard. 00:33:46.443 --> 00:33:52.923 It's ethically fraught versus having your own like thing, right? 00:33:52.933 --> 00:33:57.283 Like, honestly, even if you just had your own personal AI, that was like followed 00:33:57.283 --> 00:34:01.283 you, like literally knew all the things you've ever done in all of your companies. 00:34:01.773 --> 00:34:04.343 You know, the, the, a very specific set of skills that you've 00:34:04.373 --> 00:34:05.793 developed over a long career. 00:34:06.343 --> 00:34:06.923 And. 00:34:07.343 --> 00:34:11.123 Then you, it, it's like just able to pull up things and it knows 00:34:11.123 --> 00:34:14.113 like, yeah, it, what would CJ do? 00:34:14.233 --> 00:34:14.563 Right. 00:34:14.563 --> 00:34:19.663 And not necessarily anticipate and do it for you, but help you along the way. 00:34:19.663 --> 00:34:21.663 It's, it's your, it's your exo suit. 00:34:22.213 --> 00:34:22.663 Yeah. 00:34:23.083 --> 00:34:26.833 Are you using anything else, like in, like you mentioned email and some other things. 00:34:26.833 --> 00:34:32.703 Are you using AI in other parts of, of your day-to-Day, or is it just code? 00:34:38.688 --> 00:34:42.808 CJ: embeddings for just like a lot of semantic search type stuff. 00:34:42.858 --> 00:34:49.828 For features I've been using tools like or function calling for generating 00:34:50.168 --> 00:34:51.768 content for a couple of things. 00:34:52.308 --> 00:34:56.298 But yeah, like day to day, a lot of it is just like, the, or like the most impactful 00:34:56.308 --> 00:34:58.478 stuff is definitely editing for code. 00:34:58.908 --> 00:35:02.778 We added that feature that's just like type to complete for our sales 00:35:02.778 --> 00:35:06.968 team to do customer engagement based on like the full customer context. 00:35:07.473 --> 00:35:09.053 That's been running now for like a month. 00:35:09.083 --> 00:35:11.493 And I looked and it's quite expensive. 00:35:11.573 --> 00:35:14.873 Every single time we're making a call, we're sending like a pretty 00:35:14.873 --> 00:35:20.363 giant prompt because it embeds like the full list of frequently asked 00:35:20.373 --> 00:35:23.663 questions and previous conversation and all the history of that customer. 00:35:23.703 --> 00:35:28.063 And so it, yeah, I think we'll have to get more creative with 00:35:28.083 --> 00:35:32.163 that eventually, but try to think of other, other use cases playing 00:35:32.163 --> 00:35:33.983 around, there was like that teal draw. 00:35:34.328 --> 00:35:36.368 Tool where you could kind of like enter stuff. 00:35:36.878 --> 00:35:43.538 Yeah, I mean, I guess music just like Descript for editing, like yeah. 00:35:43.538 --> 00:35:46.578 I tried generating some like music tracks for backgrounds. 00:35:46.918 --> 00:35:49.968 So yeah, things, I guess, like on the next, definitely like the 00:35:49.968 --> 00:35:54.058 next frontier is like, how do you make it, Specialized to you. 00:35:55.028 --> 00:35:56.188 Many good examples of that. 00:35:56.198 --> 00:35:59.488 Like, how do you train stable, a stable diffusion model that looks, that can 00:35:59.488 --> 00:36:04.548 generate photos that look like you, how can you train an 11 labs voice model, 00:36:04.548 --> 00:36:08.068 or like, how can you train a voice model that sounds like you, how can you 00:36:08.068 --> 00:36:10.508 train something to write a blog post? 00:36:10.508 --> 00:36:11.788 That sounds like you wrote it. 00:36:12.578 --> 00:36:14.758 How do you train something that it's just Ruby and rails? 