1 00:00:00,340 --> 00:00:02,360 -Hello, folks! -I'm Michael Dyrynda. 2 00:00:02,360 --> 00:00:03,730 And I'm Jake Bennett. 3 00:00:04,580 --> 00:00:10,340 And this is episode 185 of the North Meet South web podcast. 4 00:00:10,340 --> 00:00:11,340 -Chaka-chaka! -Chaka. 5 00:00:12,520 --> 00:00:14,219 Hey, everybody, it's been a long time. 6 00:00:14,220 --> 00:00:18,660 Glad to hear from you again, Michael. Glad to see you. Um, okay, we're gonna talk 7 00:00:18,660 --> 00:00:19,1000 -about AI on this episode- -Mm-hmm. 8 00:00:20,000 --> 00:00:23,640 But before we do, can I tell you one thing that AI is not good at? 9 00:00:23,640 --> 00:00:26,760 -Tell me. -It is not good at helping you debug 10 00:00:26,760 --> 00:00:29,920 three-way switches in your house. Do you know what a three... Do you know what a 11 00:00:29,920 --> 00:00:33,660 -three-way switch is? -Like a three, like where... Yeah. Yes. 12 00:00:33,660 --> 00:00:34,020 Yes. 13 00:00:34,880 --> 00:00:35,730 Do, do you know what this is, or you have- 14 00:00:35,730 --> 00:00:38,330 Where you've got, like, multiple switches that all control the same light, whatever. 15 00:00:38,330 --> 00:00:40,160 That control the same thing. Yes. 16 00:00:40,160 --> 00:00:43,019 -Yeah. -Yes. Now, I do not consider myself an 17 00:00:43,020 --> 00:00:46,200 expert in this, but I'm also no slouch. Like, I've done this a number of times 18 00:00:46,200 --> 00:00:47,200 -before, right? -Mm-hmm. 19 00:00:47,200 --> 00:00:51,340 Got all the tools, got all the stuff. I've done this. But the thing that I do not 20 00:00:51,340 --> 00:00:54,360 have, and I'm not even, like, doing this for the first time, like, I, I, I put in 21 00:00:54,360 --> 00:00:58,840 the switches that previously controlled this, so I know which things went where. 22 00:00:58,840 --> 00:01:05,660 I, I know. However, the stupid switches that I got don't have the common 23 00:01:05,660 --> 00:01:12,660 screw labeled, so it's, like, kind of a guess as to what one it is, and so it's 24 00:01:12,660 --> 00:01:17,400 been extremely frustrating. I've spent probably 40 minutes on it so far tonight, 25 00:01:17,400 --> 00:01:20,680 running back and forth from the switch panel to this, so I'm just- 26 00:01:21,480 --> 00:01:23,830 -No good. -Ah. But let me tell you- 27 00:01:23,830 --> 00:01:26,220 Electrical... So this is where, this is where you and I differ. See, electrical 28 00:01:26,220 --> 00:01:27,800 -work, I- -You, you've got a sparkie over here. 29 00:01:27,800 --> 00:01:30,280 I get a sparkie. Yeah, yeah. I had one in here the other day. 30 00:01:30,280 --> 00:01:32,420 -Yeah. -We're, we're, we're redoing... We're, I, I 31 00:01:32,420 --> 00:01:36,350 wanna say redoing our laundry. We are doing our laundry, 'cause we've never had- 32 00:01:36,350 --> 00:01:36,610 Mm-hmm, mm-hmm 33 00:01:36,610 --> 00:01:39,980 ... any cupboards, or storage, or bench, or anything in there. It was very much 34 00:01:39,980 --> 00:01:44,840 when the house was built, it was what I, what I have called a bachelor's laundry, 35 00:01:44,840 --> 00:01:45,450 -in that it had- -Mm 36 00:01:45,450 --> 00:01:49,179 ... like, a tub. It's got a washing machine, and just, like, the cheapest, 37 00:01:49,180 --> 00:01:53,270 nastiest cupboard that, that I could afford when I, when I built the house. 38 00:01:53,270 --> 00:01:56,040 -Sounds like ours. Yep. -And, and that's it. So at the moment, 39 00:01:56,040 --> 00:02:01,440 we're just... We've just, um... I got the text message this morning saying that we- 40 00:02:01,440 --> 00:02:06,340 we'll get the, um, all of the cabinetry delivered on the 10th of February, and 41 00:02:06,340 --> 00:02:11,240 then they will be installing everything on the 11th. So we're getting, like, benches 42 00:02:11,240 --> 00:02:15,720 and, like, a new sink, and storage, and all kinds of stuff in there, which is, 43 00:02:15,720 --> 00:02:19,400 which is wonderful. But I had a sparkie here yesterday, and I said, "I want three 44 00:02:19,400 --> 00:02:22,970 power points, and you, you know, that's it." I'm not, 45 00:02:23,700 --> 00:02:27,080 -not touching. That's not for me. -Yep. Yeah, so I'm gonna, I'm gonna figure 46 00:02:27,080 --> 00:02:30,260 it out, and then I'll let you know. So I think, yeah, I, I think I've got it. I've 47 00:02:30,260 --> 00:02:33,739 got the, the volta- like, the voltage testers out, and, like- 48 00:02:33,740 --> 00:02:35,020 -Mm -... I thought I had it. I thought I had it 49 00:02:35,020 --> 00:02:35,840 figured out, and 50 00:02:36,480 --> 00:02:38,370 -chat was helping me, and- -Yeah 51 00:02:38,370 --> 00:02:41,640 ... then it just, kind of the wheels came off, and so I just, like, I'm so 52 00:02:41,640 --> 00:02:42,570 -frustrated- -Start a new chat 53 00:02:42,570 --> 00:02:44,290 -... with this stupid thing, so- -You've, you've done too... gone too far. 54 00:02:44,290 --> 00:02:45,210 -Oh. -You have to start a new chat. 55 00:02:45,210 --> 00:02:48,359 And the thing is, it, it learned from, like, something that I did the other day. 56 00:02:48,360 --> 00:02:50,020 -Mm-hmm. -Like, I also had a switched set of 57 00:02:50,020 --> 00:02:52,670 outlets, which is, like, where you have these outlets, where the top one is always 58 00:02:52,670 --> 00:02:54,940 powered, but the bottom one is dependent on a switch on the wall. 59 00:02:54,940 --> 00:02:56,440 -Right, yep. -And that was not working from the last 60 00:02:56,440 --> 00:02:58,340 time I installed this, like, three years ago. 61 00:02:58,340 --> 00:02:59,620 -Mm-hmm. -And so I was like: Help me out, figure out 62 00:02:59,620 --> 00:03:02,920 what's going on. And it turns out I had forgotten to break off one of the tabs on 63 00:03:02,920 --> 00:03:07,100 the things. Anyway, solved that problem. It worked great for that. 64 00:03:07,100 --> 00:03:09,060 -Yeah. -Three-way switches, not so much. 65 00:03:09,060 --> 00:03:11,329 -Yeah. -So anyway, any of you people out there, if 66 00:03:11,329 --> 00:03:14,130 you like three-way switches, hit me up. But we're not here to talk about three-way 67 00:03:14,130 --> 00:03:15,579 -switches. -You'll have to wait for GPT 5.3 to come 68 00:03:15,580 --> 00:03:18,500 -out. -Correct. Um, so I... Here's the first 69 00:03:18,500 --> 00:03:21,380 thing I wanna talk about. I wanna talk about what models we're using. So- 70 00:03:21,380 --> 00:03:24,280 -Mm-hmm -... um, let me just say, also, it's sort 71 00:03:24,280 --> 00:03:28,780 of like one of those, um... I, I feel like, I, I can't remember the exact 72 00:03:28,780 --> 00:03:32,250 phrase, but it's sort of like a little bit and then all at once. 73 00:03:32,960 --> 00:03:34,299 -Mm. -That's kind of what it feels like AI has 74 00:03:34,300 --> 00:03:40,080 been for me, and sounds like for you, too. Because it was like... What the heck? 75 00:03:42,400 --> 00:03:45,780 I have never heard that beep before. I've never heard that beep before. 76 00:03:45,780 --> 00:03:46,780 -It's gonna drive you mad. -I, I don't know- 77 00:03:46,780 --> 00:03:47,570 -That's, that's- -I don't know what that beep was. 78 00:03:47,570 --> 00:03:50,380 That's the house. It's telling you, "Stop playing with the electrical and get 79 00:03:50,380 --> 00:03:51,820 someone in here who knows what they're doing." 80 00:03:51,820 --> 00:03:54,520 Literally, I, I, I think I've never heard that beep before in my life. I don't know 81 00:03:54,520 --> 00:04:01,380 what that was. No idea. Um, anyway, okay, it was like I was sort of 82 00:04:01,380 --> 00:04:04,400 dabbling. I was using the chat interfaces to talk with the- 83 00:04:04,400 --> 00:04:05,780 -Mm-hmm -... the AIs, right? 84 00:04:05,780 --> 00:04:08,220 -Yeah. -All great, used it for three-way switches 85 00:04:08,220 --> 00:04:12,560 and stuff like that, right. But if I wanted to do actual code stuff, I was, 86 00:04:12,560 --> 00:04:14,500 like, sort of copying and pasting and, like- 87 00:04:14,500 --> 00:04:15,840 -Mm-hmm -... it just wasn't a great experience. It 88 00:04:15,840 --> 00:04:16,399 -was okay. -Mm-hmm. 89 00:04:16,399 --> 00:04:16,969 -It worked- -Mm-hmm 90 00:04:16,969 --> 00:04:22,600 ... but it was just not wonderful, right? And then I installed Cloud Code, and my 91 00:04:22,600 --> 00:04:27,140 mind was blown. Was that your experience as well, something like that? 92 00:04:27,140 --> 00:04:29,590 Yeah, so over the, over the holiday break- 93 00:04:29,590 --> 00:04:33,230 And I didn't install it until version 4.5, until Opus 4.5 was out. 94 00:04:33,230 --> 00:04:34,560 -Yeah, yeah. -So, like, I felt like I was just, like, 95 00:04:34,560 --> 00:04:36,950 immediately, like, all in, like, "This is amazing." 96 00:04:36,950 --> 00:04:38,200 -Yeah. -Go ahead. Sorry. 97 00:04:38,200 --> 00:04:41,960 Yeah, over the holiday break, I finally sat down, and I bought, like, a whole 98 00:04:41,960 --> 00:04:47,400 bunch of home assist... Uh, a whole, a whole bunch of, what are they called? Uh, 99 00:04:47,400 --> 00:04:49,340 -Zigbee temperature sensors. -Yeah. Oh, yeah. 100 00:04:49,340 --> 00:04:49,1000 -So we had a new- -Absolutely 101 00:04:50,000 --> 00:04:54,780 ... we had a new, new air conditioner put in, um, a few months ago, and there was 102 00:04:54,780 --> 00:05:00,920 the option to buy its, like, first-party temperature sensors. And not, not that it 103 00:05:00,920 --> 00:05:05,560 can control the temperature per room, but it uses the temperature sensors to 104 00:05:05,560 --> 00:05:06,130 -determine- -Mm-hmm 105 00:05:06,130 --> 00:05:08,380 ... whether it needs to increase or decrease airflow- 106 00:05:08,380 --> 00:05:09,130 -Sure -... and then open- 107 00:05:09,130 --> 00:05:11,219 -Yep, yep -... and close the actuators. And they 108 00:05:11,220 --> 00:05:12,420 were, like, $130 109 00:05:13,080 --> 00:05:14,370 -per temperature sensor- -Per temperature 110 00:05:14,370 --> 00:05:16,190 -... to hook into this thing. -Yeah, you like, forget that. 111 00:05:16,190 --> 00:05:17,300 And so I bought 112 00:05:18,240 --> 00:05:23,500 10 of them, I think, to go into every room, where we've got a, a dedicated zone, 113 00:05:23,500 --> 00:05:26,780 -and then I went to- -No, wait, did you pull, did you pull the 114 00:05:26,780 --> 00:05:29,760 -trigger on the 130 a piece? -No, no, no, no, I bought, so I bought- 115 00:05:29,760 --> 00:05:30,360 -Oh, okay, okay -... the Zigbee ones- 116 00:05:30,360 --> 00:05:31,640 -All right -... which were, like- 117 00:05:31,640 --> 00:05:32,860 -Got you, got you. All right -... $200- 118 00:05:32,860 --> 00:05:34,040 -Good man. Thank you -... for 10, or something like that. 119 00:05:34,040 --> 00:05:34,860 -Sure, sure, sure. -So- 120 00:05:34,860 --> 00:05:36,620 -Okay -... 