WEBVTT 1 00:00:05.440 --> 00:00:09.480 Hey, folks, welcome back to another episode of JavaScript Jabber. 2 00:00:09.880 --> 00:00:14.039 This week, on our panel, we have Steve Edwards Yo yo. 3 00:00:13.960 --> 00:00:17.960 Yo, coming at you aj style from just outside Washington, DC. 4 00:00:19.559 --> 00:00:21.120 Yeah. I was trying to come up with your name 5 00:00:21.160 --> 00:00:24.600 and I was looking at Eric, so my brain kind 6 00:00:24.600 --> 00:00:26.879 of went wait, wait, wait wait. I'm Charles max Wood 7 00:00:26.879 --> 00:00:29.800 from top Endevs. And this week we have two guests. 8 00:00:29.839 --> 00:00:34.240 We have Eric Hanschett, Hello, Hello, welcome back, and we 9 00:00:34.320 --> 00:00:36.600 also have Claire. I didn't get your last name, Claire 10 00:00:36.679 --> 00:00:43.840 well gory Oh cool, sounds Italian, so you're probably awesome, all. 11 00:00:43.840 --> 00:00:47.439 Right, So I sounded French, but that's just me. 12 00:00:48.119 --> 00:00:51.240 It could be anyway. Do you guys want to just 13 00:00:51.280 --> 00:00:54.320 introduce yourselves real quickly? Got Eric on before, but Claire, 14 00:00:54.359 --> 00:00:55.719 you're new to us, so we want to know how 15 00:00:55.719 --> 00:00:56.200 awesome we are. 16 00:00:56.320 --> 00:00:59.759 Sure. I'm Claire. I'm a senior Principal engineer at as 17 00:01:00.159 --> 00:01:05.319 in agentic AI, and I work on all seeing developers 18 00:01:05.599 --> 00:01:11.719 and agents basically hero ID and the strands Open Source 19 00:01:11.719 --> 00:01:12.359 Agent SDK. 20 00:01:14.640 --> 00:01:14.959 Cool. 21 00:01:16.319 --> 00:01:19.319 Well, and I'm Erica. You've been on here a few times, 22 00:01:19.359 --> 00:01:24.760 a senior developer advocate at AWUS Huge VIEWJS fan it's 23 00:01:24.840 --> 00:01:26.400 probably the last few times you may have heard me 24 00:01:26.439 --> 00:01:28.799 on this podcast talking about it and all the work 25 00:01:28.799 --> 00:01:32.000 I've done in that community and in the JavaScript community. 26 00:01:32.040 --> 00:01:34.079 So glad to be here today and talk about some 27 00:01:34.239 --> 00:01:37.840 really exciting that we released recently. That Claire can talk 28 00:01:38.319 --> 00:01:41.599 a lot about Kiro, which is our new AI I D. 29 00:01:42.560 --> 00:01:44.719 So a little bit of background before we get started. 30 00:01:44.760 --> 00:01:47.799 So we had Eric on a month and a half 31 00:01:47.799 --> 00:01:51.159 ago somewhere around there. I forgot and found out later 32 00:01:51.200 --> 00:01:53.439 that he had been planning to talk about Kiro, but 33 00:01:53.879 --> 00:01:55.959 it turned out it wasn't quite ready for release at 34 00:01:55.959 --> 00:01:58.439 that time, so we sort of scrambled and talked about 35 00:01:58.480 --> 00:02:01.239 other things sounded we sounded really good thought. But now 36 00:02:01.480 --> 00:02:04.640 Kiro's has been released, and I know I've heard a 37 00:02:04.680 --> 00:02:08.120 lot about it on other podcasts, you know, in other 38 00:02:08.159 --> 00:02:10.240 blogs and so on, and people have been really raving 39 00:02:10.280 --> 00:02:13.039 about it. So we're finally here to talk about it 40 00:02:13.039 --> 00:02:15.199 officially and all it can do for us. 41 00:02:16.719 --> 00:02:20.479 So I'm curious. Sometimes it's obvious why people name stuff 42 00:02:20.520 --> 00:02:22.960 what they name it. Right, It's like this does email, 43 00:02:23.039 --> 00:02:26.240 so we called it email thing? Right? How did you 44 00:02:26.280 --> 00:02:27.360 come up with kiro? 45 00:02:28.960 --> 00:02:35.199 That was a AI assisted process? Actually? It actually means 46 00:02:35.439 --> 00:02:39.240 crossroads in Japanese, and so we were looking around for 47 00:02:39.280 --> 00:02:43.560 a word that would reflect the sort of new development 48 00:02:43.639 --> 00:02:47.680 style that we have now. Our development has changed so much, 49 00:02:47.800 --> 00:02:50.719 so we were at a crossroads. 