1 00:00:00,000 --> 00:00:01,680 I don't consider BMad vibe coding. 2 00:00:01,680 --> 00:00:04,304 I think of it as the antithesis of vibe coding because you're 3 00:00:04,304 --> 00:00:05,910 actually putting some thought in working with a plan. 4 00:00:06,000 --> 00:00:09,300 Today's guest is Brian Madison, creator of the BMad Method, a 5 00:00:09,300 --> 00:00:11,592 spec-driven AI development framework that transformed his 6 00:00:11,592 --> 00:00:15,036 team from skeptics to 100% agentic developers, now used by 7 00:00:15,036 --> 00:00:17,160 a community of 30,000 engineers worldwide. 8 00:00:17,370 --> 00:00:20,139 What I really started to notice early on with the AI is that it 9 00:00:20,280 --> 00:00:22,742 can be an amazing facilitator. It's almost like pair 10 00:00:22,742 --> 00:00:24,652 programming. If you can explain, what is BMad 11 00:00:24,652 --> 00:00:26,801 Method? The BMad Method right now has 12 00:00:26,801 --> 00:00:28,924 multiple workflows. You work with a product 13 00:00:28,932 --> 00:00:32,107 manager to produce a PRD. You work with an architect to 14 00:00:32,107 --> 00:00:33,710 produce your architecture or solution design. 15 00:00:33,769 --> 00:00:35,885 Each of these artifacts will eventually produce what are 16 00:00:35,885 --> 00:00:37,910 user stories and epics to break your work down. 17 00:00:37,910 --> 00:00:40,195 Product manager started seeing this and saying, wow, this can 18 00:00:40,200 --> 00:00:42,980 actually help me produce better PRDs than I ever had before. 19 00:00:42,980 --> 00:00:45,542 What kind of impact do you see? Is it like productivity impact, 20 00:00:45,542 --> 00:00:47,030 the number of people becomes lesser? 21 00:00:47,105 --> 00:00:49,925 We're at a point now where a single developer can definitely 22 00:00:49,925 --> 00:00:53,176 finish a user story much faster. A user story that used to take a 23 00:00:53,176 --> 00:00:54,904 week, maybe it takes hours or a day now. 24 00:00:54,995 --> 00:00:57,574 When the work is defined, the output is gonna be quicker. 25 00:00:57,635 --> 00:00:59,974 As we develop better tools, it will get faster and faster. 26 00:01:00,004 --> 00:01:03,155 If the unit of work used to be the user story, I really believe 27 00:01:03,214 --> 00:01:06,229 very soon, the unit of work has become the feature or the epic. 28 00:01:06,380 --> 00:01:08,899 Engineers now kind of like lose their identity because code is 29 00:01:08,899 --> 00:01:10,580 largely written by AI agents now. 30 00:01:10,789 --> 00:01:13,520 A lot of developers, their identity is tied to how good or 31 00:01:13,520 --> 00:01:15,938 how fast they can type the code. It's just another abstraction, 32 00:01:15,938 --> 00:01:17,312 right? It's you're almost becoming 33 00:01:17,320 --> 00:01:20,327 more of a leader of AI. One of our greatest skills is 34 00:01:20,335 --> 00:01:23,036 the ability to take a problem and decompose it into small 35 00:01:23,036 --> 00:01:24,907 things. That is not a natural thing for 36 00:01:24,907 --> 00:01:27,530 most people, which is why we're gonna work with AI so well. 37 00:01:44,042 --> 00:01:46,488 Hey, quick pause. My goal with Tech Lead Journal 38 00:01:46,488 --> 00:01:49,072 is simple. Learn from the best in tech so 39 00:01:49,072 --> 00:01:52,442 we can all grow together. If this resonates with you, hit 40 00:01:52,442 --> 00:01:54,092 subscribe to follow the channel. 41 00:01:54,182 --> 00:01:57,068 It's the biggest way for you to support the show and help us 42 00:01:57,068 --> 00:01:59,102 keep bringing great guests and insights to you. 43 00:01:59,642 --> 00:02:01,562 Thanks for being here, and let's get back to it. 44 00:02:02,779 --> 00:02:04,559 Hello guys. Welcome back to another new 45 00:02:04,559 --> 00:02:06,419 episode of the Tech Lead Journal podcast. 