'cause I had Home Assistant running, 121 00:05:36,620 --> 00:05:41,220 and I wasn't really using it for anything. So I was like: "Okay, GPT, can you help 122 00:05:41,220 --> 00:05:46,640 me to build automation in, in Home Assistant to read the sensors, to 123 00:05:46,640 --> 00:05:52,160 determine when you should open a vent or a zone, or close a zone? You don't, don't 124 00:05:52,160 --> 00:05:57,820 mess with ones where the, where the zone is turned off." Um, we do some stuff. We 125 00:05:57,820 --> 00:06:01,460 thought, through this process, that my mas- that our master bedroom was the 126 00:06:01,460 --> 00:06:04,360 hottest room in the house, 'cause it get- it's western-facing, it gets the afternoon 127 00:06:04,360 --> 00:06:08,520 sun. Turns out it is my study that is the hottest room in the house by, like, one 128 00:06:08,520 --> 00:06:10,090 -and a half degrees. -Okay. 129 00:06:10,090 --> 00:06:13,610 Because when this, when the house was built, we insulated this room, 'cause this 130 00:06:13,610 --> 00:06:19,852 used to be my drum room.... and so once it heats up, it stays heated. And so, you 131 00:06:19,852 --> 00:06:23,171 know, but I had built all of this automation in there, like, the, the, the 132 00:06:23,171 --> 00:06:27,332 master bedroom's the hot one. We need to, like, determine in the afternoon whether 133 00:06:27,332 --> 00:06:30,692 or not we need to pre-cool it to, you know, get the airflow going early- 134 00:06:30,692 --> 00:06:32,352 -Sure, yeah -... before the sun hits, because it's 135 00:06:32,352 --> 00:06:33,382 -easier to cool- -Yep, yep 136 00:06:33,382 --> 00:06:34,311 ... a cool room than it is 137 00:06:34,991 --> 00:06:37,171 -to, like, cool down a hot room. -Of course. 138 00:06:37,171 --> 00:06:41,871 Um, and so, you know, did all of this just in, in GPT, and then it was generating 139 00:06:41,871 --> 00:06:45,751 YAML, and I was copy-pasting the YAML into Home Assistant, and then doing that. 140 00:06:45,751 --> 00:06:48,592 -Yes. -And then eventually, we needed, you know, 141 00:06:48,592 --> 00:06:51,652 uh, eventually you get to a point when you have to refactor, and even when it's just 142 00:06:51,652 --> 00:06:54,692 YAML, you still have to refactor it. You know, you wanna simplify some things. I 143 00:06:54,692 --> 00:06:55,292 wanted to see 144 00:06:55,912 --> 00:06:59,972 why it was or wasn't doing stuff, and the way that it was initially set up, it was, 145 00:06:59,972 --> 00:07:03,442 like, duplicating all of this logic. So we changed it to, to basically 146 00:07:04,551 --> 00:07:08,212 give you a reason, and then whether or not it did something was based on if that 147 00:07:08,212 --> 00:07:11,642 reason was, like, no blockers kind of thing. And then in this process- 148 00:07:11,642 --> 00:07:13,322 And at this point, you're still copying and pasting or no? 149 00:07:13,322 --> 00:07:14,842 -I'm still copying and pasting, yeah. -Okay. 150 00:07:14,842 --> 00:07:16,832 And I'm, I'm trying to figure out the best way, 'cause 151 00:07:17,631 --> 00:07:19,512 it, it started to get so much YAML 152 00:07:20,251 --> 00:07:24,832 that it was like syntax highlighting this, and the tab with the chat just kept on 153 00:07:24,832 --> 00:07:28,452 crawling. But I was too scared to start a new chat, lest it lose all of the history 154 00:07:28,452 --> 00:07:29,662 -and the- -Yeah, all the context 155 00:07:29,662 --> 00:07:30,871 -... context of that first thing. -Yeah, yeah, yeah. 156 00:07:30,871 --> 00:07:34,472 So I'm like: Okay, just give me, like, snippets, give me patches, you know, 157 00:07:34,472 --> 00:07:37,871 diffs, and I'll just copy and paste them. And then it... That became difficult, so I 158 00:07:37,871 --> 00:07:40,142 thought, "Okay, this is where we're gonna, uh, bust-" 159 00:07:40,142 --> 00:07:42,251 You've gotta... Now you've gotta go. Yep, you've gotta make the- 160 00:07:42,251 --> 00:07:42,522 -This is- -You've gotta make the jump 161 00:07:42,522 --> 00:07:45,751 ... this is where we go. We're gonna put this into a, into a Git repo. 162 00:07:45,751 --> 00:07:48,150 -Yep, yep. -We're gonna bust out OpenCode, and we're 163 00:07:48,152 --> 00:07:51,171 gonna use their, their free whatever models- 164 00:07:51,171 --> 00:07:52,501 -Okay, so now real quick- -To do this 165 00:07:52,501 --> 00:07:56,332 ... real quick, so I think this is important, right? So, uh, for me, there 166 00:07:56,332 --> 00:07:59,032 was some confusion around some of these things, so I just wanna be, like, crystal 167 00:07:59,032 --> 00:07:59,332 -clear. -Mm-hmm. 168 00:07:59,332 --> 00:08:01,751 For those of you who have not made the jump yet, look, we're gonna, we're gonna 169 00:08:01,751 --> 00:08:05,032 try and do our best to define some things here. Okay, so 170 00:08:05,772 --> 00:08:10,181 we have models, um, let's call the providers. Let's talk about some providers 171 00:08:10,181 --> 00:08:16,412 real quick. So OpenAI is the creator of the GPT models, which we all know, 172 00:08:16,412 --> 00:08:19,952 ChatGPT. They, I think, were the first ones to popularise, you know, the chat 173 00:08:19,952 --> 00:08:21,192 -interface- -Yeah 174 00:08:21,192 --> 00:08:24,631 ... that you can use with their GPT models. So ChatGPT, everybody uses Chat, 175 00:08:24,631 --> 00:08:25,491 -right? -Mm-hmm. 176 00:08:25,491 --> 00:08:29,511 So that is OpenAI. There's also Anthropic- 177 00:08:29,511 --> 00:08:31,631 -Mm -... which is now pushing commercials for 178 00:08:31,631 --> 00:08:34,300 -their models, which are Claude, right? -Mm-hmm. 179 00:08:34,300 --> 00:08:39,052 And, uh, with Claude, you have a couple of different flavours of models. You have 180 00:08:39,652 --> 00:08:41,881 -Sonnet, and you have Opus. -Mm. 181 00:08:41,881 --> 00:08:47,891 And Sonnet is, uh, less... I- it's more, um, it, it's, it's a faster, uh, model. 182 00:08:47,891 --> 00:08:51,081 It, it has less parameters, but it, so it returns values faster. 183 00:08:51,081 --> 00:08:52,872 -Mm. -But Opus is more of a, like a deep thring- 184 00:08:52,872 --> 00:08:55,512 a deep think model. It'll take a little bit longer, but it's also much more 185 00:08:55,512 --> 00:08:57,112 -comprehensive, okay? So- -Yeah 186 00:08:57,112 --> 00:09:01,632 ... those are your providers and your models. Then what we're talking about is, 187 00:09:01,632 --> 00:09:04,852 we've said, like, you know, we were just pasting things into an interface in chat, 188 00:09:04,852 --> 00:09:07,550 whatever. Fine, fair enough. A lot of these providers have their own chat 189 00:09:07,552 --> 00:09:11,012 interfaces to interact with their models. However, what Michael's talking about now, 190 00:09:11,012 --> 00:09:11,732 OpenCode 191 00:09:12,352 --> 00:09:12,862 -is- -Mm 192 00:09:12,862 --> 00:09:18,712 -... what? Tell me about OpenCode. -So OpenCode is like a multiprovider 193 00:09:19,672 --> 00:09:24,791 version of... If you've used Claude, it's a, a command line interface that you go 194 00:09:24,791 --> 00:09:28,652 in, and you have access to the models, and it does things to your code on your 195 00:09:28,652 --> 00:09:34,612 computer, or you can tell it, you know, do whatever. OpenCode is similar in vein, 196 00:09:34,612 --> 00:09:39,172 um, there is a distinction. Claude is a command line interface, and it's like- 197 00:09:39,172 --> 00:09:39,422 -Mm-hmm -... just 198 00:09:40,052 --> 00:09:45,232 doing and repainting things in your terminal, whereas OpenCode is a terminal 199 00:09:45,232 --> 00:09:50,612 user interface, a TUI, where it's doing much more modern things it, in, in a way 200 00:09:50,612 --> 00:09:55,912 that it's, like, able to render pretty interfaces and scrolling and, um, colour 201 00:09:55,912 --> 00:09:58,972 support and all this other stuff, command palettes, notifications, et cetera, et 202 00:09:58,972 --> 00:10:04,652 cetera. But the, the, the main difference, Claude locks you to the Anthropic models. 203 00:10:04,652 --> 00:10:05,752 -You can only use- -Mm-hmm 204 00:10:05,752 --> 00:10:08,072 -... what is provided by Anthropic. -Right. 205 00:10:08,072 --> 00:10:11,192 -OpenCode is multimodal, multi- -And you have to have a subscription 206 00:10:11,192 --> 00:10:12,502 -... provider. -You have to have a subscription- 207 00:10:12,502 --> 00:10:13,752 -Yes, you've gotta have a subscription- -to Claude stuff 208 00:10:13,752 --> 00:10:14,771 -... to that, yeah. -Yes, to Claude. 209 00:10:14,771 --> 00:10:16,541 -So with OpenCode- -Sorry 210 00:10:16,541 --> 00:10:22,652 ... they, um, they host some models themselves that, depending on what plan 211 00:10:22,652 --> 00:10:25,852 you're on, there are some free ones, there are some part of their Zen plan. They 212 00:10:25,852 --> 00:10:30,271 just launched, a few weeks ago, their Black plan, which is, uh, you know, more 213 00:10:30,271 --> 00:10:34,271 similar to what you'd be familiar with Claude Pro, Claude Max. But they allow you 214 00:10:34,271 --> 00:10:38,852 to bring your own keys and use any provider, and they support OpenAI, they 215 00:10:38,852 --> 00:10:43,212 support, um, Anthropic to some degree. There was a bit of hubbub about that- 216 00:10:43,212 --> 00:10:43,811 -Right -... over the break because- 217 00:10:43,811 --> 00:10:45,832 Anthropic sort of shut that down a little bit, yeah. 218 00:10:45,832 --> 00:10:51,372 Yep. They support GitHub Copilot. They support, um, ZAI is another one I tried. 219 00:10:51,372 --> 00:10:55,472 There's Minimax. There's, there's a list of, I think, probably 30 or 40 different 220 00:10:55,472 --> 00:10:59,172 providers that you can bring your own keys, and they basically just act as an 221 00:10:59,172 --> 00:11:00,941 -interface, and they give you a- -So- 222 00:11:00,941 --> 00:11:06,631 ... a much nicer interface to your models. And the, the, the best bit about it is 223 00:11:06,632 --> 00:11:11,412 that you can switch between models from the same interface. So you might find 224 00:11:11,412 --> 00:11:18,252 that, um, you know, Opus is the best for reasoning and for, um, for code, but you 225 00:11:18,252 --> 00:11:22,652 might find that GPT is better for, you know, general house things. You might find 226 00:11:22,652 --> 00:11:26,412 that Minimax is better for one thing, that, you know, all these different models 227 00:11:26,412 --> 00:11:30,472 are, are suited for different purposes, and it kinda allows you to, to wield them 228 00:11:31,811 --> 00:11:35,452 -quite easily. -And you can also use it with, like, local 229 00:11:35,452 --> 00:11:37,732 -LLMs, right? Well, using Ollama. -Yeah, yeah, Ollama. 