50 00:02:52.159 --> 00:02:55.800 Gotcha. It's funny because I go for walks and I 51 00:02:55.879 --> 00:02:58.120 talked to Grock the whole time, so I know how 52 00:02:58.199 --> 00:03:00.879 this goes. Sometimes it's like, oh, that's a good idea, 53 00:03:01.039 --> 00:03:02.599 So do you want to explain what it is and 54 00:03:02.879 --> 00:03:05.400 how it works, maybe how it's different from some other 55 00:03:05.439 --> 00:03:08.159 things that it sounds like do the same thing. 56 00:03:08.840 --> 00:03:13.360 Well, Kiro is an AI powered ide like Eric said, 57 00:03:13.479 --> 00:03:17.719 but one of the things that makes it different is 58 00:03:17.759 --> 00:03:20.240 that we've all been kind of vibe coding for a 59 00:03:20.280 --> 00:03:22.319 bit now, and one of the things that we were 60 00:03:22.360 --> 00:03:26.439 noticing is that people really struggled to get what they 61 00:03:26.479 --> 00:03:29.280 wanted out of vibe coding, especially when it came to 62 00:03:29.360 --> 00:03:34.800 more complex tasks. And there was a kind of an 63 00:03:34.879 --> 00:03:37.879 aha moment for me when we and was the reason 64 00:03:37.879 --> 00:03:42.319 that we started Kiro, which was what if developers instead 65 00:03:42.520 --> 00:03:46.719 of saying kind of commands to an AI assistant of 66 00:03:46.759 --> 00:03:49.560 how to change their code, what if they just declared 67 00:03:49.599 --> 00:03:53.280 what they wanted? What if they defined requirements and then 68 00:03:53.319 --> 00:03:56.280 worked with the AI on requirements and how that broke 69 00:03:56.319 --> 00:03:59.439 down into individual tasks. And that was really the birth 70 00:03:59.840 --> 00:04:03.319 of what we call spectrum and development, which we have 71 00:04:03.960 --> 00:04:08.039 dedicated UX around in KIRO. That's really what makes it different. 72 00:04:08.199 --> 00:04:11.000 Spec based development is kind of what it sounds like. 73 00:04:11.080 --> 00:04:13.840 You create a specification for what you want. That could 74 00:04:13.840 --> 00:04:16.959 be a brand new application, it could be a feature, 75 00:04:17.079 --> 00:04:20.399 it could be a massive refactor in your code base. 76 00:04:20.920 --> 00:04:24.639 But you go through this process where you define the requirements, 77 00:04:24.720 --> 00:04:27.399 the use cases, user stories. You can have the AI 78 00:04:27.519 --> 00:04:31.240 generate all of those, because I remember writing these by 79 00:04:31.319 --> 00:04:34.360 hand ten fifteen years ago and it wasn't that much fun. 80 00:04:34.480 --> 00:04:38.480 And then you go through a process of working through 81 00:04:38.519 --> 00:04:41.759 the technical design with the AI and finally what tasks 82 00:04:41.759 --> 00:04:44.000 that breaks down to, and then you have a lot 83 00:04:44.000 --> 00:04:50.199 of control over steering the actual implementation of those individual tasks. 84 00:04:50.279 --> 00:04:52.920 And we've heard from folks that this has enabled them 85 00:04:53.000 --> 00:04:56.160 to at the end of the day, produce much better 86 00:04:56.480 --> 00:05:00.040 quality code and actually get to what they wanted. I 87 00:05:00.040 --> 00:05:03.279 said of this very frustrating experience with vicuding, we were 88 00:05:03.279 --> 00:05:06.720 hearing about where it's like you go ten cycles and 89 00:05:06.759 --> 00:05:08.920 that it's almost like the AI is not getting it 90 00:05:09.000 --> 00:05:09.879 getting what you want. 91 00:05:12.279 --> 00:05:14.720 Yeah, I've experienced that a little bit. I've been using 92 00:05:14.759 --> 00:05:17.959 cursor and a co pilot. I'm depending on what i'm 93 00:05:17.959 --> 00:05:21.519 working on. And you talked about it breaking it down 94 00:05:21.560 --> 00:05:24.959 into lists, And it's funny because cursor will give me 95 00:05:25.040 --> 00:05:26.600 a these are the things I'm going to do, and 96 00:05:26.639 --> 00:05:28.800 then it'll check them off as it goes right. And yeah, 97 00:05:28.839 --> 00:05:31.759 it doesn't always get it right. If it's something simple, 98 00:05:32.040 --> 00:05:34.519 probably the vast majority of the time it gets that right. 