46 00:02:06,419 --> 00:02:09,889 Today, I'm very excited to meet someone who is the creator 47 00:02:09,889 --> 00:02:15,271 of the BMad method, B-M-A-D. So I haven't heard about BMad 48 00:02:15,271 --> 00:02:18,235 method before, but someone in my community actually mentioned 49 00:02:18,235 --> 00:02:22,204 about it, and he found it really interesting and useful to 50 00:02:22,204 --> 00:02:25,984 use it to apply agentic AI and all that for your software 51 00:02:25,984 --> 00:02:28,543 development. So I think, today, I'm really 52 00:02:28,543 --> 00:02:30,109 looking forward for this conversation. 53 00:02:30,109 --> 00:02:31,424 Brian Madison, welcome to the show. 54 00:02:32,975 --> 00:02:34,760 Thank you, Henry. I'm so excited to be here. 55 00:02:34,760 --> 00:02:37,430 And that's so cool that somebody in your community, you know, 56 00:02:37,430 --> 00:02:40,350 let you know about the BMad. So I'm happy to be here and talk 57 00:02:40,350 --> 00:02:42,650 to it about, you know, talk about it with you. 58 00:02:42,650 --> 00:02:44,614 It's very exciting. Yeah. 59 00:02:45,184 --> 00:02:48,616 So maybe before we go into the BMad, so when I did a little bit 60 00:02:48,616 --> 00:02:50,699 of research looking at your background, one interesting 61 00:02:50,699 --> 00:02:54,061 thing that I found is that you actually started in the US 62 00:02:54,069 --> 00:02:56,704 Army first before going into software engineering, software 63 00:02:56,704 --> 00:02:59,542 development. So maybe tell us this unique 64 00:02:59,550 --> 00:03:01,594 background, how did it shape you? 65 00:03:02,044 --> 00:03:05,524 And what actually brought you, you know, the knowledge from US 66 00:03:05,524 --> 00:03:08,510 Army to software engineering? Yeah, sure. 67 00:03:08,510 --> 00:03:11,180 Henry. Yeah, that's a great question. 68 00:03:11,180 --> 00:03:14,260 I started out actually in electronics and engineering, 69 00:03:14,260 --> 00:03:17,000 so I've pretty much been in engineering all my life. 70 00:03:17,000 --> 00:03:19,275 You know, I guess I should go back even before the Army when 71 00:03:19,275 --> 00:03:22,674 I was a child, you know, I grew up with a Texas Instruments 72 00:03:22,682 --> 00:03:25,130 computer and I started, you know, probably programming when 73 00:03:25,130 --> 00:03:28,490 I was, I don't know, I don't even — Like I remember being in 74 00:03:28,490 --> 00:03:30,710 kindergarten and typing away on computers. 75 00:03:31,670 --> 00:03:34,910 But I really had a love for computers and electronics and 76 00:03:35,120 --> 00:03:40,190 went in the army to do that. It really did shape me in ways 77 00:03:40,190 --> 00:03:42,490 I never expected. I know this is gonna sound a 78 00:03:42,498 --> 00:03:45,650 little cliche, but it really did teach me leadership. 79 00:03:46,910 --> 00:03:50,048 I probably grew up as a child with not a lot of 80 00:03:50,048 --> 00:03:53,248 self-confidence maybe, and I really developed a sense of 81 00:03:53,256 --> 00:03:57,265 self-confidence and leadership in the army that I don't know if 82 00:03:57,265 --> 00:04:00,440 I - maybe I always had it - but I really developed that. 83 00:04:01,490 --> 00:04:05,780 In the Army though, you really do turn, you know, get to lead 84 00:04:05,790 --> 00:04:10,704 in interesting situations. And there's some things that I 85 00:04:10,704 --> 00:04:14,834 learned there that have really — even now I was thinking about 86 00:04:14,834 --> 00:04:17,720 this the other day — it really plays in my career right now. 87 00:04:18,769 --> 00:04:22,427 When you're leading, you know, you have to show 88 00:04:22,427 --> 00:04:25,647 self-confidence don't always have that for the benefit of the 89 00:04:25,647 --> 00:04:28,523 people following you — and this really applies even in the 90 00:04:28,523 --> 00:04:30,280 workplace when you're, you know, leading, when