00:36:15.258 --> 00:36:19.758 So yeah, that, that definitely feels like The next rabbit hole worth jumping down. 00:36:19.828 --> 00:36:20.558 So, Mm. 00:36:20.808 --> 00:36:24.278 Colin: Yeah, I've been thinking about, I don't know how to put the words around it 00:36:24.278 --> 00:36:29.688 yet, but there's this, like, those are all aids in creating something at the end of 00:36:29.688 --> 00:36:33.778 the day that someone wants to listen to or wants to watch or wants to learn from. 00:36:34.338 --> 00:36:34.718 CJ: Mm. 00:36:35.398 --> 00:36:39.908 Colin: feels like a dumping ground of just AI generated garbage right now. 00:36:39.978 --> 00:36:43.198 But, like, there's a person specifically that I have in mind 00:36:43.228 --> 00:36:45.868 that has been doing these, like, 00:36:45.915 --> 00:36:46.736 CJ: Mm. 00:36:46.918 --> 00:36:48.908 Colin: they're definitely, I think, prompts. 00:36:49.338 --> 00:36:51.878 that they asked for like, Hey, give me some prompts that I 00:36:51.878 --> 00:36:53.668 can answer as LinkedIn posts. 00:36:53.668 --> 00:36:55.158 They're like thought leadership posts. 00:36:55.158 --> 00:36:55.388 Right. 00:36:55.908 --> 00:36:59.148 But they, so then they like do this, like waxing philosophical, 00:36:59.148 --> 00:37:00.588 like fucking question. 00:37:00.648 --> 00:37:02.098 We're getting our first leap in there. 00:37:02.478 --> 00:37:06.708 And then they generate an image with grok and it's like, okay, so you 00:37:06.708 --> 00:37:10.838 just created like this most, like lowest common denominator milk toast, 00:37:10.838 --> 00:37:13.258 like thing for people to consume. 00:37:13.258 --> 00:37:16.258 I had to unfollow them because I'm like, every time this pops up, 00:37:16.258 --> 00:37:17.678 it's like, It's a waste of my time. 00:37:17.728 --> 00:37:19.748 It's like literally no value. 00:37:20.168 --> 00:37:24.628 Some thought leadership in entrepreneurship or startups or we'll 00:37:24.628 --> 00:37:28.678 save founder mode, startup mode or whatever manager mode for another day. 00:37:28.678 --> 00:37:32.398 But like no adding to the conversation whatsoever. 00:37:32.408 --> 00:37:36.768 And then the image that you generated was this like, you know, guy in a field with 00:37:36.768 --> 00:37:38.908 five, you know, with five extra fingers. 00:37:39.358 --> 00:37:42.678 And it's just not, what anyone needed. 00:37:42.788 --> 00:37:46.068 Like, so, and everyone's doing it on LinkedIn. 00:37:46.618 --> 00:37:48.138 They're probably doing it on Twitter. 00:37:48.718 --> 00:37:51.508 You have probably had people who are like, Oh, I'll be your copywriter and they're 00:37:51.508 --> 00:37:54.828 going and generating a bunch of things and you know, sending you a bill for it. 00:37:55.298 --> 00:37:58.768 So that stuff that I'm just like trying to wrestle with, you 00:37:58.768 --> 00:38:00.588 know, is it just like that stuff? 00:38:01.568 --> 00:38:05.048 And people are going to ignore it, or is it really going to, you 00:38:05.048 --> 00:38:08.588 know, be something that we have to like, maybe it's just a fad. 00:38:08.668 --> 00:38:11.438 Cause I know people were generating images with mid journey for a long time. 00:38:11.438 --> 00:38:13.008 And then they're like, cool. 00:38:13.048 --> 00:38:16.958 There's only so many times we can do like pictures of your cat around space. 00:38:16.998 --> 00:38:17.338 Right. 00:38:17.378 --> 00:38:19.848 Before we start to try to use it for other things. 00:38:20.198 --> 00:38:24.788 I don't use a lot of image generation stuff just because it's like, I don't 00:38:24.798 --> 00:38:29.228 have It's not good usually like if I want I need art I need to go jump into 00:38:29.228 --> 00:38:33.518 figma and make some illustrations and stuff or work with a designer and have 00:38:33.518 --> 00:38:36.898 them make something from scratch because so that's that's been rolling around 00:38:36.898 --> 00:38:41.548 and I'm just like now that I see it I just unfollow or hide it or whatever 00:38:41.608 --> 00:38:43.288 but have you seen any of this stuff? 00:38:43.718 --> 00:38:44.628 CJ: Yeah, totally. 