99 00:05:34.560 --> 00:05:37.959 But yeah, the second it gets complicated, it's like, Okay, 100 00:05:37.959 --> 00:05:41.639 this isn't working. Sometimes I'm still not quite sure what 101 00:05:41.680 --> 00:05:44.759 it did or how to fix it, and so I'll 102 00:05:44.800 --> 00:05:48.120 ask it to fix it, and then it still doesn't work. Yeah, 103 00:05:48.120 --> 00:05:49.920 So I get that. It sounds like this is a 104 00:05:49.959 --> 00:05:53.000 little bit more in depth than sort of the process of, 105 00:05:53.079 --> 00:05:55.639 you know, giving it a single prompt and then having 106 00:05:55.680 --> 00:05:57.639 it come up with I'm going to do this, and 107 00:05:57.680 --> 00:05:58.879 then I'm going to do that, and then I'm going 108 00:05:58.920 --> 00:06:01.120 to do that and then I'll be done. 109 00:06:01.439 --> 00:06:04.040 Yeah, and it still starts with a single prompt. But 110 00:06:04.120 --> 00:06:06.759 one of the things that I've noticed is that it's 111 00:06:06.879 --> 00:06:09.839 really hard for us to describe what's in our head 112 00:06:11.079 --> 00:06:15.079 from the get go, from a blank prompt text input box. 113 00:06:15.240 --> 00:06:17.319 And so often what I find is that there's this 114 00:06:17.399 --> 00:06:21.160 iterative process of effectively the AI trying to guess what 115 00:06:21.199 --> 00:06:25.439 I meant by this simple sentence and then being able 116 00:06:25.560 --> 00:06:28.360 to iterate on that and say, no, no, that's not 117 00:06:28.399 --> 00:06:31.000 exactly what I meant. I actually meant this, or oh, 118 00:06:31.040 --> 00:06:33.360 that's an interesting idea I didn't even think of. And 119 00:06:33.399 --> 00:06:37.360 so that process of creating this spec is really helpful 120 00:06:37.439 --> 00:06:40.759 for refining what it is you want. Often if you 121 00:06:40.839 --> 00:06:44.439 go to Vibe coding, it's kind of like you know, 122 00:06:44.720 --> 00:06:48.319 when you have an engineer who's a bit a bit fresh, 123 00:06:48.360 --> 00:06:52.160 and they always jump right to writing the code before 124 00:06:52.199 --> 00:06:55.319 actually thinking about the problem. That's kind of the Vibe 125 00:06:55.319 --> 00:06:59.040 coding experience is the AI jumped into writing the code 126 00:06:59.079 --> 00:07:02.199 before kind of understand the problem. And so this spec 127 00:07:02.319 --> 00:07:06.519 driven development helps you to kind of make sure you're 128 00:07:06.519 --> 00:07:08.720 on the same page and that you know everything that's 129 00:07:08.759 --> 00:07:10.959 in your head of what you're thinking about the application 130 00:07:11.079 --> 00:07:15.040 or the future is actually out on paper without necessarily 131 00:07:15.079 --> 00:07:17.600 having to type it out all yourself. You kind of 132 00:07:17.639 --> 00:07:20.720 just go through this iterative process of the AI kind 133 00:07:20.720 --> 00:07:23.240 of guesses writes the spec for you, and then you say, no, no, 134 00:07:23.279 --> 00:07:24.519 that's not exactly what I meant. 135 00:07:26.079 --> 00:07:28.000 Yeah. It sounds a little bit like the process I 136 00:07:28.040 --> 00:07:31.319 went through more as a freelancer than really is a 137 00:07:31.399 --> 00:07:35.120 full time employee, where you know, I'd have the client 138 00:07:35.199 --> 00:07:37.160 come to me and say, hey, we want this feature, 139 00:07:37.199 --> 00:07:38.920 and then I'd ask questions, do you want it to 140 00:07:38.920 --> 00:07:40.279 do this? Do you want it to do that? And 141 00:07:40.319 --> 00:07:41.560 you want it to look like this? Do you want 142 00:07:41.600 --> 00:07:43.480 it to blah blah blah, right, and then there was 143 00:07:43.519 --> 00:07:46.560 further iteration once we got going where it was okay, 144 00:07:46.600 --> 00:07:48.800 I have a rough version of this, and then they 145 00:07:48.879 --> 00:07:51.000 go look at it. Oh, no, this isn't quite what 146 00:07:51.040 --> 00:07:53.040 I wanted. And so by the time we're done, yeah, 147 00:07:53.040 --> 00:07:55.160 I had a fully fleshed out understanding of what they 148 00:07:55.199 --> 00:07:57.480 wanted and could build it. And it sounds like that's 149 00:07:57.519 --> 00:07:58.199 what this is doing. 