you're the 91 00:04:30,280 --> 00:04:34,301 tech lead on a team — you have to show some confidence and 92 00:04:34,301 --> 00:04:37,955 leadership, but you also have to place trust on your team. 93 00:04:38,915 --> 00:04:42,947 You're responsible for the team. You are the owner at the end of 94 00:04:42,947 --> 00:04:44,425 the day. You are responsible for the 95 00:04:44,425 --> 00:04:48,210 actions of your team, but you have to trust in your team and 96 00:04:48,210 --> 00:04:51,035 delegate. And I think, I learned a lot of 97 00:04:51,035 --> 00:04:55,339 that because it's very easy to think you have to do it all, but 98 00:04:55,339 --> 00:04:58,697 you really learn when you're managing a large group of team 99 00:04:58,697 --> 00:05:01,882 in the army or a squad. You have to really rely on the 100 00:05:01,882 --> 00:05:04,061 people on your team. And that's, you know, something 101 00:05:04,061 --> 00:05:06,425 that's hard to do at first. And I really learned that. 102 00:05:06,425 --> 00:05:08,198 Another great thing that happened to me in the Army 103 00:05:08,198 --> 00:05:10,745 though, Henry, is I got very lucky. 104 00:05:10,745 --> 00:05:15,281 I was assigned to a lot of NATO units, which means I was 105 00:05:15,281 --> 00:05:19,250 working with the Dutch and the French and Canadians, and so 106 00:05:19,250 --> 00:05:22,627 many different countries. So I was assigned in these units 107 00:05:22,627 --> 00:05:25,565 with very diverse groups of people that I would've never 108 00:05:25,565 --> 00:05:27,921 been exposed to. I got to travel the world and 109 00:05:27,921 --> 00:05:29,615 get outside of the bubble of America. 110 00:05:29,615 --> 00:05:31,736 A lot of people that grew up in America, they're kind of like 111 00:05:31,736 --> 00:05:35,061 in this bubble. I loved traveling the world and 112 00:05:35,061 --> 00:05:37,925 just meeting people everywhere, and I really have 113 00:05:37,925 --> 00:05:39,905 carried that through all my life. 114 00:05:41,375 --> 00:05:45,327 I really strongly believe in good, diverse teams with 115 00:05:45,327 --> 00:05:47,893 interesting perspectives. And I don't — I mean, maybe I 116 00:05:47,893 --> 00:05:50,279 would've developed that over time —but I really enjoyed just 117 00:05:50,279 --> 00:05:53,444 that opportunity to see so many different perspectives and 118 00:05:53,444 --> 00:05:55,865 meet people from so, you know, so many places. 119 00:05:55,865 --> 00:06:00,594 So that's really all. Probably indirectly, shaped my 120 00:06:00,594 --> 00:06:03,722 career. And then when I got out of the 121 00:06:03,722 --> 00:06:06,355 military, it was an easy transition into, still working 122 00:06:06,355 --> 00:06:09,051 for the Department of Defense. That's what really transitioned 123 00:06:09,051 --> 00:06:11,620 me into working on simulations and computers. 124 00:06:12,240 --> 00:06:16,708 And I realized I really loved just working with computers 125 00:06:16,708 --> 00:06:19,886 and software and programming, and kind of made a career out 126 00:06:19,886 --> 00:06:21,728 of it since then. So it really, you know, it 127 00:06:21,736 --> 00:06:26,090 really shaped where I am now. So it's been a long time, Henry. 128 00:06:26,090 --> 00:06:28,740 It's been, you know, over 20 years now since I've been in 129 00:06:28,740 --> 00:06:31,670 the military. And so it's kind of interesting 130 00:06:31,670 --> 00:06:34,266 thinking back on it and how it has kind of shaped where I've 131 00:06:34,266 --> 00:06:36,219 ended up. Yeah. 132 00:06:36,579 --> 00:06:38,679 Thanks for sharing your background and your story. 133 00:06:38,679 --> 00:06:40,929 So definitely very interesting journey, I would say. 134 00:06:41,289 --> 00:06:43,089 And you mentioned about self-confidence. 