230 00:11:37,732 --> 00:11:40,271 Right. And so to be clear, we're gonna talk about this real quick. So Ollama, 231 00:11:40,271 --> 00:11:44,352 what is Ollama, right? So most of the models that we're talking about are hosted 232 00:11:44,352 --> 00:11:49,732 for you. So Claude is hosted, um, or, uh, ChatGPT or the GPT models are hosted, 233 00:11:49,732 --> 00:11:56,372 right? Um, Co- Copilot is hosted, but you can run your own large language models on 234 00:11:56,372 --> 00:11:58,112 your own hardware if you'd like to. 235 00:11:58,752 --> 00:12:00,531 -Mm-hmm. -So Ollama gives you the ability to do 236 00:12:00,531 --> 00:12:04,952 that. You download a model, and then you run it on your machine, and then you can 237 00:12:04,952 --> 00:12:09,732 use something like OpenCode to point to any of these hosted models if you'd like 238 00:12:09,732 --> 00:12:12,102 -to, or to your local model. So if you're- -Mm-hmm 239 00:12:12,102 --> 00:12:16,051 ... very concerned with or if your company is very concerned with making sure that 240 00:12:16,051 --> 00:12:20,072 everything stays local and does not ever get leaked out, that might be an option 241 00:12:20,072 --> 00:12:21,462 -for you as well. -Yeah. 242 00:12:21,462 --> 00:12:26,932 So, uh, running your own stuff on Ollama. Also, if you need more firepower than what 243 00:12:26,932 --> 00:12:28,742 your machine can give you, um, 244 00:12:29,672 --> 00:12:35,311 we actually just bought, at work, a NVIDIA DGX Spark, which is pretty freaking sick. 245 00:12:35,311 --> 00:12:35,872 Mm. 246 00:12:35,872 --> 00:12:40,612 So it's like a $3,000, $4,000 machine, which has got, like, a 128 gigabyte 247 00:12:40,612 --> 00:12:42,771 NVIDIA, uh, GPU in it, 248 00:12:43,531 --> 00:12:45,551 -and just ships with, like, Ubuntu on it. -Mm-hmm. 249 00:12:45,551 --> 00:12:50,332 So you just set it up, and then you plug it in, and away you go. You SSH into the 250 00:12:50,332 --> 00:12:55,743 box, and then you can run Ollama-... on there. You can put any of the models that 251 00:12:55,743 --> 00:12:59,534 you want on there. Uh, I think you can run all the big ones as well. And so- 252 00:12:59,534 --> 00:13:02,123 -Mm-hmm -... you can use that as a way to not 253 00:13:02,123 --> 00:13:06,584 necessarily run it on your hardware on the same machine, but on, like, a piece of 254 00:13:06,584 --> 00:13:09,843 hardware that's sitting next to your machine, right? So that's an option as 255 00:13:09,843 --> 00:13:13,404 well. So, you know, if your company is really interested in investing in the AI 256 00:13:13,404 --> 00:13:18,464 thing, but they're not-- they, like, you know, they have things with your... They 257 00:13:18,464 --> 00:13:19,993 -have agreements maybe that says- -Yeah 258 00:13:19,993 --> 00:13:22,444 ... you can't do this. So they're like, "Well, AI is off the table." It's not off 259 00:13:22,444 --> 00:13:24,814 the table. You just have to be creative, right? 260 00:13:24,814 --> 00:13:26,873 -Yeah. -So this NVIDIA DGX Spark is another option 261 00:13:26,873 --> 00:13:31,524 that you could use to allow yourselves to use AI, and you can use something like 262 00:13:31,524 --> 00:13:34,444 OpenCode to only push things to that thing. 263 00:13:34,444 --> 00:13:35,824 -Mm. -So that's an option as well- 264 00:13:35,824 --> 00:13:37,843 -Yeah -... um, which has been interesting. Sorry. 265 00:13:37,843 --> 00:13:37,884 Yeah. 266 00:13:37,884 --> 00:13:40,924 Go ahead. So you're using OpenCode. You've got all these different models and stuff. 267 00:13:40,924 --> 00:13:40,964 Yeah. 268 00:13:40,964 --> 00:13:43,003 So what's, what's the journey been like for you? What have you been building? 269 00:13:43,003 --> 00:13:43,064 So 270 00:13:43,743 --> 00:13:46,964 just be- before we move on as well, OpenCode gives you access to different 271 00:13:46,964 --> 00:13:51,694 providers and different models. Different providers can also offer the same models. 272 00:13:51,694 --> 00:13:53,554 So obviously, Anthropic owns, 273 00:13:54,223 --> 00:13:55,593 -owns Claude- -Mm-hmm 274 00:13:55,593 --> 00:13:58,763 ... and the Claude models, so it's gonna want you to use Claude. But there are 275 00:13:58,763 --> 00:14:03,843 other providers out there. For example, um, GitHub Copilot does give you access 276 00:14:03,843 --> 00:14:09,424 to, obviously, their compute, but also to the Claude models, to, you know, different 277 00:14:09,424 --> 00:14:14,144 models that way. There's Amp Code, which gives you access to, um, Opus and Sonnet 278 00:14:14,144 --> 00:14:16,624 -and things like that through there, so- -Tell me about that real quick. I don't, I 279 00:14:16,624 --> 00:14:18,233 -don't know what that is. Tell... Can you- -So- 280 00:14:18,233 --> 00:14:19,624 -explain that to me a little bit? I don't know what- 281 00:14:19,624 --> 00:14:20,164 -No -... what that means. 282 00:14:20,164 --> 00:14:22,084 No, I can't, 'cause I started using it three days ago. 283 00:14:22,084 --> 00:14:23,664 -Oh. -I'll tell you what I did use it for. Um, 284 00:14:23,664 --> 00:14:27,444 it was Aaron Francis who said, "Get Amp Code. They give you $10 free per day or 285 00:14:27,444 --> 00:14:31,204 something, and use its Oracle." And so what I did, uh, which we'll probably touch 286 00:14:31,204 --> 00:14:35,884 on in a little bit, is I built this application in, in Go. I've never written 287 00:14:35,884 --> 00:14:37,444 -a line of Go in my life. I- -Okay 288 00:14:37,444 --> 00:14:41,223 ... it's simple enough a language to read, but I used, um, 289 00:14:41,904 --> 00:14:46,324 the Minimax model to build this out, and then I opened Amp and I said, "Hey, Amp, 290 00:14:46,324 --> 00:14:47,304 this is what I've built. 291 00:14:48,363 --> 00:14:53,084 Go and look at the code and tell me, is this idiomatic Go? And use your Oracle." 292 00:14:53,084 --> 00:14:57,524 And the Oracle is Amp's whatever, it's an oracle. And it went and looked at the code 293 00:14:57,524 --> 00:15:01,424 that was built, and it gave me a, a plan effectively. "These are the things that 294 00:15:01,424 --> 00:15:06,263 you can change to make this more idiomatic Go, to make it leverage, um, some 295 00:15:06,263 --> 00:15:08,763 things." You know, over time, you're building this, and then you build this, 296 00:15:08,763 --> 00:15:15,553 and then you build this, and then even as, you know, um, chisel, chisel in stone 297 00:15:15,553 --> 00:15:19,704 developers, old, old people like us, we would build, you know, feature A, feature 298 00:15:19,704 --> 00:15:23,044 B, feature C, and then you would look at that and then go, and, "Okay, where are 299 00:15:23,044 --> 00:15:26,233 the abstractions? Where are the things that I can hoist up?" So Amp kind of did 300 00:15:26,233 --> 00:15:32,003 that, and then I took that plan and fed it back into, um, the Minimax model in 301 00:15:32,003 --> 00:15:34,694 OpenCode, to then go and implement it. So, 302 00:15:35,624 --> 00:15:39,223 uh, the, the, the crux of the thing, I think is fairl-- there's two main ways of 303 00:15:39,223 --> 00:15:42,303 doing this now, three main ways. You can vibe it, where you just, like, whatever 304 00:15:42,304 --> 00:15:48,024 the code is generated, I'm just gonna use whatever's generated. There's, um, there's 305 00:15:48,024 --> 00:15:52,363 this notion of, of roughing, where you kind of just give it a plan, you give it 306 00:15:52,363 --> 00:15:56,954 a, a list of tasks, and you just run it literally in a bash loop, and it will just 307 00:15:56,954 --> 00:16:00,483 pick the next thing off that it thinks it should work on. It'll have its acceptance 308 00:16:00,483 --> 00:16:03,824 criteria test, static analysis, whatever you want to do, and it will just keep 309 00:16:03,824 --> 00:16:07,644 running that loop until it's run out of things to do. And then there's the thing 310 00:16:07,644 --> 00:16:09,343 that I've settled on that I think most 311 00:16:10,024 --> 00:16:13,444 proper, quote-unquote, "software engineers" would do, and that is to plan 312 00:16:13,444 --> 00:16:17,584 out your feature. Get, get this document. It's literally a markdown document that 313 00:16:17,584 --> 00:16:21,243 you can put into your, uh, applica- or, you know, version control, store it in a 314 00:16:21,243 --> 00:16:25,084 computer, whatever, and tell Claude, OpenCode, whatever, "Go read this and 315 00:16:25,084 --> 00:16:26,584 implement it exactly." And so 316 00:16:27,343 --> 00:16:32,384 I would spend thirty, forty minutes coming up with a plan, looking at it, deciding 317 00:16:32,384 --> 00:16:35,284 this is, you know, these are the files I want to create, this is the folder 318 00:16:35,284 --> 00:16:37,784 structure I want, these are the things I do and don't want to do, et cetera, et 319 00:16:37,784 --> 00:16:38,124 cetera. 320 00:16:38,863 --> 00:16:42,242 So you can use this plan mode, and then you can use different models. So, you 321 00:16:42,243 --> 00:16:46,564 know, we use Opus for planning at work because it's, you know, more of a thinking 322 00:16:46,564 --> 00:16:50,904 model. It's able to find its way through things. Once you come up with the plan, 323 00:16:50,904 --> 00:16:56,103 save it, um, start a new session, switch over to build mode, use Sonnet, 'cause it, 324 00:16:56,103 --> 00:16:58,444 it's quicker. It doesn't need to think about what it's doing, it just needs to 325 00:16:58,444 --> 00:17:01,884 execute the plan. So it reads the plan, and it goes, "Yep, no worries," knock 326 00:17:01,884 --> 00:17:08,223 tasks off until it's done. So, um, yeah, it's been, it's been a very interesting 327 00:17:08,223 --> 00:17:11,862 month or so. You know, I took some time over Christmas and the New Year- 328 00:17:11,864 --> 00:17:12,134 -Double- -Three- 329 00:17:12,134 --> 00:17:15,263 ... Can I, can I interject one question or one thing real quick? 330 00:17:15,263 --> 00:17:16,743 -Go. Go. -The reason why you might want to switch 331 00:17:16,743 --> 00:17:19,913 between models is because it's more expensive to use certain models, right? 332 00:17:19,913 --> 00:17:21,543 -Yeah. -So something like a deep think model, 333 00:17:21,543 --> 00:17:22,523 like, like Opus- 334 00:17:23,324 --> 00:17:26,263 -Mm-hmm -... requires more tokens, and so you can 335 00:17:26,263 --> 00:17:29,214 exhaust those tokens much faster if you're using Opus for everything. 336 00:17:29,214 --> 00:17:29,864 -Yeah. -Whereas if you're just- 337 00:17:29,864 --> 00:17:30,964 -Yeah -... using it for the planning mode, and 338 00:17:30,964 --> 00:17:32,574 -then use Sonnet for the execution- -Yeah 339 00:17:32,574 --> 00:17:34,313 -... that's fine too. -Mm. 340 00:17:34,313 --> 00:17:37,464 Um, if you're on the max plan, like $200 plan, maybe it doesn't really matter. 341 00:17:37,464 --> 00:17:39,184 Maybe you just use Opus for everything, right? 