150 00:07:58.279 --> 00:08:01.079 Yeah, And I love that idea of the rough version 151 00:08:01.120 --> 00:08:04.199 of it, is this what you wanted? Because so often 152 00:08:04.439 --> 00:08:07.279 we don't know what we want until we actually see 153 00:08:07.319 --> 00:08:10.839 it when it comes to application development. And you know, 154 00:08:10.920 --> 00:08:15.959 we've seen across moving from Waterfall, where we tried to 155 00:08:16.519 --> 00:08:18.959 you know, have all the requirements ironed out at the beginning, 156 00:08:19.000 --> 00:08:21.480 and that never really worked because those were not Those 157 00:08:21.480 --> 00:08:24.279 were never the actual requirements that you ended up with 158 00:08:24.360 --> 00:08:27.120 at the end once you saw the application. We went 159 00:08:27.160 --> 00:08:31.279 to agile, where we're attempting to move into sprints, where 160 00:08:31.319 --> 00:08:36.240 we have some capability of refining those requirements as we build. 161 00:08:37.039 --> 00:08:42.200 But even with something like agile development, it still takes 162 00:08:42.240 --> 00:08:44.960 so long to get to something that's working that you 163 00:08:44.960 --> 00:08:47.360 can actually show to a customer, show to a user. 164 00:08:47.480 --> 00:08:49.480 So it's going to take multiple months. And one of 165 00:08:49.480 --> 00:08:53.440 the powerful things about AI assisted coding is you can 166 00:08:53.519 --> 00:08:57.919 get to that tangible result so much faster. You can 167 00:08:58.000 --> 00:09:02.240 build entire application may not work exactly right, it may 168 00:09:02.279 --> 00:09:05.200 not be exactly what you had pictured in your head 169 00:09:05.200 --> 00:09:07.759 to begin with, but you can actually play with it. 170 00:09:07.840 --> 00:09:10.000 You can you can touch it, use it, and go 171 00:09:10.080 --> 00:09:12.960 back to your requirements and say, oh, actually this this 172 00:09:13.080 --> 00:09:16.720 part wasn't exactly right, or oh we forgot this, let's 173 00:09:16.799 --> 00:09:20.919 change these a little bit and regenerate. And so that's 174 00:09:20.960 --> 00:09:26.279 been really powerful for getting into kind of prototype driven development, 175 00:09:26.399 --> 00:09:28.799 and I've found that that ends up with a better 176 00:09:28.840 --> 00:09:31.000 product at the end of the day because you are 177 00:09:31.200 --> 00:09:35.200 able to really really use the product very early on 178 00:09:35.279 --> 00:09:37.519 while you're still forming the requirements. 179 00:09:37.600 --> 00:09:43.799 Really yeah, I'm kind of imagining how this flows. I'm 180 00:09:43.799 --> 00:09:46.759 a little curious once it has the spec if it's 181 00:09:46.759 --> 00:09:49.480 something that's a little more complicated. One of the things 182 00:09:49.480 --> 00:09:52.759 that I've experienced more with Cursor is that it will 183 00:09:52.799 --> 00:09:55.879 go and it will do the whole job right, and 184 00:09:55.919 --> 00:09:59.320 so it gets done and I'm like, I have a 185 00:09:59.360 --> 00:09:59.879 lot of stuff. 186 00:10:00.320 --> 00:10:02.840 Yeah, yeah, yeah. 187 00:10:02.360 --> 00:10:05.279 Does it do that or does it kind of so 188 00:10:06.000 --> 00:10:07.879 march through a piece at a time there too. 189 00:10:08.279 --> 00:10:13.480 We call that supervised versus autopilot mode. So sometimes you know, 190 00:10:13.679 --> 00:10:18.279 you just want to get something done right. You just 191 00:10:18.399 --> 00:10:21.120 want a really simple tool and it doesn't need to 192 00:10:21.159 --> 00:10:24.000 be exactly right. And I find that autopilot mode is 193 00:10:24.000 --> 00:10:26.440 going to be the fastest way. That means exactly what 194 00:10:26.480 --> 00:10:28.879 you just said. It's just going to plow through the work. 195 00:10:29.120 --> 00:10:33.919 It's going to completely build the entire thing. Supervised mode 196 00:10:34.000 --> 00:10:37.039 is where you want to give feedback one step at 197 00:10:37.080 --> 00:10:41.159 a time. So inspec mode, what you can do is 198 00:10:42.639 --> 00:10:44.879 it generates a task liss similar to what you were 199 00:10:44.919 --> 00:10:47.320 saying that the AI says, here's what I'm going to do, 200 00:10:48.039 --> 00:10:52.519 