135 00:06:43,569 --> 00:06:47,585 From many engineers that I know, some of them actually kind of 136 00:06:47,585 --> 00:06:50,255 like lack self-confidence, especially, you know, in a big 137 00:06:50,255 --> 00:06:52,809 team set up or maybe talking to executives. 138 00:06:52,959 --> 00:06:56,799 Some engineers really did have challenges, especially, you 139 00:06:56,799 --> 00:07:00,159 know, when conveying messages and they're somehow not 140 00:07:00,159 --> 00:07:02,199 confident, especially talking about technical stuff. 141 00:07:02,324 --> 00:07:05,160 Maybe from your point of view, right, from your experience, 142 00:07:05,160 --> 00:07:08,895 because you have turned yourself to more self-confident, how 143 00:07:08,895 --> 00:07:12,649 would you advise people to kind of like be more self-confident? 144 00:07:14,396 --> 00:07:17,392 You know, that can be something that's really hard at first, 145 00:07:17,392 --> 00:07:19,516 especially if you do struggle with self-confidence. 146 00:07:19,556 --> 00:07:24,050 But a few things that I use, and I have shared this with other 147 00:07:24,050 --> 00:07:29,263 people, is if you cannot explain an idea, it doesn't matter how 148 00:07:29,263 --> 00:07:32,635 good it is, if you cannot convey the idea. 149 00:07:32,635 --> 00:07:36,265 So sometimes you have to believe in what you're doing and just 150 00:07:36,535 --> 00:07:39,991 share the idea with confidence. Some people will say fake it 151 00:07:39,991 --> 00:07:41,515 till you make it. I don't know. 152 00:07:41,515 --> 00:07:43,675 That's, I don't know if that's really the right way to say it, 153 00:07:43,675 --> 00:07:47,388 but you just, you gotta develop that, whether you feel it 154 00:07:47,388 --> 00:07:49,237 inside. Everybody suffers from imposter 155 00:07:49,237 --> 00:07:51,025 syndrome. I still do, Henry. 156 00:07:51,135 --> 00:07:52,615 I mean, I'm feeling it right now, right? 157 00:07:52,615 --> 00:07:54,685 But like, should I really be on this podcast? 158 00:07:55,885 --> 00:07:58,435 But no, I, actually, I'm happy to be here and I believe it. 159 00:07:58,435 --> 00:08:01,685 But everybody's gonna have that voice of self-doubt or many 160 00:08:01,685 --> 00:08:06,040 people will. You just gotta remember that 161 00:08:06,048 --> 00:08:10,040 not everybody sees, you know. They're listening to you because 162 00:08:10,040 --> 00:08:11,960 they believe you have something to tell them. 163 00:08:12,380 --> 00:08:14,360 I believe I have something to share with you, Henry. 164 00:08:14,360 --> 00:08:15,950 I believe you have stuff to share with me. 165 00:08:16,340 --> 00:08:20,060 So whether we have that voice in our head, just remember that 166 00:08:20,060 --> 00:08:23,480 is just the voice in your head and not everybody feels that 167 00:08:23,480 --> 00:08:25,735 way because or else they wouldn't even be listening to 168 00:08:25,735 --> 00:08:27,430 you. So feel good about yourself, 169 00:08:27,438 --> 00:08:30,517 share your message. But as engineers, a lot of times 170 00:08:30,517 --> 00:08:34,840 we focus on the code and, you know, the beauty of it and the 171 00:08:34,840 --> 00:08:38,419 technical details. But share it with confidence to 172 00:08:38,419 --> 00:08:42,587 really explain what it is you're doing to your audience so 173 00:08:42,587 --> 00:08:46,763 they can appreciate it. I think just sometimes trying to 174 00:08:46,763 --> 00:08:51,062 get out of your head, it's easier said than done, but this 175 00:08:51,062 --> 00:08:54,830 is something that can be dealt with and built up over time. 176 00:08:55,010 --> 00:08:57,880 And sometimes it takes time, but I think everybody can get 177 00:08:57,880 --> 00:08:59,095 there. If you work at it, it's a 178 00:08:59,103 --> 00:09:02,304 skill that you can develop. Yeah, speaking about imposter 179 00:09:02,312 --> 00:09:05,303 syndrome, definitely, I also feel it. you know, even though 180 00:09:05,303 --> 00:09:06,841 how many years of experience you have worked. 