342 00:17:39,184 --> 00:17:40,984 -Yeah. -Um, if you're a little bit more selective, 343 00:17:40,984 --> 00:17:43,704 you can save yourself some tokens that way. Um- 344 00:17:43,704 --> 00:17:46,563 -Yeah. -And, and one other thing, which is that 345 00:17:46,563 --> 00:17:50,624 the big difference between these things and sort of using the chat window feature, 346 00:17:50,624 --> 00:17:55,083 is that these have access to do things on your behalf- 347 00:17:55,083 --> 00:17:57,004 -Mm-hmm -... but also they have the entire context 348 00:17:57,004 --> 00:17:58,913 of the repository at their fingertips, right? 349 00:17:58,913 --> 00:18:00,563 -Yeah. -So instead of having to p- paste the 350 00:18:00,563 --> 00:18:04,684 context, you can just say, "Hey, over in this class here, I'm u- I'm seeing- 351 00:18:04,684 --> 00:18:05,724 -Mm-hmm -... this sort of thing. I don't really 352 00:18:05,724 --> 00:18:08,864 like how we're doing this. Instead of this, instead of doing this, could we use, 353 00:18:08,864 --> 00:18:10,793 -like, Laravel's upsert method?" -Yeah. 354 00:18:10,793 --> 00:18:13,824 "And, you know, inspect all the other places where we're using the suspend 355 00:18:13,824 --> 00:18:16,083 system, and just get rid of those. I don't need those anymore if we use this 356 00:18:16,083 --> 00:18:16,533 -upsert." -Mm-hmm. 357 00:18:16,533 --> 00:18:19,583 And it'll just say like: "Yep, let me take a look," and it'll go through everything. 358 00:18:20,404 --> 00:18:23,504 And sometimes it'll even be like: "Uh, here's the plan." When you go to execute 359 00:18:23,504 --> 00:18:28,104 it, it'll say things like, uh, "Do you mind if I execute a artisan migrate or 360 00:18:28,104 --> 00:18:31,694 like a PHP artisan make command to make this migration?" 361 00:18:31,694 --> 00:18:33,484 -Mm-hmm. -So it'll ask, like, "Yes. Yeah, you can do 362 00:18:33,484 --> 00:18:36,083 that, and don't bother asking again for this session," whatever. So 363 00:18:37,043 --> 00:18:39,324 it's, it's doing things on your behalf. That- that's... 364 00:18:39,324 --> 00:18:40,684 -Yeah -... And, and the big, you know, the 365 00:18:40,684 --> 00:18:43,714 context, all the context that it has, that's, that's what makes it just 366 00:18:44,524 --> 00:18:46,264 -super powerful. Um- -Yeah. 367 00:18:46,936 --> 00:18:51,195 ... super, super powerful. And one other thing, which we can talk about- 368 00:18:51,196 --> 00:18:53,636 -Mm-hmm. -Which is MCPs, right? So- 369 00:18:53,636 --> 00:18:56,596 -Yeah. -This was a big one that was really helpful 370 00:18:56,596 --> 00:19:00,486 for me, is on the thing that I was building, I ended up having... Previously, 371 00:19:00,486 --> 00:19:04,636 I had a designer friend that I had worked with, that had built some screenshots for 372 00:19:04,636 --> 00:19:07,006 -me in Figma for how this should look. -Mm-hmm. 373 00:19:07,006 --> 00:19:11,316 So I described the feature and basically the whole application to Claude, wrote up 374 00:19:11,316 --> 00:19:15,656 like a, like a page of, like, instructions and said, "Make a plan." Made a plan, 375 00:19:15,656 --> 00:19:19,396 executed the plan, had a great MVP, but it didn't look anything like what I wanted 376 00:19:19,396 --> 00:19:24,425 it to. So I said, "Um, let's install the Figma MCP." Installed it- 377 00:19:24,425 --> 00:19:26,756 -Mm. -Installed it, authenticated with it, 378 00:19:26,756 --> 00:19:31,086 copied the link to the particular design, and I said, "I want it to look like this." 379 00:19:31,086 --> 00:19:31,376 Mm-hmm. 380 00:19:31,376 --> 00:19:34,605 And it said, "Yep, sure, no problem. You, you good with me executing this?" "Yes." 381 00:19:34,605 --> 00:19:36,936 And it was like, "Okay," and it started downloading screenshots. It started 382 00:19:36,936 --> 00:19:37,785 -downloading logos- -Mm. 383 00:19:37,785 --> 00:19:41,016 - placing in my asset files, and then started matching screenshot or s- you 384 00:19:41,016 --> 00:19:44,736 know, colors, offered to do fonts, Tailwind. "Can I configure your Tailwind 385 00:19:44,736 --> 00:19:47,916 -config?" "Yes." Boom! Done. -Yeah. 386 00:19:47,916 --> 00:19:49,556 -Freaking amazing! -Yeah. 387 00:19:49,556 --> 00:19:51,726 -Right? That is wild. That is the- -Yeah 388 00:19:51,726 --> 00:19:54,416 ... That is the stuff that's just, like, so insane. 389 00:19:54,416 --> 00:19:58,116 One of my colleagues... So we use ClickUp at work for project management, and we use 390 00:19:58,116 --> 00:20:02,736 Figma for design. So one of, one of my, uh, engineering colleagues hooked up his 391 00:20:02,736 --> 00:20:08,666 Claude to talk to ClickUp and to talk to Figma, and all of our application is using 392 00:20:08,666 --> 00:20:10,916 the Mui framework for 393 00:20:12,256 --> 00:20:12,906 -design- -Mui? 394 00:20:12,906 --> 00:20:14,566 -- and things like that. -Mui? 395 00:20:14,566 --> 00:20:19,536 Yeah, Mui. Yep. And so basically, what he said was, "Go and look in ClickUp and get 396 00:20:19,536 --> 00:20:25,476 all of the requirements for this ticket." We then have a link to the Figma. It will 397 00:20:25,476 --> 00:20:29,986 then go into Figma, find the components it needs to build, and because- 398 00:20:29,986 --> 00:20:31,896 -Mm -... we're using Mui components in, in 399 00:20:31,896 --> 00:20:37,076 Figma, and we're using Mui components in our React front end, it was able to just 400 00:20:37,076 --> 00:20:39,136 snap everything together from the ticket 401 00:20:39,796 --> 00:20:45,416 via Figma into actual React code. And so this is, you know, where we're at now, 402 00:20:45,416 --> 00:20:51,126 getting things that are basically built. And, and, um, it's an, it's an interesting 403 00:20:51,126 --> 00:20:53,546 -time to be alive because- -It sure is 404 00:20:53,546 --> 00:20:57,315 ... most of your time now is like, the other day, I spent thirty, forty minutes 405 00:20:57,315 --> 00:21:01,656 prompting out a plan to implement something. I was gonna go for a walk to go 406 00:21:01,656 --> 00:21:05,216 and get some lunch. I'm like: All right, I've spent forty minutes planning this. 407 00:21:05,216 --> 00:21:09,536 It'll probably take ten, fifteen to get going. This damn thing was finished 408 00:21:09,536 --> 00:21:10,526 -implementing- -Right 409 00:21:10,526 --> 00:21:15,736 ... my plan before I had finished putting my shoes on. And you may not know this 410 00:21:15,736 --> 00:21:16,906 -about me, but I- -Oh 411 00:21:16,906 --> 00:21:20,996 ... do not have laces on my shoes. I've got slip-ins, usually, that I wear during 412 00:21:20,996 --> 00:21:21,986 -the day. -You wear Kiziks? 413 00:21:21,986 --> 00:21:23,656 -Around the house. -You got Kiziks? 414 00:21:23,656 --> 00:21:26,256 -Uh, Skechers. I got Skechers. -Oh, okay, okay. 415 00:21:26,256 --> 00:21:30,296 Skechers slip-ins. So I didn't even get a chance to put my slip-ins on before this 416 00:21:30,296 --> 00:21:34,576 damn thing, after forty minutes of refining a plan, had finished, finished 417 00:21:34,576 --> 00:21:39,356 implementing. So it is a, it is a wild time, but it's just funny. Over Christmas 418 00:21:39,356 --> 00:21:42,976 and, and New Year, we, we had this workshop at the end of last year at work, 419 00:21:42,976 --> 00:21:46,516 where we talked about what AI meant in our business, and how we wanted to use it, 420 00:21:46,516 --> 00:21:49,116 and what we did and didn't want to do, and what tools we could and couldn't use 421 00:21:49,116 --> 00:21:51,736 because of compli- you know, we're finance, with there's, there's compliance 422 00:21:51,736 --> 00:21:56,796 things in there. So here's a set of tools that have passed our, um, you know, vendor 423 00:21:56,796 --> 00:22:00,256 onboarding and all that kind of stuff that we're allowed to use. And so 424 00:22:01,816 --> 00:22:04,656 I'm on the computer all day at work. It gets to the end of the day, I don't wanna 425 00:22:04,656 --> 00:22:09,276 look at the computer again until the next day. So during the break, where it was 426 00:22:09,276 --> 00:22:13,176 like family stuff, kid stuff, you know, relaxing, when I had some time in the 427 00:22:13,176 --> 00:22:16,516 evenings, I was more inclined to wanna touch the computer. And I was like: Well, 428 00:22:16,516 --> 00:22:19,776 now is a good a time as any to do this. So I started with this, this home assistant 429 00:22:19,776 --> 00:22:23,736 thing, and that just blew my mind what it could do. And I thought, "Okay, what's, 430 00:22:23,736 --> 00:22:27,956 what's the next thing we can do?" So I built, like, a, a small SaaS. I did a 431 00:22:27,956 --> 00:22:33,536 Laravel new with a, with a Livewire project. I... And I said, "Okay, I have a 432 00:22:33,536 --> 00:22:37,576 hundred and forty open tabs on my phone for articles that I'm never going to read 433 00:22:37,576 --> 00:22:43,236 later. What I want to do, what I do all the time, is I listen to podcasts," right? 434 00:22:43,236 --> 00:22:44,385 -When I walk, when I run- -Mm-hmm 435 00:22:44,385 --> 00:22:45,406 -... when I'm doing stuff- -Mm 436 00:22:45,406 --> 00:22:45,836 ... around the house, 437 00:22:46,656 --> 00:22:48,306 podcasts. I say, "Okay, what we're gonna do 438 00:22:48,936 --> 00:22:53,476 is we're gonna build this little micro SaaS for myself, where I will give it a 439 00:22:53,476 --> 00:22:59,656 link. It will go and fetch the link. It will summarize it, just, you know, to see, 440 00:22:59,656 --> 00:23:03,956 okay, what is this all about? It will then generate speech." So it's not gonna 441 00:23:03,956 --> 00:23:05,336 read word for word 442 00:23:05,956 --> 00:23:09,235 what was written, but it's going... Because, you know, you- people have lists, 443 00:23:09,235 --> 00:23:12,606 people have code, people have parentheticals, all of this kind of stuff. 444 00:23:12,606 --> 00:23:17,315 So it will generate speech, and then it will generate an audio file, and it will 445 00:23:17,316 --> 00:23:20,536 put it into a podcast feed so that it lands on my phone, and then I don't even 446 00:23:20,536 --> 00:23:23,055 have to think about it. I can just listen to this article. 447 00:23:23,056 --> 00:23:26,616 -Hmm. -Um, so built that, like four hours one 448 00:23:26,616 --> 00:23:30,485 day. Start to finish, done. Right. Never would've built that myself. Something I've 449 00:23:30,485 --> 00:23:34,596 always wanted, but nothing that I'm gonna spend, you know, days to weeks building 450 00:23:34,596 --> 00:23:35,716 myself. Um, 451 00:23:37,296 --> 00:23:41,336 I built, uh, this thing for... We used it for Laravel News. I said to you, when we 452 00:23:41,336 --> 00:23:44,576 recorded the Laravel News podcast, "I don't wanna go through the last two weeks 453 00:23:44,576 --> 00:23:48,096 of articles and find the articles, and generate the markdown, and put it into 454 00:23:48,096 --> 00:23:50,706 Notion." And, uh, so I built... 