but it will actually stop for your feedback after each 201 00:10:52.600 --> 00:10:57.240 individual task, and so you can say, whoa, whoa, you 202 00:10:57.679 --> 00:11:01.840 are starting off on the wrong foot here, let's rewind 203 00:11:01.919 --> 00:11:04.919 and get this right. And I find that that does 204 00:11:05.440 --> 00:11:08.879 end up with more of the code that I already 205 00:11:08.919 --> 00:11:12.200 wanted because I'm steering it along the way. In between 206 00:11:12.200 --> 00:11:17.120 these little individual tests. I will say that I've noticed 207 00:11:19.679 --> 00:11:23.000 AI models really like to do a lot of work. 208 00:11:24.000 --> 00:11:26.559 And so what I find is when they when they 209 00:11:26.879 --> 00:11:32.159 generate these spec files, it's following all the software engineering 210 00:11:32.240 --> 00:11:35.919 best practices. It's got high unit test coverage, it's got 211 00:11:35.960 --> 00:11:39.399 integration tests. It might sometimes say I'm going to try 212 00:11:39.399 --> 00:11:44.639 to deploy this before before deciding I'm done. And that's 213 00:11:44.679 --> 00:11:48.480 really great if you are building a production app. Sometimes 214 00:11:48.480 --> 00:11:50.960 I'm just building a little side project or I'm building 215 00:11:51.080 --> 00:11:54.000 a little utility that I just kind of need to 216 00:11:54.039 --> 00:11:57.759 get done and I don't need integration tests. So being 217 00:11:57.799 --> 00:12:01.080 able to review that list of tasks for it goes 218 00:12:01.120 --> 00:12:03.360 off and does a lot more work than I wanted 219 00:12:03.360 --> 00:12:04.840 it to do is also really helpful. 220 00:12:07.279 --> 00:12:11.679 Nice I feel like it's like an eager intern. It 221 00:12:11.799 --> 00:12:13.720 just wants, it wants to help you so much and 222 00:12:13.759 --> 00:12:17.039 do everything for you, and you can definitely be like, no, 223 00:12:17.279 --> 00:12:20.240 let's supervise this one at a time. YEA, let's list 224 00:12:20.279 --> 00:12:23.279 delete things we don't need. We don't need three thousand 225 00:12:23.639 --> 00:12:24.519 integration tests. 226 00:12:25.200 --> 00:12:28.799 Definitely. I love the eager intern analogy. You sent me 227 00:12:28.840 --> 00:12:32.759 for coffee and I came back with doughnuts and sandwiches. 228 00:12:32.200 --> 00:12:35.399 And the whole coffee machine did. 229 00:12:35.279 --> 00:12:41.799 Your dry clean. So so do you use it now 230 00:12:41.879 --> 00:12:43.000 or go ahead? 231 00:12:43.720 --> 00:12:45.519 I was going to say so in reading through the 232 00:12:45.559 --> 00:12:47.879 instruction chareing what you were just talking about. It sounds 233 00:12:47.919 --> 00:12:50.679 like there's a lot of it's like dock writing, almost 234 00:12:50.679 --> 00:12:52.879 like you're planning and figuring out I'm going to do this. 235 00:12:53.480 --> 00:12:55.840 How much of it? How much does it actually write 236 00:12:55.840 --> 00:12:59.440 code versus just doing the planning? Is it all of 237 00:12:59.440 --> 00:13:02.799 the above? I mean could say, for instance, if you 238 00:13:02.879 --> 00:13:06.279 are a design challenge back in engineer like me or 239 00:13:06.320 --> 00:13:09.600 a code writer like me who has no design chops whatsoever, 240 00:13:10.440 --> 00:13:12.440 you know, and I actually have a project I'm working 241 00:13:12.480 --> 00:13:14.559 on that I really would love to do, this can 242 00:13:15.200 --> 00:13:17.320 do something like come up with at least a somewhat 243 00:13:17.320 --> 00:13:20.960 workable UI design based on what you're doing, you know, 244 00:13:20.960 --> 00:13:23.639 in terms of page layouts and elements, and then you 245 00:13:23.919 --> 00:13:27.919 naw bar structured tables formed that kind of stuff as well. 246 00:13:28.399 --> 00:13:32.720 Yeah, it can both write the specs and write the 247 00:13:32.840 --> 00:13:36.960 actual code for you. And the technical design step is 248 00:13:37.000 --> 00:13:40.000 really helpful for some of the UX as well, and 249 00:13:40.080 --> 00:13:42.799 kind of the code architecture, so you can say, what 250 00:13:42.919 --> 00:13:45.639 front end framework do you want to use, what kind 251 00:13:45.639 --> 00:13:48.279 of component library do you want to use? You know, 252 00:13:48.360 --> 00:13:52.559 compared to I mean when I first started out in 253 00:13:52.600 --> 00:13:58.200 the industry, I was literally writing waterfall requirements documents. What 254 00:13:58.240 --> 00:14:01.279 I like is that it's not so much about you 255 00:14:01.600 --> 00:14:06.919 writing these documents as they are kind of artifacts for 256 00:14:07.320 --> 00:14:10.559 a shared understanding between you and the and the AI, 257 00:14:11.320 --> 00:14:13.440 and so you don't have to actually write any of it. 258 00:14:13.519 --> 00:14:16.720 You can keep prompting that, you can keep prompting in 259 00:14:16.799 --> 00:14:20.000 KIRO until you have the requirements that you want, or 260 00:14:20.080 --> 00:14:22.120 you can you know, kind of edit them yourself where 261 00:14:22.240 --> 00:14:24.919 need be. But it's really helpful and really the whole 262 00:14:25.120 --> 00:14:29.480 life cycle of kind of ideation and then design and 263 00:14:29.519 --> 00:14:32.320 then actually writing the code. 264 00:14:33.399 --> 00:14:35.799 Yeah, I'll just jump in. I assumed that I guess 265 00:14:35.840 --> 00:14:38.000 I've been working from the assumption that it's as capable 266 00:14:38.039 --> 00:14:40.000 as any of the other systems out there, right, So, 267 00:14:40.279 --> 00:14:43.200 like I said, cursor and copilots seem to be able 268 00:14:43.240 --> 00:14:45.919 to do the UI stuff just fine, because I have 269 00:14:45.960 --> 00:14:49.480 the same problem, right, I work almost exclusively on the 270 00:14:49.519 --> 00:14:52.159 back end. In fact, for my full time job, I'm 271 00:14:52.240 --> 00:14:55.200 basically building APIs and then somebody else is doing all 272 00:14:55.240 --> 00:14:58.840 of the front end work. And so when I'm working 273 00:14:58.919 --> 00:15:01.679 on my own projects, yeah, I tend to let it 274 00:15:01.799 --> 00:15:05.960 do the UI UX stuff and it does fine. I mean, 275 00:15:06.039 --> 00:15:10.440 it's not always beautiful or whatever. But the thing is 276 00:15:10.440 --> 00:15:13.399 is I can more easily than tell it I want 277 00:15:13.440 --> 00:15:17.000 this over here, or hey, this isn't working quite the 278 00:15:17.000 --> 00:15:19.840 way I expect. And a lot of times again back 279 00:15:19.879 --> 00:15:21.519 to the point of you have to see it before 280 00:15:21.600 --> 00:15:24.279 you know what you want. Like I'll go in and 281 00:15:24.320 --> 00:15:27.559 I'll just do something instinctively and I didn't tell it 282 00:15:27.600 --> 00:15:30.120 I needed that, But then I'm like, oh, I need 283 00:15:30.159 --> 00:15:31.759 it so that when I click here, it does this 284 00:15:32.840 --> 00:15:35.200 and so. And it's usually pretty good about that. 285 00:15:36.360 --> 00:15:38.240 And you can also give it. You can give it 286 00:15:38.320 --> 00:15:41.360 kind of a Napkin diagram of what you want the 287 00:15:41.879 --> 00:15:44.559 UI to look like, and it will actually translate that 288 00:15:44.679 --> 00:15:49.600 into requirements or design statements. So that's pretty cool as well. 289 00:15:51.840 --> 00:15:54.840 So one thing that I'm wondering having used. Like I said, 290 00:15:54.840 --> 00:15:57.600 I've used Cursor and I've used Copilot. I'm working on 291 00:15:57.679 --> 00:15:59.919 figuring out how to get Copilot to work with Emacs. 292 00:16:00.080 --> 00:16:03.960 So I like that a little more than Visual Studio code. 293 00:16:04.360 --> 00:16:06.679 But Cursor's based on DS code. Is that is that 294 00:16:06.720 --> 00:16:08.919 what you did with kiro or is kiro something else? 295 00:16:09.200 --> 00:16:13.200 It's based on code OSS similar to the Yeah, so 296 00:16:13.240 --> 00:16:16.360 it feels very familiar. But we were also able to, 297 00:16:16.759 --> 00:16:21.159 for example, with this this speck layout. One of our 298 00:16:21.639 --> 00:16:26.159 hopes is to a little bit take the focus of 299 00:16:26.360 --> 00:16:30.360 the environment away from just the code. So one of 