181 00:09:06,841 --> 00:09:08,899 Especially these days, especially all these 182 00:09:08,899 --> 00:09:10,951 technologies seems to advance rapidly. 183 00:09:11,381 --> 00:09:12,901 You know, you won't know everything, right? 184 00:09:12,901 --> 00:09:15,781 So definitely always there's a imposter syndrome, especially 185 00:09:15,781 --> 00:09:18,556 people who work in tech. And yeah, being confident is 186 00:09:18,564 --> 00:09:21,617 something that can be trained. So I think for everyone who 187 00:09:21,617 --> 00:09:24,027 listen, if you're not confident, you can train that 188 00:09:24,027 --> 00:09:26,323 maybe in a smaller scale. I really, really I believed 189 00:09:26,323 --> 00:09:27,501 that too, Henry. Yeah. 190 00:09:27,661 --> 00:09:29,650 And I think that's a, you know, I think that is a good 191 00:09:29,650 --> 00:09:31,845 reminder for people. And I, still tell that to 192 00:09:31,845 --> 00:09:34,407 people, because I'm pretty confident when I'm speaking in 193 00:09:34,407 --> 00:09:37,157 front of people, when I'm presenting to a board or to 194 00:09:37,157 --> 00:09:39,734 leadership and people see that, but they might not realize 195 00:09:39,742 --> 00:09:43,387 that as leaders we still suffer from that weird voice in our 196 00:09:43,387 --> 00:09:46,217 head, right? it doesn't necessarily go away. 197 00:09:46,937 --> 00:09:49,069 But when you're — I know when you're new in a career, you 198 00:09:49,069 --> 00:09:51,967 feel like, wow, do I really belong here, right? 199 00:09:52,357 --> 00:09:54,637 Or wow, I got lucky getting this job. 200 00:09:54,637 --> 00:09:57,427 But just know that everybody has gone through that. 201 00:09:57,877 --> 00:10:00,547 It's a universal feeling that I think most go through. 202 00:10:00,547 --> 00:10:04,499 And I think as leaders, if we share that we even still go 203 00:10:04,499 --> 00:10:07,420 through that sometimes, I think that can actually help build 204 00:10:07,420 --> 00:10:10,519 some self-confidence and make people not feel like, oh, I must 205 00:10:10,519 --> 00:10:12,187 be weird because I'm feeling this. 206 00:10:12,547 --> 00:10:14,587 Because I used to feel like that and I'm like, wow. 207 00:10:15,007 --> 00:10:17,107 Like I remember when I was, you know, new in my career. 208 00:10:17,287 --> 00:10:19,315 I used to have those thoughts and not think that anybody 209 00:10:19,315 --> 00:10:21,421 felt that way. But then you kind of start to 210 00:10:21,421 --> 00:10:24,116 realize it's universal. Yeah. 211 00:10:24,716 --> 00:10:27,476 So let's go into your the BMad thing, right? 212 00:10:27,476 --> 00:10:31,154 So first of all, I would like to clarify what does BMad stand 213 00:10:31,154 --> 00:10:33,320 for? Because I see a couple of 214 00:10:33,328 --> 00:10:36,326 different definition, but I also kind of like see your name. 215 00:10:36,326 --> 00:10:39,196 Brian Madison. Is this like Brian Madison as 216 00:10:39,196 --> 00:10:41,036 BMad? So tell us what does it stand 217 00:10:41,036 --> 00:10:44,582 for actually? It really started out as just 218 00:10:44,590 --> 00:10:48,062 people — first of all, Henry, if you like, you can't call me 219 00:10:48,062 --> 00:10:49,897 BMad. People actually call me BMad 220 00:10:49,897 --> 00:10:52,700 in my day-to-day life. It is just been a nickname for 221 00:10:52,700 --> 00:10:56,414 a long time. And when I was starting BMad and 222 00:10:56,414 --> 00:11:00,377 I think we'll talk a little bit more about why I started 223 00:11:00,377 --> 00:11:03,646 BMad and how it started. It did, it did not start out as 224 00:11:03,646 --> 00:11:06,648 this grand vision where I needed to put a good name on it 225 00:11:06,648 --> 00:11:10,022 or anything. So I stuck with BMad and I made 226 00:11:10,022 --> 00:11:14,117 the acronym match something. When I was starting BMad though, 227 00:11:14,117 --> 00:11:18,132 I really was trying to map processes onto a way of working 228 00:11:18,132 --> 00:11:20,372 that I've always worked, which was Agile. 