455 00:23:51,416 --> 00:23:55,956 I initially, like, the, quote-unquote, "MVP," was just asking GPT, "Hey, go and 456 00:23:55,956 --> 00:23:59,696 fetch all of the articles from the last two weeks from the Laravel News website. 457 00:23:59,696 --> 00:24:02,636 Give me a list," and I just copy-pasted that into Notion. I thought, "We're onto 458 00:24:02,636 --> 00:24:03,896 -something here." -Yeah, yeah. 459 00:24:03,896 --> 00:24:08,376 So I built out this little thing that for Laravel News, I just say, "Go and find me 460 00:24:08,376 --> 00:24:10,816 all of the articles that were published between date X, date Y. 461 00:24:11,436 --> 00:24:17,916 Classify it as releases, news, tutorials, packages. Um, generate the PDF. Generate 462 00:24:17,916 --> 00:24:21,656 the show notes. Okay, we've recorded the episode. We've got the, the markers in 463 00:24:21,656 --> 00:24:26,976 the, in the video file for YouTube. Go and match up the markers with the original 464 00:24:26,976 --> 00:24:28,416 thing. 465 00:24:29,836 --> 00:24:33,476 Put, put together the show notes that we can put into Transistor, into YouTube. Go 466 00:24:33,476 --> 00:24:35,376 and generate the transcription. 467 00:24:36,396 --> 00:24:38,875 Generate me the, you know, the, the chapter markers," all of that kind of 468 00:24:38,876 --> 00:24:44,056 stuff. That's done. So now the, the only time spent on the Laravel News podcast is 469 00:24:44,056 --> 00:24:47,056 you and I recording it, and then me editing it, which is really these days, 470 00:24:47,056 --> 00:24:50,968 'cause we've been doing it for-... donkey's years, um, is just putting the 471 00:24:50,968 --> 00:24:54,768 chapter markers in there. We-- Very low touch editing. I put chapter markers in, I 472 00:24:54,768 --> 00:24:58,928 put titles into the video, and that's it. And so another thing done. Um, I built 473 00:24:58,928 --> 00:24:59,568 this, this 474 00:25:00,348 --> 00:25:02,808 thing in Go. I've never written Go before. I l-- I read it- 475 00:25:02,808 --> 00:25:04,008 -Right -... and I'm like, "That looks like it's 476 00:25:04,008 --> 00:25:09,768 doing what I want it to do." It's just... And, like, I've been hammering it at work, 477 00:25:09,768 --> 00:25:13,348 you know. Work hours, uh, work days only last five hours now, because that's, 478 00:25:13,348 --> 00:25:15,998 that's what your Claude limit is, if you can stretch it for five hours. 479 00:25:18,448 --> 00:25:20,208 -No, I keep- -Yeah, that's how long until your tokens 480 00:25:20,208 --> 00:25:21,118 -wear out, right? -Yeah. 481 00:25:21,118 --> 00:25:22,558 You get five hours until it, until it resets again. 482 00:25:22,558 --> 00:25:26,208 Yeah, we're on the API plan, and, you know, there's six of us... There's still a 483 00:25:26,208 --> 00:25:28,688 couple of guys on, on leave at the moment. So there's six of us just 484 00:25:28,688 --> 00:25:31,728 hammering this API plan and just burning through tokens. So we finally got onto the 485 00:25:31,728 --> 00:25:32,008 -team- -Yeah 486 00:25:32,008 --> 00:25:35,868 ... plan now, and it's like: Okay, you've got five hours, use it wisely, kind of 487 00:25:35,868 --> 00:25:41,428 thing. So it is, it has definitely been productive. It has... Like, I had to do 488 00:25:41,428 --> 00:25:45,467 something with Scramble yesterday, where we've got spicy enums in our 489 00:25:47,068 --> 00:25:50,588 resources, but it doesn't read them properly. And we had to try and figure 490 00:25:50,648 --> 00:25:51,288 out: Well, how do we get 491 00:25:51,928 --> 00:25:54,668 the values from a spicy enum to render in the same way that 492 00:25:55,428 --> 00:25:58,848 native enums render in those Scramble docs? And, you know, off it went. It went 493 00:25:58,848 --> 00:26:01,748 digging through the vendor code. It went figuring out, okay, I need to write this. 494 00:26:01,748 --> 00:26:05,308 I need to, you know, hook into Scramble over here. I need to register the 495 00:26:05,308 --> 00:26:10,548 extension. I, I would've spent a day on that, probably, myself, figuring out. I 496 00:26:10,548 --> 00:26:13,968 said, "I don't wanna do that." You know, all of this stuff we're just delegating, 497 00:26:13,968 --> 00:26:17,168 -um, and shipping stuff. -Yeah. 498 00:26:17,168 --> 00:26:18,958 -It's just- -It opens so many doors. 499 00:26:18,958 --> 00:26:19,808 -It's crazy. -And so, like- 500 00:26:19,808 --> 00:26:20,928 -It's crazy -... the, the, I just read this article 501 00:26:20,928 --> 00:26:22,568 -called Perfect Software, right? -Mm-hmm. 502 00:26:22,568 --> 00:26:28,318 It's this idea of, uh, software for an audience of one, right? And so the idea is 503 00:26:28,318 --> 00:26:32,708 that you don't have to buy something off the shelf anymore that is, like, cooked up 504 00:26:32,708 --> 00:26:37,688 in Silicon Valley. Instead, you can say, like this guy did, he's like: "I want to 505 00:26:37,688 --> 00:26:43,518 have... Like, I enjoy reading text that is justified, like left to right justified." 506 00:26:43,518 --> 00:26:43,518 Yeah. 507 00:26:43,518 --> 00:26:45,088 -Like, that's what they are. -Mm-hmm. Yeah, text justification. 508 00:26:45,088 --> 00:26:45,228 -And he's like- -Yeah, true. 509 00:26:45,228 --> 00:26:49,328 So, so I just basically said, "Hey, build a Chrome plugin that justifies text with 510 00:26:49,328 --> 00:26:51,408 -one click on a website." Done. -Mm-hmm. 511 00:26:51,408 --> 00:26:53,818 Built it, of course. I mean, like, of course, it could build that. 512 00:26:53,818 --> 00:26:55,178 -Yeah. -Installed the plugin, and he's done. 513 00:26:55,178 --> 00:26:57,057 Doesn't have to go find it, doesn't have to worry about, like- 514 00:26:57,057 --> 00:26:58,328 -Mm -... it's selling his data, whatever. It's 515 00:26:58,328 --> 00:27:01,988 just, it's done, it's installed, and he can use it. 516 00:27:01,988 --> 00:27:03,468 -Mm-hmm. -So cool. Like, exactly the things that 517 00:27:03,468 --> 00:27:06,628 you're talking about. It's like, I want to be able to do this, X, and all I have to 518 00:27:06,628 --> 00:27:06,858 -do- -Mm 519 00:27:06,858 --> 00:27:08,268 -... is prompt it. -Mm-hmm. 520 00:27:08,268 --> 00:27:11,807 It's so awesome. So it's like you don't have to be, like, a 10X developer anymore. 521 00:27:11,868 --> 00:27:14,208 You know, it used to be so jealous of these guys who had this time to just 522 00:27:14,208 --> 00:27:18,288 kinda... I wanna, you know, Nuno Maduro, like: "Oh, I don't like the link-sharing 523 00:27:18,288 --> 00:27:21,848 services that are out there. I'm gonna build Pyre," right? It's like, man, it 524 00:27:21,848 --> 00:27:25,168 would be so cool to just be able to have the ti-- Like, we all have the time now to 525 00:27:25,168 --> 00:27:25,668 -do that. -Yeah. 526 00:27:25,668 --> 00:27:26,188 -You don't have to- -True. 527 00:27:26,188 --> 00:27:27,217 -Like, you can be a- -Yeah, that's right 528 00:27:27,217 --> 00:27:29,298 -... 1X engineer and just an itch- -Mm 529 00:27:29,298 --> 00:27:30,488 -... to solve a problem, right? -Yeah. 530 00:27:30,488 --> 00:27:31,868 -And it's, and that's it. That's all you- -Yeah. 531 00:27:31,868 --> 00:27:34,388 And in your LLM, and you're good to go. And it's really- 532 00:27:34,388 --> 00:27:36,008 -Yeah -... it's just like, it feels like you're 533 00:27:36,008 --> 00:27:39,688 tinkering again. It feels like you're playing with code again. It's like- 534 00:27:39,688 --> 00:27:42,088 -Yeah -... um, you know, I was talking to Andy- 535 00:27:42,088 --> 00:27:43,028 -I've had some late nights. -Yeah, yeah. 536 00:27:43,028 --> 00:27:44,448 I've had some late nights for the last few weeks because- 537 00:27:44,448 --> 00:27:45,698 I was up till 2:00. I was up till 2:00 the other night doing this. 538 00:27:45,698 --> 00:27:46,128 -I can do this. Yeah. -Just- 539 00:27:46,128 --> 00:27:49,738 Because it's like, oh, I'm gonna spend three or four hours on this, but it'll be 540 00:27:49,738 --> 00:27:52,408 -done. Before it was like- -It'll be done. Something that would've 541 00:27:52,408 --> 00:27:53,368 -taken forever -... I would spend three or four hours on 542 00:27:53,368 --> 00:27:56,148 it. Yeah, I would spend three or four hours on it before, and I would be tired, 543 00:27:56,148 --> 00:27:58,948 and I would be angry at the computer, 'cause, you know, when you get tired, it 544 00:27:58,948 --> 00:28:02,728 just... it's this, you know, missing a semicolon or pro- you know, calling the 545 00:28:02,728 --> 00:28:03,178 -wrong method- -Yeah 546 00:28:03,178 --> 00:28:04,748 ... or something, you just don't even realize it- 547 00:28:04,748 --> 00:28:05,557 -Or even, like, I don't wanna look- -... when you're tired 548 00:28:05,557 --> 00:28:06,918 ... at the documentation. I just wanted to- 549 00:28:06,918 --> 00:28:08,108 -Yeah -... like, I know what I need. 550 00:28:08,108 --> 00:28:09,708 -It's done. -I know what the words are that I need to 551 00:28:09,708 --> 00:28:10,378 -do. -Yeah, yeah. 552 00:28:10,378 --> 00:28:11,628 I just don't have to go look at the documentation- 553 00:28:11,628 --> 00:28:13,208 -Yeah -... to do it manually. 