300 00:16:30.360 --> 00:16:34.320 the things that I noticed as we were working with 301 00:16:34.360 --> 00:16:38.000 code OSM is it's interesting the way that the colors 302 00:16:38.039 --> 00:16:42.080 are where code is very much the focus. So in 303 00:16:42.120 --> 00:16:45.639 that I use a light a light mode, and all 304 00:16:45.799 --> 00:16:50.200 everything in the UI is gray except for the code, 305 00:16:50.240 --> 00:16:54.000 which is in white white background. And so we were 306 00:16:54.039 --> 00:16:57.559 we were looking at this and thinking, you know, actually, 307 00:16:57.639 --> 00:17:00.639 you don't want necessarily the code to be the focus 308 00:17:00.679 --> 00:17:02.519 in your ID, and I think we're not We're not 309 00:17:02.600 --> 00:17:05.039 quite there yet. I think our first step is with 310 00:17:05.599 --> 00:17:09.319 this kind of dedicated UX around spec based development, where 311 00:17:09.319 --> 00:17:13.720 the focus is on the spec the design and the tasks. 312 00:17:13.839 --> 00:17:17.000 But I think my hope at least is that the 313 00:17:17.079 --> 00:17:20.960 ide becomes a place it is not just about writing code. 314 00:17:21.240 --> 00:17:26.480 It becomes a place that is more tailored around, you know, 315 00:17:26.960 --> 00:17:29.960 defining what you want, and then the code is is 316 00:17:30.039 --> 00:17:31.480 kind of an artifact that comes from that. 317 00:17:34.960 --> 00:17:42.920 MM I'm trying to imagine to myself what that looks like. Also, 318 00:17:43.079 --> 00:17:45.160 you're the only person I know that uses a light mode. 319 00:17:47.000 --> 00:17:50.039 I just don't like staring at black all day. 320 00:17:52.720 --> 00:17:55.400 So can we maybe go at a high level? I 321 00:17:55.440 --> 00:17:57.359 don't know how detailed do you want to get. Granted, 322 00:17:57.400 --> 00:18:00.480 everybody can't really see the ID since we're not screen 323 00:18:00.519 --> 00:18:04.480 sharing with you can just sort of walk some steps through. 324 00:18:04.920 --> 00:18:06.160 Let's say you're starting a. 325 00:18:06.119 --> 00:18:08.680 Project completely from scratch and you want to build some 326 00:18:09.039 --> 00:18:11.680 you know, maybe something more complex than a two do app. 327 00:18:12.359 --> 00:18:14.920 You know, who knows what are the steps? Because now, 328 00:18:14.960 --> 00:18:16.599 for instance, if I come, I'm looking at the ID 329 00:18:16.839 --> 00:18:19.799 right now, and it looks like you have two different 330 00:18:19.799 --> 00:18:22.240 ways possible ways to start, it gives you the option 331 00:18:22.319 --> 00:18:25.880 to be the vibe respect right, So one is chatting 332 00:18:25.880 --> 00:18:28.599 and one is planning first, So it looks like those 333 00:18:28.640 --> 00:18:31.200 are two different ways you can start. 334 00:18:31.200 --> 00:18:32.359 Am I reading that correctly? 335 00:18:32.400 --> 00:18:35.119 That's exactly right? Yeah. 336 00:18:35.519 --> 00:18:37.680 Okay, So let's say I want to start with the 337 00:18:37.759 --> 00:18:41.319 vibe mode. So what am I doing or how what 338 00:18:41.359 --> 00:18:42.400 are the steps I'm taking. 339 00:18:43.279 --> 00:18:46.079 That's vibe coding, as I think we're all kind of 340 00:18:46.119 --> 00:18:50.519 familiar with. Now you're you're prompting the AI to generate code, 341 00:18:50.599 --> 00:18:52.920 it's going to you know, if you do have any example, 342 00:18:53.359 --> 00:18:55.920 any existing code in there, it's going to take a 343 00:18:55.960 --> 00:18:58.599 look at that and bring it into the context. But 344 00:18:58.680 --> 00:19:01.039 if not, it's just going to start writing code for you. 345 00:19:02.240 --> 00:19:06.559 Right, So there's this chat panel down here, so this 346 00:19:06.640 --> 00:19:08.640 is where you start saying, hey, yeah, this is my code. 347 00:19:08.720 --> 00:19:11.920 I want to add this feature that's right, and make 348 00:19:12.000 --> 00:19:13.880 it look like this. This is my stack, you know, 349 00:19:13.920 --> 00:19:16.400 and I'm using you with you know, a level in 350 00:19:16.440 --> 00:19:20.039 the back end with nursia, jas and tailwind, et cetera. 