229 00:11:20,372 --> 00:11:24,350 I've been working, you know, for the last 15+ years out of my 230 00:11:24,350 --> 00:11:28,562 25+ year career using Agile methodologies. 231 00:11:28,562 --> 00:11:30,752 I've been involved in Agile transformations. 232 00:11:30,752 --> 00:11:34,272 And for me it was a very good thought framework, so I came up 233 00:11:34,272 --> 00:11:37,232 with the breakthrough method of Agile AI-driven development. 234 00:11:38,672 --> 00:11:41,012 Over time I realized it's very hard to say. 235 00:11:41,072 --> 00:11:45,682 It's very long, and it actually ends up spelling like 236 00:11:45,682 --> 00:11:47,891 B-M-A-A-D-D. So it doesn't even work that 237 00:11:47,891 --> 00:11:50,940 well. But for what I was doing, it 238 00:11:50,940 --> 00:11:54,842 actually does make sense. So, yes, it is my name. 239 00:11:56,162 --> 00:11:59,252 It is the breakthrough method of Agile AI-driven development. 240 00:11:59,702 --> 00:12:02,870 But as the platform and what we've been building as a 241 00:12:02,870 --> 00:12:06,277 community has evolved, I now just call it Build More 242 00:12:06,277 --> 00:12:09,905 Architect Dreams because — we'll talk a little bit about 243 00:12:09,905 --> 00:12:12,938 what the BMad method is, but it's become more than what it 244 00:12:12,938 --> 00:12:15,242 originally meant. So yes, there's not a clear 245 00:12:15,242 --> 00:12:17,522 answer, Henry. It means multiple things. 246 00:12:17,522 --> 00:12:21,441 But, you know, it's funny, I was watching another podcast or a 247 00:12:21,441 --> 00:12:23,537 video, and I didn't even know about it. 248 00:12:23,537 --> 00:12:27,099 I'm watching it. And it's funny because like 249 00:12:27,099 --> 00:12:30,827 halfway through the host just realized, oh, wait a minute, did 250 00:12:30,827 --> 00:12:32,297 this guy just name it after himself? 251 00:12:33,232 --> 00:12:36,772 And he, so, so yes. So I don't know. 252 00:12:36,792 --> 00:12:40,173 It's funny story, I think, but yes, the BMad method means 253 00:12:40,173 --> 00:12:44,728 multiple things. But it really is about taking a 254 00:12:44,728 --> 00:12:50,152 way of structuring Agile, and, you know, a way to think about 255 00:12:50,152 --> 00:12:53,117 working in an Agile method with AI to do context 256 00:12:53,257 --> 00:12:54,597 engineering at the end of the day. 257 00:12:55,327 --> 00:12:57,789 Right. So definitely it's a fun trivia, 258 00:12:57,789 --> 00:12:59,924 right? So for people who are kind of 259 00:12:59,924 --> 00:13:02,036 like maybe confused, what does it stand for? 260 00:13:02,036 --> 00:13:04,916 I think maybe, we'll, let's just call it whatever that is name 261 00:13:04,916 --> 00:13:07,451 in the GitHub, which is like, Build More Architect Dreams, 262 00:13:07,451 --> 00:13:09,566 right? Or maybe simply your name as the 263 00:13:09,566 --> 00:13:12,929 creator. You know what's funny, Henry, is 264 00:13:12,929 --> 00:13:16,847 a lot of people will like use AI just to create like an 265 00:13:16,847 --> 00:13:19,133 article about BMad. Maybe they wanna just like, you 266 00:13:19,133 --> 00:13:20,897 know, put something up on their social or whatever. 267 00:13:21,527 --> 00:13:24,621 And the funny thing is, it's a little bit better known now by 268 00:13:24,621 --> 00:13:27,320 the models, but the older models, like every time I see 269 00:13:27,320 --> 00:13:29,777 a different article about BMad, it has a different acronym. 