554 00:28:13,208 --> 00:28:14,247 -We have this, um- -And so- 555 00:28:14,248 --> 00:28:14,818 -... iPad. -Yeah. 556 00:28:14,818 --> 00:28:18,068 I had this iPad in the kitchen, and we had this, like, little dashboard on it. And I 557 00:28:18,068 --> 00:28:22,008 used to have it so we could see the current power price, you know, to 558 00:28:22,008 --> 00:28:24,888 determine, oh, should I, should the, should I put the car on to charge? Should 559 00:28:24,888 --> 00:28:28,208 I run the air conditioner? Should I do the dishwasher, or whatever else? And then 560 00:28:28,208 --> 00:28:31,048 this was another thing. I was like: Oh, I wanna do this up a little bit. We've got a 561 00:28:31,048 --> 00:28:34,488 new, new solar system put in. We've got a new battery put in. We need to change it 562 00:28:34,488 --> 00:28:38,688 over. So I said, "Hey, you know, AI, go and, go and do that. Rebuild it, change it 563 00:28:38,688 --> 00:28:43,788 up." And so typically, we'd have just the calendar on this iPad, and I'm like: "No, 564 00:28:43,788 --> 00:28:47,828 I want to see this power dashboard." And Rhi would be like: "But I need to see 565 00:28:47,828 --> 00:28:51,908 what, what's on for tomorrow." So she-- So I, I said, "Hey, AI, go and hook into 566 00:28:51,908 --> 00:28:55,928 iCal, you know? Go and look at our iCloud calendars, put them all on there. Show us 567 00:28:55,928 --> 00:28:57,988 everything that's today. When there's nothing for today, show us what's 568 00:28:57,988 --> 00:29:02,568 tomorrow." Oh, Rhi no longer switches to the calendar. And she, she mocked me 569 00:29:02,568 --> 00:29:04,867 initially. She's like: "I can't believe you're doing this, like, this AI thing," 570 00:29:04,868 --> 00:29:08,948 whatever. Then she comes to me, and she goes: "Hey, can you put tomorrow's weather 571 00:29:08,948 --> 00:29:13,047 on there?" "Yeah, done." Thirty seconds later, weather's there. "Done. No 572 00:29:13,048 --> 00:29:17,968 problem." So, uh, she mocks me, but there's, there's just utility, and, you 573 00:29:17,968 --> 00:29:21,448 know, for me to figure out how to get the weather in there and, you know, not 574 00:29:21,448 --> 00:29:24,988 something that I couldn't do, something that I probably couldn't do in thirty 575 00:29:24,988 --> 00:29:26,698 -minutes, you know? -Right. Yeah, but now you can. 576 00:29:26,698 --> 00:29:29,588 Go and find the weather service, find that API, figure out... You know, nah, doesn't 577 00:29:29,588 --> 00:29:31,068 matter. The AI can just do it. 578 00:29:31,908 --> 00:29:35,528 Yeah, and it's actually been a little bit scary, uh, because it's been-- there's 579 00:29:35,528 --> 00:29:38,668 been a couple of things where I would talk to one of my developers, and it'd be 580 00:29:38,668 --> 00:29:43,908 like: "Hey, I need to do this thing." Be like: "Okay. Well, I'm... Hey, I'm heading 581 00:29:43,908 --> 00:29:46,388 out for a doctor's appointment. Can I work on it tomorrow morning?" "Yeah, 582 00:29:46,388 --> 00:29:51,427 sure." He's like: "Just, just write it up for me." "Cool. Well, I'm actually just 583 00:29:51,428 --> 00:29:53,608 gonna write it up for the LLM." And I did- 584 00:29:53,608 --> 00:29:55,968 -Yeah, save some time writing it out -... and I had two... Literally, I'm, I 585 00:29:55,968 --> 00:29:58,848 wrote it up, and exactly what I would have written for him, I wrote for the LLM, 586 00:29:58,848 --> 00:30:02,368 except for a little bit sloppier, and pasted it in. And it was like I had two 587 00:30:02,368 --> 00:30:05,948 different versions of, like, experiments that I had thought about the day before, 588 00:30:05,948 --> 00:30:08,628 and was able to basically get feedback from the team. Which ones do we-- Which 589 00:30:08,628 --> 00:30:09,828 -one do we prefer? -Mm-hmm. 590 00:30:09,828 --> 00:30:12,608 Okay, throw the other one away. I spent ten minutes on it. Like- 591 00:30:12,608 --> 00:30:14,218 -Yeah -... doesn't matter. It, it's so cool. 592 00:30:14,218 --> 00:30:16,047 -Yeah. -Another s- another cool strategy, which 593 00:30:16,048 --> 00:30:18,938 I've seen some people do, is they basically... So Jordan Brill, actually, on 594 00:30:18,938 --> 00:30:23,228 our team, said: "Do not do anything on the back end." Like, he instructed the 595 00:30:23,228 --> 00:30:29,248 LLM, "Do not do anything s- like, um, back-end related. I- I'm just interested 596 00:30:29,248 --> 00:30:33,228 in front-end, front-end exploration here. Here are the challenges I'm, I'm having. 597 00:30:33,228 --> 00:30:36,228 Build me a front-end sort of, like, mock-up, but don't, don't worry about 598 00:30:36,228 --> 00:30:38,018 actually persisting any of these things in databases-" 599 00:30:38,018 --> 00:30:40,588 -Yeah -... whatever." Unbelievable, what it was 600 00:30:40,588 --> 00:30:44,168 able to generate for him. And again, it's just throwaway. If you don't like it, 601 00:30:44,168 --> 00:30:44,968 -don't worry about it. -Mm. 602 00:30:44,968 --> 00:30:45,808 -Like, chuck it. -Mm. 603 00:30:45,808 --> 00:30:47,818 Doesn't matter. Like, you've spent no time on it. 604 00:30:47,818 --> 00:30:49,548 -Yeah. -Like, you've spent some, some tokens, and 605 00:30:49,548 --> 00:30:51,098 -that's it. Um- -Yeah 606 00:30:51,098 --> 00:30:55,107 ... and so it's just great for doing explorations to sort of say, like, "Do we 607 00:30:55,108 --> 00:30:58,318 like this? Do we not? Oh, that gives us some good ideas. Let's kind of move that 608 00:30:58,318 --> 00:31:02,208 direction. Uh, we don't really need... Like, if we do need that part, how would 609 00:31:02,208 --> 00:31:04,238 we structure that in the back end? Uh, we should- 610 00:31:04,238 --> 00:31:05,428 -Mm -... adjust that this way." You know, so 611 00:31:05,428 --> 00:31:10,848 like, um, it just allows you to sort of-... experiment, kinda go down the road 612 00:31:10,848 --> 00:31:14,098 a little bit. Before, I felt like I would've had to have much more 613 00:31:14,098 --> 00:31:17,078 justification to try something out, because you know, you're develop-- you're, 614 00:31:17,078 --> 00:31:19,688 you're basically committing a day to it, like, one of your developers'- 615 00:31:19,688 --> 00:31:21,818 -Mm-hmm --days to it. So you kinda would waffle 616 00:31:21,818 --> 00:31:24,158 back and forth and like, "Ah, which approach is gonna be the best? I don't 617 00:31:24,158 --> 00:31:26,808 know." Now, you literally just try both. Try them both. 618 00:31:26,808 --> 00:31:27,988 -Mm. -Let's see which one we like better. 619 00:31:27,988 --> 00:31:29,758 -Mm-hmm. -Ah, I actually like that one better. Let's 620 00:31:29,758 --> 00:31:30,708 -go with that. -And with- 621 00:31:30,708 --> 00:31:31,798 -And so- -With the, with the- 622 00:31:31,798 --> 00:31:34,418 -Really, really cool -... with, with, you know, the advent and 623 00:31:34,418 --> 00:31:39,418 the robustness now and the, the power of these, these tools, coupled with the power 624 00:31:39,418 --> 00:31:40,498 of Git worktrees, 625 00:31:41,598 --> 00:31:43,378 -you can, you can- -Ah, let's talk about Git worktrees. 626 00:31:43,378 --> 00:31:46,678 You can, you can have... You know, you said you had two different approaches, why 627 00:31:46,678 --> 00:31:47,518 -not three approaches? -Yeah. 628 00:31:47,518 --> 00:31:50,838 Why not the same two approaches with four different models, you know? 629 00:31:50,838 --> 00:31:53,898 -Yeah. -So worktrees then allow you to create a 630 00:31:53,898 --> 00:31:56,968 branch and a working environment for each of these things. So previously, you 631 00:31:56,968 --> 00:31:59,898 would've just cloned down the same repository four times. Well, now you can 632 00:31:59,898 --> 00:32:03,798 clone down the repository once with your worktrees, and have four worktrees all 633 00:32:03,798 --> 00:32:09,038 working on, like, feature A one, feature A two, feature A three, feature A four. 634 00:32:09,038 --> 00:32:11,118 -Okay, we're gonna send Opus- -Can we, can we pause for a second to talk 635 00:32:11,118 --> 00:32:12,378 -about this? -Sure. 636 00:32:12,378 --> 00:32:15,148 Like, what Git, what a Git worktree is. So, like, I think we all know, like, a Git 637 00:32:15,148 --> 00:32:17,258 repository, you pull down, right? But the, the- 638 00:32:17,258 --> 00:32:19,258 -Mm -... the limitation with a Git repository 639 00:32:19,258 --> 00:32:22,938 is that y- you can only be working one branch at a time, right? If you're on a 640 00:32:22,938 --> 00:32:25,138 branch, and you have something in your working directory, and you need to, 641 00:32:25,138 --> 00:32:28,448 like... Oh, an emergency bug comes in, you have to, like, either commit, or you have 642 00:32:28,448 --> 00:32:31,168 to Git stash, and then you have to switch back to master, pull down master- 643 00:32:31,168 --> 00:32:32,398 -Mm -... and make a new branch and do all that 644 00:32:32,398 --> 00:32:35,548 stuff, and it's like, you can only have one thing working at a time. That's 645 00:32:35,548 --> 00:32:38,858 challenging, especially if you want to have two different things going on at the 646 00:32:38,858 --> 00:32:41,288 same time on your own machine. How do you do that? 647 00:32:41,288 --> 00:32:42,938 -Yeah. -So what you said is, the solution used to 648 00:32:42,938 --> 00:32:46,458 be, you know, just clone down multiple things, and then kind of it-- you know, 649 00:32:46,458 --> 00:32:49,208 you'd Git cd into each directory and do it that way. 650 00:32:49,208 --> 00:32:51,578 -Yeah. -Worktrees are the solution to this, and 651 00:32:51,578 --> 00:32:54,048 they are baked straight into Git. This is not a- 652 00:32:54,048 --> 00:32:55,938 -Yeah -... like, extra feature. This is a Git 653 00:32:55,938 --> 00:32:59,568 -thing. So, um, go ahead. Like, a, a Git- -Yeah 654 00:32:59,568 --> 00:33:04,118 -... tree then is what? A worktree. -Yeah, a worktree is just a, basically a 655 00:33:04,118 --> 00:33:10,218 directory that points to some branch in your repository, and you can have multiple 656 00:33:10,218 --> 00:33:12,418 worktrees all pointing at the same branch 657 00:33:13,038 --> 00:33:15,578 if you really wanted to. I probably wouldn't do it at the same time, but, you 658 00:33:15,578 --> 00:33:19,258 know, four branches, four worktrees, and like I said, you could have Opus working 659 00:33:19,258 --> 00:33:23,218 on one, you could have, uh, GPT working on one, you could have Sonnet working on 660 00:33:23,218 --> 00:33:28,198 one, and you could Minimax, GLM, whatever, pick another of, of the models. "Here is 661 00:33:28,198 --> 00:33:32,258 the same plan for all four of you. Go implement." And it-- and you've got four 662 00:33:32,258 --> 00:33:35,998 agents, four models, all working on the same thing. "Look, pick which one you 663 00:33:35,998 --> 00:33:39,758 like. Pick which is, which is the one you wanna have in there," and they all do it. 664 00:33:39,758 --> 00:33:44,858 So, um, yeah, I, I'd been using them for years and years because I was often 665 00:33:44,858 --> 00:33:47,798 reviewing other people's code, looking at other things. So if I was working on 666 00:33:47,798 --> 00:33:48,218 something, 667 00:33:48,898 --> 00:33:52,407 and I would have to stop because I wanted to pull down a PR to look at it locally, 668 00:33:52,407 --> 00:33:56,078 'cause, you know, sometimes, most of the time, it's not great to just look at it in 669 00:33:56,138 --> 00:33:59,118 the, in the GitHub UI. You want to pull the code down, feel around, click around, 670 00:33:59,118 --> 00:34:03,438 see, see what it is. You'd have to, you know, do a whip commit, stash your 671 00:34:03,438 --> 00:34:08,588 changes, change branches, come back to it, you know, half a day later, forgetting, 672 00:34:08,588 --> 00:34:11,527 "Oh, and I forgot... forget about the stash. Where'd that code go? I don't know. 673 00:34:11,527 --> 00:34:14,818 It must be lost somewhere," you know. The worktree solved that problem, because you 674 00:34:14,818 --> 00:34:18,257 can just have everything and just switch directories and, um, and do all that. So 675 00:34:18,257 --> 00:34:21,518 I, I built... This is the, this little Go thing that I built, um, 676 00:34:22,578 --> 00:34:22,757 which 677 00:34:23,398 --> 00:34:27,538 doesn't-- like, it wraps over the top of Git worktrees and just gives you, like, a 678 00:34:27,538 --> 00:34:33,578 nice UI, a nice, um, interface to then create worktrees, spin up new projects, 679 00:34:33,578 --> 00:34:39,697 um, work and move between all of these different things in a simpler way. With 680 00:34:39,697 --> 00:34:44,947 the, with the, like, idea in my head that you could say, "Hey," um, tell your AI 681 00:34:44,947 --> 00:34:48,217 model, "Go and, go and create a new worktree and go and explore this plan in, 682 00:34:48,217 --> 00:34:50,757 in that directory," and off it goes and, and does its thing. So 683 00:34:51,498 --> 00:34:57,298 it was, it was... I saw, I saw OpenCode has built this functionality into, I 684 00:34:57,298 --> 00:35:01,118 think, their GUI. I don't know if it's in their command line version at the moment, 685 00:35:01,118 --> 00:35:04,378 but I saw a video. I literally finished exporting the video that, that I published 686 00:35:04,378 --> 00:35:08,558 on, on Twitter, and I saw that the OpenCode guys had, had posted a video 687 00:35:08,558 --> 00:35:11,868 saying they've got workspaces in OpenCode now. So, um, it was more- 688 00:35:11,868 --> 00:35:13,278 Worktrees in, in OpenCode, it'll do it in there. 