351 00:19:20.160 --> 00:19:23.839 That's my preferred stack, yep. And then it's going to 352 00:19:23.920 --> 00:19:26.720 start generating what is it going to start generating documents 353 00:19:26.759 --> 00:19:29.279 there that with the output, you know. 354 00:19:29.599 --> 00:19:31.359 Yeah, so in vibe mode, it's just going to start 355 00:19:31.359 --> 00:19:34.839 writing code. It's that that very familiar experience I was 356 00:19:34.880 --> 00:19:37.160 seeing before, where it's just going to jump right into 357 00:19:37.160 --> 00:19:40.519 writing the code and then kind of iterate from there 358 00:19:40.920 --> 00:19:45.359 of trying to get the AI to update the code 359 00:19:45.440 --> 00:19:48.000 into what it is you want it to build. And 360 00:19:48.039 --> 00:19:50.400 so I find that that's very useful. I find that 361 00:19:50.480 --> 00:19:53.839 the so the Cure team built uses kiro to build kiro, 362 00:19:54.799 --> 00:19:59.319 and I find kind of across the team that we 363 00:19:59.440 --> 00:20:01.960 tend to jump into vibe mode if we know exactly 364 00:20:02.000 --> 00:20:04.880 what we want. You know, we know the code that 365 00:20:04.920 --> 00:20:07.400 needs to be written. We're just kind of steering the 366 00:20:07.440 --> 00:20:12.160 AI to generate that code. And it's a pretty simple, 367 00:20:12.599 --> 00:20:16.720 simple problem. But if we don't know what we want, 368 00:20:18.240 --> 00:20:20.440 we don't know exactly what code needs to be written, 369 00:20:20.680 --> 00:20:24.200 will jump into spec mode or if it's just a 370 00:20:24.279 --> 00:20:30.720 more complex problem. And I think that that takes a 371 00:20:30.759 --> 00:20:33.720 little bit of intuition today, It kind of takes a 372 00:20:33.720 --> 00:20:37.160 little bit of playing, and it requires knowing what do 373 00:20:37.279 --> 00:20:40.039 I think the AI is going to be good at 374 00:20:40.319 --> 00:20:43.519 kind of one shot right vibe mode to get what 375 00:20:43.559 --> 00:20:45.319 you want. You want it to be pretty close to 376 00:20:45.359 --> 00:20:47.319 one shot. You don't want to have to do a 377 00:20:47.400 --> 00:20:51.240 ton of prompt iterations. But if it, you know, you 378 00:20:51.319 --> 00:20:53.720 kind of get this intuition of what is simple enough 379 00:20:54.000 --> 00:20:56.880 for me to give to vibe mode and then anything 380 00:20:56.920 --> 00:20:59.039 more complex than that that needs to be broken down 381 00:20:59.079 --> 00:21:03.759 into individuals or they don't exactly know what I need yet. 382 00:21:03.960 --> 00:21:07.079 It is a really good starting place for a spec. 383 00:21:08.440 --> 00:21:11.960 And maybe I'm jumping ahead here, but as we're using bimode, 384 00:21:12.480 --> 00:21:15.400 as Claire said, there's it reads the files, but you 385 00:21:15.440 --> 00:21:17.680 can't create steering files or kind of the same thing 386 00:21:17.720 --> 00:21:21.079 you might see and other code editors like rules files 387 00:21:21.640 --> 00:21:24.680 that you can set up so it follows certain patterns 388 00:21:24.920 --> 00:21:27.039 as you're vibe coding back and forth. And you can 389 00:21:27.079 --> 00:21:32.519 even have here generate these steering files for you to 390 00:21:32.599 --> 00:21:33.160 start off with. 391 00:21:35.599 --> 00:21:37.759 Yeah. I find that that's really the first thing to 392 00:21:37.799 --> 00:21:42.079 do if you're starting in an existing project is generate 393 00:21:42.079 --> 00:21:45.720 the steering files because it creates this really great summary 394 00:21:45.720 --> 00:21:47.759 of the code base and what it is and what 395 00:21:47.799 --> 00:21:51.920 it does and what the architecture is. That really helps 396 00:21:51.920 --> 00:21:54.680 you to save on contact space, so it's not having 397 00:21:54.680 --> 00:21:56.240 to read a ton of files every time. 398 00:21:59.559 --> 00:22:01.480 Yeah, outputs better code. 399 00:22:02.160 --> 00:22:06.799 I feel like, Yeah, so a couple of other questions 400 00:22:06.799 --> 00:22:08.319 that I have, I'm going to change the topic a 401 00:22:08.359 --> 00:22:13.759