270 00:13:29,987 --> 00:13:32,867 Because the AI would just like guess what BMad stands for. 271 00:13:33,167 --> 00:13:37,231 So like if you Google it, you might find like a hundred 272 00:13:37,231 --> 00:13:39,462 different answers just because people keep putting up 273 00:13:39,462 --> 00:13:41,907 different articles and it just whatever it spits out, it's 274 00:13:41,907 --> 00:13:43,972 like. So it's actually pretty fun 275 00:13:43,972 --> 00:13:45,917 'cause sometimes I'll see those and I'm like, dang, I 276 00:13:45,917 --> 00:13:48,107 should have used that one, I actually like that one better. 277 00:13:49,291 --> 00:13:51,393 Right. Speaking of, which, right, so 278 00:13:51,393 --> 00:13:53,931 maybe if you can explain what is BMad method? 279 00:13:53,931 --> 00:13:57,087 Because I assume some people might have heard about it, but 280 00:13:57,087 --> 00:13:59,247 some may not, right? Especially even like for people 281 00:13:59,247 --> 00:14:01,943 who have used, you know, a lot of AI in their software 282 00:14:01,943 --> 00:14:04,081 development, some may not have heard about it. 283 00:14:05,087 --> 00:14:07,037 Yeah, lemme give you a good explanation. 284 00:14:08,027 --> 00:14:11,849 The BMad method as it is right now, if you go to, you can 285 00:14:11,849 --> 00:14:14,777 look at my YouTube channel, there's a masterclass on there. 286 00:14:14,777 --> 00:14:16,895 Or you can go to the — if you just go to GitHub and you 287 00:14:16,895 --> 00:14:19,397 search for the BMad method, you'll find it. 288 00:14:20,647 --> 00:14:26,312 What it is right now is a way to do context and spec 289 00:14:26,312 --> 00:14:29,369 engineering. But what it does, as I 290 00:14:29,369 --> 00:14:32,417 mentioned, is it maps these Agile personas and ways of 291 00:14:32,417 --> 00:14:36,703 working into doing Agile. So when you start out with a 292 00:14:36,703 --> 00:14:39,999 project, if you've done vibe coding before, Henry, and I 293 00:14:39,999 --> 00:14:42,663 think everybody's tried vibe coding at this point, you 294 00:14:42,663 --> 00:14:46,089 eventually hit a wall. Even now the models are very 295 00:14:46,089 --> 00:14:48,431 good. You know, they can almost one 296 00:14:48,439 --> 00:14:53,225 shot very simple applications. But as you keep working with the 297 00:14:53,225 --> 00:14:56,961 agent mode in these tools, eventually you hit a wall and 298 00:14:56,961 --> 00:15:00,272 you still hit a wall where — and this is kind of the vibe 299 00:15:00,272 --> 00:15:03,337 coding thing, right? You just, you tell it to do one 300 00:15:03,337 --> 00:15:05,857 thing and it starts doing it three different ways. 301 00:15:06,357 --> 00:15:09,699 Or it gets stuck in these death spirals where it starts testing 302 00:15:09,699 --> 00:15:12,770 over and over again and then it will just randomly start 303 00:15:12,770 --> 00:15:16,773 throwing stuff away. The BMad method right now has 304 00:15:16,781 --> 00:15:20,407 multiple what will be skills, but they're just workflows right 305 00:15:20,407 --> 00:15:23,025 now, organized in a way to work through. 306 00:15:23,025 --> 00:15:27,705 So you work with a product manager to produce a PRD. 307 00:15:28,155 --> 00:15:30,650 And you know, a lot of people produce PRDs and then they 308 00:15:30,650 --> 00:15:34,131 jump right into the coding. But what I found is just a PRD 309 00:15:34,139 --> 00:15:37,502 on its own always isn't enough. So then you can work with a 310 00:15:37,502 --> 00:15:40,305 design agent, a UX expert to help produce your design. 311 00:15:40,665 --> 00:15:43,965 You work with an architect to produce your architecture or 312 00:15:43,965 --> 00:15:46,486 solution design. And I can go into the benefits 313 00:15:46,494 --> 00:15:48,855 of some of these and why these different artifacts matter. 