689 00:35:13,278 --> 00:35:14,338 -Yeah. -Yeah. 690 00:35:14,338 --> 00:35:16,268 Yeah, workspaces, I think they're calling it, 'cause then they're saying- 691 00:35:16,268 --> 00:35:17,377 -Okay -... you know, your workspace may not 692 00:35:17,378 --> 00:35:21,598 necessarily be a Git worktree. It could be, like, a, a Docker environment or, you 693 00:35:21,598 --> 00:35:22,718 -know, some third- -Sure 694 00:35:22,718 --> 00:35:27,368 ... um, remote service or whatever. But it was more an exploration in, um, 695 00:35:28,358 --> 00:35:34,658 how, how do I use these tools to, to craft something where I'm not, I'm not the 696 00:35:34,658 --> 00:35:36,118 expert in the language? 697 00:35:37,398 --> 00:35:41,698 Putting myself in the shoes of, like, "I know how to develop software. How would I 698 00:35:41,698 --> 00:35:45,478 apply this to build something that I would never build myself?" Um, 699 00:35:46,298 --> 00:35:49,438 which, which was an interesting experience. I learnt, I learnt a lot. Um, 700 00:35:49,438 --> 00:35:53,058 you know, having, having one model build it and then having another model come 701 00:35:53,058 --> 00:35:56,017 through and review it and say, "You know, this is not the way to do it. I would do 702 00:35:56,018 --> 00:35:59,498 it this other way." Creating a plan and then, you know, bouncing, and then you 703 00:35:59,498 --> 00:36:02,818 learn, you know, which, which models are good for which things. I mean, you could, 704 00:36:02,818 --> 00:36:07,398 if you wanted to burn all your tokens and all of the trees in the world, use Opus 705 00:36:07,398 --> 00:36:10,118 -and just only use Opus for everything. -That's what I do. 706 00:36:10,118 --> 00:36:14,718 Great! But- ... Opus, you know, and this, and this is where the, you know, the... 707 00:36:14,718 --> 00:36:17,858 Yes, you can do that, but it's ex- it's more expensive in terms of, like- 708 00:36:17,858 --> 00:36:20,678 -Sure -... consumption of tokens. So you use the 709 00:36:20,678 --> 00:36:25,538 smart models, the reasoning, the thinking, whatever models you wanna call them, to 710 00:36:25,538 --> 00:36:28,258 come up with the plans, 'cause they will think about the edge cases. You know, 711 00:36:28,258 --> 00:36:31,277 you'd say: What have I missed here? What do you think I should explore? Is this a 712 00:36:31,278 --> 00:36:35,158 good idea? Yada, yada, yada. Um, and then once the plan is written, you get it to 713 00:36:35,158 --> 00:36:38,138 write it to a markdown file. You start a new session, and you say, "Okay, let's 714 00:36:38,138 --> 00:36:43,978 switch to Sonnet or Minimax or GLM or whatever, whatever, GPT, whatever. Um, 715 00:36:43,978 --> 00:36:48,337 here's the plan. Read it." It may have other questions. Ask me, you know, if you 716 00:36:48,338 --> 00:36:53,358 think there's anything missing, um, and then off you go. And then, you know... So 717 00:36:53,358 --> 00:36:59,138 where, let's say, an Opus token counts for one, you know, one, one token is, is, is 718 00:36:59,138 --> 00:37:04,538 $1. Well, a Sonnet token, for example, is 50 cents. So you could do twice as many 719 00:37:04,538 --> 00:37:06,767 -tokens with your Sonnet model- -Yeah, yeah 720 00:37:06,767 --> 00:37:10,598 ... than you can with the Opus model. But when it's, when it doesn't have to think 721 00:37:10,598 --> 00:37:15,389 or-... determine, you know, do I want A or B? Do I wanna create this file or that 722 00:37:15,390 --> 00:37:19,710 file? No, it's just executing the plan that you've already written. So there's no 723 00:37:19,710 --> 00:37:23,090 thinking involved. It's got fresh context. It can just burn through it. 724 00:37:23,090 --> 00:37:26,850 Takes a couple of minutes. Um, you all know this 'cause you've all been using AI 725 00:37:26,850 --> 00:37:31,890 for six, 12 months, whatever, already. I'm late to the game. I'm as, as excited as 726 00:37:31,890 --> 00:37:35,890 -you were. I'm sorry that I missed this. -Mm-hmm. Well, I mean, honestly, I feel 727 00:37:35,890 --> 00:37:38,210 -like, for me- -I feel like it's only come along in the 728 00:37:38,210 --> 00:37:41,530 last two months or so, though. Like, since Opus came out, which was Oct... I think 729 00:37:41,530 --> 00:37:46,100 October, November last year, like, that's where w- where we saw this real shift. Um, 730 00:37:46,100 --> 00:37:46,720 -and, you know- -Mm-hmm 731 00:37:46,720 --> 00:37:49,970 ... the start of this year, where we started to see a little bit of fallout 732 00:37:49,970 --> 00:37:53,190 from all of the, the changes come to a head, I think so. 733 00:37:53,190 --> 00:37:56,430 Yeah. I read an article the other day, and I'm trying to find it right now, but it 734 00:37:56,430 --> 00:38:01,420 was essentially, you know, a year ago, you know, people were saying, like: "Oh, it's 735 00:38:01,420 --> 00:38:05,530 gonna, you know, replace all these jobs of m- mid-level developers," and it was 736 00:38:05,530 --> 00:38:09,230 like, not a chance. And I read an article the other day where, like, after Opus 4.5 737 00:38:09,230 --> 00:38:12,670 was released, they're like: "Nope, that's actually true. That's actually true now. 738 00:38:12,670 --> 00:38:14,020 -It could." -Mm-hmm. Mm. 739 00:38:14,020 --> 00:38:16,250 Um, I don't think it's gonna replace jobs. I think it's just going to... People who 740 00:38:16,250 --> 00:38:19,830 don't know how to use AI are gonna be outpaced by those who do, and so there'll 741 00:38:19,830 --> 00:38:24,070 still be people at-- There will still be a human at the wheel a lot of times, right? 742 00:38:24,070 --> 00:38:26,850 Um, if you're using it responsibly and wanna maintain a code base that actually 743 00:38:26,850 --> 00:38:28,589 is gonna live for longer than a year, right? 744 00:38:28,590 --> 00:38:29,770 -Yeah, yeah. -You're gonna have to sort of be 745 00:38:29,770 --> 00:38:33,710 intelligent about how you're using it, but, um, yeah, I mean, I've already seen 746 00:38:33,710 --> 00:38:37,430 this in our own organisation. It's like, you know, one of our guys was just like 747 00:38:37,430 --> 00:38:41,070 flying. I'm like: "What is he doing?" Well, yeah, he's, he's using a lot of, uh, 748 00:38:41,070 --> 00:38:44,030 tooling, and it's not that he was being lazy. He's literally just being efficient, 749 00:38:44,030 --> 00:38:44,640 -just work- -Yeah 750 00:38:44,640 --> 00:38:47,810 ... ripping through this stuff. So shout-out, big shout-out on our team, 751 00:38:47,810 --> 00:38:52,180 honestly, Andy Hinkle is, uh, the dude who's been pushing this so hard. 752 00:38:52,180 --> 00:38:53,810 -Mm. -Like: "Hey, you guys gotta start using 753 00:38:53,810 --> 00:38:55,700 -this stuff." Um- -Yeah 754 00:38:55,700 --> 00:38:56,440 -... and so- -Yeah 755 00:38:56,440 --> 00:38:57,080 -... really, really neat. -Well, I just mean, uh- 756 00:38:57,080 --> 00:38:59,650 And he's actually doing some... I, we should have him on the show sometime. He, 757 00:38:59,650 --> 00:39:04,350 he's building, um, some of our own models, like, to, you know, like I was talking 758 00:39:04,350 --> 00:39:07,070 -about, like, on our, on code stuff- -Mm 759 00:39:07,070 --> 00:39:09,630 ... um, no big deal. Like, we're not actually hi-- you know, handling any 760 00:39:09,630 --> 00:39:13,369 private information, but when we start dealing with actual data- 761 00:39:13,370 --> 00:39:14,790 -Mm -... that's where you've gotta be careful. 762 00:39:14,790 --> 00:39:15,720 -Interesting. -And that's why we have that- 763 00:39:15,720 --> 00:39:17,530 -Yeah -... um, DGX. It's like we're, we're 764 00:39:17,530 --> 00:39:20,399 pushing stuff to that because that's internal, not going anywhere. 765 00:39:20,399 --> 00:39:22,550 -Mm-hmm. Mm. -And, uh, we're actually training-- he's 766 00:39:22,550 --> 00:39:22,890 training 767 00:39:23,830 --> 00:39:27,250 some of our own models on some of this data, which is really exciting. Super cool 768 00:39:27,250 --> 00:39:30,550 stuff. He understands it so much better than I do, but yeah, we should have him 769 00:39:30,550 --> 00:39:33,709 talk about it, uh, sometime, have him on to talk about it. And he also has been 770 00:39:33,710 --> 00:39:36,770 doing this really cool... He introduced me to this thing called the circuit breaker 771 00:39:36,770 --> 00:39:41,150 pattern today. Um, yeah, I'd have to have him on to talk about that sometime, too, 772 00:39:41,150 --> 00:39:42,410 -but- -Interesting. 773 00:39:42,410 --> 00:39:44,470 Yeah, good, good stuff. Really exciting. Okay. 774 00:39:44,470 --> 00:39:45,900 -Um, yeah- -Real... Go ahead. 