00:38:44.648 --> 00:38:47.498 And just, I think it was yesterday or today. 00:38:48.158 --> 00:38:53.218 I got a pop up from LinkedIn that said, you can, you know, Do do premium 00:38:53.218 --> 00:38:58.838 for 30 days or something and in the pitch for premium, it now says AI 00:38:58.838 --> 00:39:00.478 writing assistance and whatever. 00:39:00.478 --> 00:39:03.978 And so I'm like, I wonder if that's where it's coming from is like, maybe 00:39:04.018 --> 00:39:08.598 LinkedIn has some feature that's like actually making it so that people can 00:39:08.868 --> 00:39:11.148 more easily create that kind of content. 00:39:11.338 --> 00:39:14.888 And it also comes to mind too, that like, we've been trained that putting 00:39:14.888 --> 00:39:20.168 a picture in your post will make your post more likely to like get attention. 00:39:20.398 --> 00:39:23.938 And so if people don't have a good picture, they're just making shit 00:39:23.948 --> 00:39:26.628 up with like, you know, or they're, yeah, they're generating it. 00:39:26.658 --> 00:39:27.088 And 00:39:27.603 --> 00:39:31.593 Colin: Yeah, but it's like shovelware, like they're just putting stuff out to, 00:39:31.733 --> 00:39:33.803 to try to build their brand or whatever. 00:39:33.803 --> 00:39:35.613 And it's like, you're just building this one. 00:39:35.623 --> 00:39:41.753 Like you did not do anything of talent or skill, which I do think that there is 00:39:41.753 --> 00:39:47.753 going to be this diff of people who can and people who have to prompt for sure. 00:39:47.763 --> 00:39:48.823 Like it's going to happen. 00:39:49.223 --> 00:39:52.443 And if you're an engineer out there, you're relying on these tools, make 00:39:52.443 --> 00:39:56.668 sure you can still do it without it, because I'm pretty sure Pretty sure that 00:39:56.668 --> 00:40:00.988 like when we do interviews, we do not allow you to use AI in your interviews. 00:40:01.188 --> 00:40:06.598 So if you are coming out of a bootcamp and you used AI to get through your bootcamp, 00:40:08.238 --> 00:40:09.948 turn it off, make sure you do it. 00:40:09.948 --> 00:40:12.938 I think you and I both are used to like doing the same thing over and over again 00:40:12.938 --> 00:40:16.078 until it just becomes a thing we can do. 00:40:16.128 --> 00:40:16.438 Right? 00:40:16.448 --> 00:40:20.798 Like we're prompting ourselves into like, all right, build this, build 00:40:20.798 --> 00:40:22.778 it again, delete it, build it again. 00:40:22.963 --> 00:40:24.183 until you really know it. 00:40:24.233 --> 00:40:28.143 And it's okay if you don't need to touch that for a long time to come back to 00:40:28.143 --> 00:40:33.783 it and refresh or ask chatgbt to prompt you on like, how do you do that again? 00:40:33.783 --> 00:40:38.063 Like, most people don't do a lot every day and couldn't rebuild it from scratch. 00:40:39.223 --> 00:40:39.893 And that's okay. 00:40:40.043 --> 00:40:42.483 But knowing how it works is a good idea. 00:40:43.143 --> 00:40:43.353 CJ: Yeah. 00:40:43.353 --> 00:40:44.133 Definitely. 00:40:44.163 --> 00:40:47.193 From using Cursor and saying like, generate this feature for me. 00:40:48.243 --> 00:40:52.203 And having it only work like one in three ish, I don't know, maybe, 00:40:52.263 --> 00:40:54.063 maybe the hit rate is one in three where it generates something. 00:40:54.063 --> 00:40:54.573 I'm like, yeah. 00:40:54.573 --> 00:40:55.563 Like, let's accept that. 00:40:55.563 --> 00:40:56.643 That seems legit. 00:40:57.203 --> 00:41:00.283 That makes me a little bit nervous for people who are like, I don't really 00:41:00.283 --> 00:41:04.413 know how to code, but I built this thing 'cause I used Cursor and, you know, 00:41:04.413 --> 00:41:06.393 I, I just prompted my way through it. 00:41:06.393 --> 00:41:08.193 I'm like, yeah, cool. 00:41:08.193 --> 00:41:08.673 Like, 00:41:09.183 --> 00:41:10.023 Colin: one to accept? 