775 00:39:45,900 --> 00:39:48,570 I think there, there's been a lot of doom and gloom about developers being r- I 776 00:39:48,570 --> 00:39:53,549 think... Yes, I, I mean, I've got, I've got a friend who's built, um, he built, 777 00:39:53,550 --> 00:39:57,410 like, an iOS app that he sent me a test flight link for, which is just when he 778 00:39:57,410 --> 00:40:02,570 makes coffee, he will record the grind size, the brewing time, what it tasted 779 00:40:02,570 --> 00:40:05,090 like, whether he liked it, whether he didn't like it. He'll take a photo of the 780 00:40:05,090 --> 00:40:06,550 latte art that he did on it. 781 00:40:07,390 --> 00:40:12,390 This is a UX guy, no coding experience. Just this, this is just what people can do 782 00:40:12,390 --> 00:40:16,209 now. No one-- no, no market for it. It doesn't matter. I will use it, you know? 783 00:40:16,210 --> 00:40:20,510 It's always been... Laravel wouldn't exist if Taylor wasn't building it for his own 784 00:40:20,510 --> 00:40:23,990 needs. The good things will find a market for themselves. I think we're gonna see a 785 00:40:23,990 --> 00:40:27,959 lot more stuff come out, and we have been seeing a lot more stuff come out. I think 786 00:40:28,610 --> 00:40:32,770 the doom and gloom is, is an interesting prospect. Like, Salesforce fired four 787 00:40:32,770 --> 00:40:36,680 thousand people, and then I think, uh, the CEO or whatever came out a couple of days 788 00:40:36,680 --> 00:40:39,450 ago and was like: "Whoops, made a mistake. We actually need all those people 789 00:40:39,450 --> 00:40:45,230 back. AI is not quite what we thought it was." So, um, I think there's still going 790 00:40:45,230 --> 00:40:50,110 to be space for people like us because people still need to understand. Um, yes, 791 00:40:50,110 --> 00:40:52,910 you can ask the LLM to just go and fix the problem, but eventually, 792 00:40:53,650 --> 00:40:57,530 any, any of us that have used it, that have just, like, built without thought, 793 00:40:58,150 --> 00:41:02,070 um, you'll find yourself in, in a hole. And so there's still gonna be a market for 794 00:41:02,070 --> 00:41:09,050 us to, to kind of do that kind of work. But as developers, I think we 795 00:41:09,050 --> 00:41:14,510 always found ways to optimise our job. You know, for years I've been given, you 796 00:41:14,510 --> 00:41:18,370 know, "Do this task, do this data entry. We need to ingest this," you know, 797 00:41:18,370 --> 00:41:22,210 whatever. We write scripts to do that. Like, this is the next evolution of 798 00:41:22,210 --> 00:41:25,490 writing a script to do that, so that we can focus on, less on the, the 799 00:41:25,490 --> 00:41:29,440 boilerplate, the boring stuff, all these conversations that have been had- 800 00:41:29,440 --> 00:41:31,110 -Mm -... over and over again, you know? And 801 00:41:31,170 --> 00:41:35,910 again, I know I'm late to this, and I know I've, I was a naysayer, but I've caught 802 00:41:35,910 --> 00:41:39,070 up, all right? We're here. We're ready to go. 803 00:41:39,070 --> 00:41:41,770 -Yep, and it's great. It's exciting. -This will not turn into the AI podcast, 804 00:41:41,770 --> 00:41:42,040 -though- -I love it 805 00:41:42,040 --> 00:41:46,430 ... but it's just, I think, given, given that we took the plunge, both of us, kind 806 00:41:46,430 --> 00:41:48,790 -of at the same time, separately- -Yeah 807 00:41:48,790 --> 00:41:54,090 ... but together, um, we took the plunge. We, "Oh, okay, we missed-- we, we, we were 808 00:41:54,090 --> 00:41:58,550 -late," but I think we're there now, so. -Yeah. I mean, I've always, I've always 809 00:41:58,550 --> 00:42:00,680 been a late adopter. Honestly, I, I don't feel- 810 00:42:00,680 --> 00:42:02,870 -Mm -... like it's, it's killed me. I, I feel 811 00:42:02,870 --> 00:42:08,370 like, um, I, I can't say I've always been a late adopter, but it's like specifically 812 00:42:08,370 --> 00:42:11,810 with these things, it's just we've got a lot of stuff that we need to actually get 813 00:42:11,810 --> 00:42:15,870 done, and so if there's something that's a flash in the pan, new thing, like, I'm 814 00:42:15,870 --> 00:42:19,010 probably gonna wait for a bit to kinda jump on board with it, right? 815 00:42:19,010 --> 00:42:20,090 -Mm-hmm. -Um, and 816 00:42:21,310 --> 00:42:25,830 i- in this case, I don't really think I missed much. I think we got on about at 817 00:42:25,830 --> 00:42:29,060 the time where it's like, it's reaching really good productivity levels. 818 00:42:29,060 --> 00:42:30,750 -Exactly. -And it's like, I, I missed, I missed sort 819 00:42:30,750 --> 00:42:34,490 of the experimentation phase of a lot of things and let some other folks figure 820 00:42:34,490 --> 00:42:38,310 that out, and now it's like, yep, I can jump right in and be very productive. All 821 00:42:38,310 --> 00:42:40,169 -the tooling is already built out. -Yeah. 822 00:42:40,170 --> 00:42:44,390 All the MCP stuff for Laravel is built out, all done, and all the bugs and kinks 823 00:42:44,390 --> 00:42:48,729 are worked out, like, it's, it's a really great time to jump in. So if you have not 824 00:42:48,730 --> 00:42:55,130 done so yet, try it out. Um, suggestions for if you're just getting started, um, I 825 00:42:55,130 --> 00:43:00,170 think that if you want to just dip your toe in, I think OpenCode is a really good 826 00:43:00,170 --> 00:43:04,310 one to just get started with. Just try it. You can use some of their hosted models 827 00:43:04,310 --> 00:43:09,530 or, um, you know, like you said, AMP has some free stuff. So find one that's free 828 00:43:09,530 --> 00:43:13,870 and just try it out and see if you like it. If you do, I think that 829 00:43:14,790 --> 00:43:18,630 you should try Opus. I think that you should get like a... It's, it's a twenty 830 00:43:18,630 --> 00:43:20,729 -dollar a month plan. So twenty bucks. -Yeah. 831 00:43:20,730 --> 00:43:23,510 Try it out. You're gonna get rid of-- You're gonna eat all your tokens right 832 00:43:23,510 --> 00:43:25,590 away at first, 'cause you're not gonna know what you're doing, and you're just 833 00:43:25,590 --> 00:43:26,970 -gonna exhaust them. -Mm-hmm. 834 00:43:26,970 --> 00:43:30,010 You'll figure out how to switch between, um, between models, between Sonnet and 835 00:43:30,010 --> 00:43:35,014 Opus and all that stuff, using Cloud Code on your desktop. Um-... but you're gonna 836 00:43:35,014 --> 00:43:38,564 be able to create some pretty crazy stuff pretty quick, and you're gonna realize the 837 00:43:38,564 --> 00:43:43,573 power of it very fast. So if it's a $20 investment, try it. It's really, really 838 00:43:43,574 --> 00:43:46,934 worth it, and it's not bad. It's not bad to install it and get it going. It's, it's 839 00:43:46,934 --> 00:43:48,604 -pretty simple. -Yeah. 840 00:43:48,604 --> 00:43:52,214 Um, so yeah, the, the learning curve is just there isn't one anymore. It's very 841 00:43:52,214 --> 00:43:56,214 simple. It's, it's like if you can install anything from Homebrew, you can install 842 00:43:56,214 --> 00:43:58,033 -Claude Code and get going. -Mm-hmm. 843 00:43:58,034 --> 00:43:58,774 Um, so 844 00:44:00,054 --> 00:44:06,494 well worth the time, folks. Jump in. It's good stuff. All right. Um, I was going to 845 00:44:06,494 --> 00:44:09,674 say one more thing about sort of what, what I've built with mine. So I had a 846 00:44:09,674 --> 00:44:13,714 friend who was asking me to help with a project. 847 00:44:13,774 --> 00:44:16,364 I had committed to it. This is the one I was talking about, and I was- 848 00:44:16,364 --> 00:44:17,934 -Mm -... supposed to meet with him, like, the 849 00:44:17,934 --> 00:44:21,954 next day to sort of show this project off. I had all the designs, I had all the 850 00:44:21,954 --> 00:44:25,924 concepts, but I hadn't even started building it at all. Like, that's why I was 851 00:44:25,924 --> 00:44:30,914 like, okay, started at, like, 8:00 o'clock, and I had it done by, like, 852 00:44:30,914 --> 00:44:36,414 midnight. Like, completely done, front end, back end, saving to the database, 853 00:44:36,414 --> 00:44:41,874 doing local storage stuff on the front end, validation, view. Um, I don't even 854 00:44:41,874 --> 00:44:43,344 -write view stuff, but it was all- -Yeah 855 00:44:43,344 --> 00:44:45,764 ... like, single, you know, single view component. 856 00:44:45,764 --> 00:44:47,954 -Yeah. -So it was really snappy. Um, 857 00:44:49,274 --> 00:44:54,434 unbelievable! Unbelievable. And that was my first time using it in a three-hour 858 00:44:54,434 --> 00:44:55,494 span, and I was like, 859 00:44:56,234 --> 00:44:58,214 -"Sold, done, never looking back." -Yeah. 860 00:44:58,214 --> 00:45:01,554 And I immediately bought... Like, I was on the $100 plan by the end of the night. I 861 00:45:01,554 --> 00:45:02,214 -was like, "I'm- -Mm 862 00:45:02,274 --> 00:45:05,224 -... I'm in." So crazy. -Yeah. 863 00:45:05,224 --> 00:45:07,714 -So crazy. So- -The interesting thing is, it's, it's 864 00:45:07,714 --> 00:45:08,404 possible, and 865 00:45:09,214 --> 00:45:13,124 I, I wonder how far the pendulum will swing to, like, everyone can build 866 00:45:13,124 --> 00:45:15,913 -whatever they think of. -Mm-hmm. 867 00:45:15,914 --> 00:45:20,514 Um, 'cause, like, lots of people build stuff now and find no users. So yes, th- 868 00:45:20,514 --> 00:45:23,214 this is gonna enable people to build anything that they dream of. That doesn't 869 00:45:23,214 --> 00:45:26,974 mean that it solves all the other problems: marketing, having, finding a 870 00:45:26,974 --> 00:45:30,494 customer base, you know, people to use your thing. Maybe it doesn't matter. Maybe 871 00:45:30,494 --> 00:45:31,044 just building 872 00:45:31,714 --> 00:45:36,194 20, 30 things for yourself to, to solve, you know, miscellaneous little problems. 873 00:45:36,194 --> 00:45:39,734 But, uh, it'd be interesting to watch the pendulum swing. 2026- 874 00:45:39,734 --> 00:45:41,174 -Yeah -... will be an interesting year, I think. 875 00:45:42,954 --> 00:45:48,554 Absolutely. All right, folks, well, hey, let's wrap it up there. Uh, episode 185, 876 00:45:48,554 --> 00:45:51,854 -is that what you said? -Hmm. Yep. 877 00:45:52,094 --> 00:45:56,234 Yeah, 185. Find show notes for this episode at northmeetsouth.audio/185. Hit 878 00:45:56,234 --> 00:46:01,404 us up on Twitter, on X, @jacobbennett and @michaeldrenda, @northsouthaudio? 879 00:46:01,404 --> 00:46:04,254 -Mm-hmm. -Think so. Um, that's it, everybody. Rate 880 00:46:04,254 --> 00:46:07,754 us up five, five stars in your podcatcher of choice. 881 00:46:08,514 --> 00:46:11,094 -Loving this stuff, man. Loving it. -Right. 882 00:46:11,094 --> 00:46:13,923 -It's fun again. It's fun again. -It's, seriously- 883 00:46:13,923 --> 00:46:16,694 It gives me autonomy. I don't have to ask everybody to do stuff for me, like, 'cause 884 00:46:16,694 --> 00:46:17,554 -I don't have time to do it. -Yeah. 885 00:46:17,554 --> 00:46:19,814 I literally just ask my agent to do it, and it's done. 886 00:46:19,814 --> 00:46:20,794 -Yep. -So fun. 887 00:46:20,794 --> 00:46:21,814 -Here's a proof of concept. -I'll stop. 888 00:46:21,814 --> 00:46:25,834 -Go, go deal with it. But yeah, it's good. -Exactly. Exactly. All right, everybody. 889 00:46:25,834 --> 00:46:28,934 -Yeah. -Sounds good. We'll see you next time. 890 00:46:28,934 --> 00:46:30,694 -Peace. -Bye.