00:41:10.318 --> 00:41:11.008 CJ: Exactly. 00:41:11.028 --> 00:41:11.298 Yeah. 00:41:11.298 --> 00:41:12.958 They're just like, that looks good. 00:41:12.988 --> 00:41:15.328 You know, maybe they try to run it and they're like, I ran into an error. 00:41:15.358 --> 00:41:16.348 Copy paste the error. 00:41:16.998 --> 00:41:17.878 Oh yeah, you're right. 00:41:17.888 --> 00:41:18.948 Change it to this thing. 00:41:18.978 --> 00:41:22.538 And then you end up with in the same way that it's not 00:41:22.538 --> 00:41:25.538 fun to read AI generated text. 00:41:25.948 --> 00:41:29.678 It probably will be very hard to maintain an AI generated application. 00:41:29.748 --> 00:41:33.538 And so I dunno, there there's, I don't want to poo it too much. 00:41:33.558 --> 00:41:38.788 Cause I guess like, if you are using it as a way to like build a business 00:41:38.788 --> 00:41:41.378 and you don't actually care about the underlying tech, that's cool. 00:41:41.468 --> 00:41:41.788 Right. 00:41:42.173 --> 00:41:43.043 Colin: It is, yeah. 00:41:43.683 --> 00:41:46.553 And I think that's, Josh Pickford and some others have been having that 00:41:46.553 --> 00:41:50.663 conversation of there's definitely some purists who are thinking of coding 00:41:50.663 --> 00:41:53.813 as craft versus coding as a result. 00:41:54.113 --> 00:41:57.693 And Josh wants what's on the other side of the code. 00:41:57.693 --> 00:42:01.153 The code allows him to build a business, to build a family, 00:42:01.163 --> 00:42:02.383 you know, a life for his family. 00:42:03.123 --> 00:42:05.193 Yes, he wants it to be good code. 00:42:05.193 --> 00:42:06.213 Yes, whatever. 00:42:06.553 --> 00:42:11.353 But he's not going to be over here fighting some like language semantic 00:42:11.353 --> 00:42:16.933 wars because he wants to ship a product and make number go up, right? 00:42:16.963 --> 00:42:18.063 That's, that's his goal. 00:42:18.213 --> 00:42:19.553 And he wants to solve problems, right? 00:42:19.553 --> 00:42:22.053 He builds lots of apps to solve his problems. 00:42:22.733 --> 00:42:23.743 He makes those solutions. 00:42:23.743 --> 00:42:24.623 People want them. 00:42:24.643 --> 00:42:25.593 He charges for them. 00:42:25.633 --> 00:42:30.013 And at the end of the day, it doesn't matter what the code was, right? 00:42:30.033 --> 00:42:30.893 As long as it works. 00:42:31.513 --> 00:42:31.963 So, 00:42:32.923 --> 00:42:35.843 CJ: Yeah, I think that was like the big takeaway from the Peter levels argument 00:42:35.843 --> 00:42:38.163 too, is like, he's using PHP and jQuery. 00:42:38.163 --> 00:42:38.883 That's what he knew. 00:42:39.343 --> 00:42:40.533 And like, 00:42:40.893 --> 00:42:42.403 Colin: react people are so angry. 00:42:43.623 --> 00:42:48.353 CJ: yeah, I mean, yeah, that's, I don't know, it's, it's funny to think 00:42:48.363 --> 00:42:55.163 about, but yeah, trying to keep a laser focus on growing the business or yeah, 00:42:55.163 --> 00:42:59.443 growing your own sidekick or your own sidekick, growing your own side hustle. 00:42:59.443 --> 00:43:01.683 Like all of that is really Yeah. 00:43:01.683 --> 00:43:03.653 It doesn't really matter what, what the code looks like. 00:43:03.653 --> 00:43:05.183 If people are just looking at the interface, so 00:43:06.968 --> 00:43:07.298 Colin: Cool. 00:43:07.358 --> 00:43:08.178 Should we leave it there? 00:43:08.883 --> 00:43:09.993 CJ: Yeah, totally. 00:43:09.993 --> 00:43:13.343 If you want to check out the links to the resources for stuff we chatted about 00:43:13.343 --> 00:43:14.863 today, head over to build and learn. 00:43:15.093 --> 00:43:16.113 dev and yeah, we'll catch you next time. 00:43:17.138 --> 00:43:17.798 Colin: Bye friends.