314 00:15:48,855 --> 00:15:51,611 But each of these artifacts, if you think of it as a funnel, 315 00:15:51,611 --> 00:15:54,933 Henry, they kind of each feed into the next one to 316 00:15:54,941 --> 00:15:58,109 eventually produce what our user stories and epics to break your 317 00:15:58,109 --> 00:16:00,827 work down. And then those get filled with 318 00:16:00,827 --> 00:16:03,405 context, with a different agent and a different process. 319 00:16:03,765 --> 00:16:07,548 So then when you actually develop with the agent mode, 320 00:16:07,548 --> 00:16:11,434 you take your story, it has everything it needs in that one 321 00:16:11,434 --> 00:16:14,904 user story so the agent mode doesn't have to spend a lot of 322 00:16:14,904 --> 00:16:17,296 context reading all of these different documents and trying 323 00:16:17,296 --> 00:16:20,664 to figure out what to do. So the theory is when it 324 00:16:20,664 --> 00:16:24,198 actually finally picks up that user story, it can pretty much 325 00:16:24,198 --> 00:16:26,826 know everything it needs to know to implement the story 326 00:16:26,826 --> 00:16:29,985 without making bad decisions. It's gonna know exactly what 327 00:16:29,993 --> 00:16:33,420 libraries and frameworks you're using and make good decisions. 328 00:16:34,410 --> 00:16:39,175 But the difference I think between the BMad Method and 329 00:16:39,175 --> 00:16:41,345 other frameworks, especially when it first started, there's 330 00:16:41,345 --> 00:16:44,131 a lot of, you know, similar stuff out there now. 331 00:16:45,681 --> 00:16:48,737 But I didn't wanna just come up with something where I just 332 00:16:48,737 --> 00:16:52,549 give the AI my idea and it spits out all these documents and 333 00:16:52,549 --> 00:16:55,761 give them to me. One of the things I really 334 00:16:55,761 --> 00:16:59,309 realized, it's funny but the BMad method has been going about 335 00:16:59,309 --> 00:17:04,769 a year now, almost a year. A lot of people treat the AI 336 00:17:04,769 --> 00:17:08,201 like a ChatGPT or, you know, even the agent mode. 337 00:17:08,580 --> 00:17:10,941 They ask it a question, give it an answer. 338 00:17:11,271 --> 00:17:13,671 I have this function, how would you refactor it? 339 00:17:14,781 --> 00:17:18,471 What libraries should I use to do X, Y, Z? 340 00:17:19,881 --> 00:17:23,391 What I really started to notice early on with the AI is that it 341 00:17:23,391 --> 00:17:27,771 can be an amazing facilitator. I come from a background like 342 00:17:27,771 --> 00:17:29,679 you, Henry. I've been an engineer, like I 343 00:17:29,679 --> 00:17:33,355 said, for 25 years. I know a lot of things and I 344 00:17:33,355 --> 00:17:36,221 have opinions. And I think especially as 345 00:17:36,221 --> 00:17:39,111 engineers, and this is something else we can talk about 346 00:17:39,111 --> 00:17:43,706 if we have time, but don't think of the AI as just a 347 00:17:43,706 --> 00:17:47,332 replacement for your own skills. So the way I've built the BMad 348 00:17:47,340 --> 00:17:50,181 method is when it goes through these workflows, it's having 349 00:17:50,181 --> 00:17:53,931 multi-turn conversations with you and trying to pull the best 350 00:17:53,931 --> 00:17:57,610 information out of you. It's gonna ask you questions and 351 00:17:57,610 --> 00:18:01,531 it will give you suggestions, but it really tries to teach you 352 00:18:01,531 --> 00:18:04,431 as the expert, working with the AI as a partner. 353 00:18:04,731 --> 00:18:06,171 It's almost like pair programming. 354 00:18:06,501 --> 00:18:08,991 So you might not be an expert product person because you're an 355 00:18:08,991 --> 00:18:12,891 engineer, but working through the PRD, it's gonna ask you 356 00:18:12,891 --> 00:18:15,021 things that you might not normally think of. 357 00:18:15,021 --> 00:18:17,121 Like, you know, do you need admin? 358 00:18:17,391 --> 00:18:19,753 What are the user journeys? What are the flows through your 359 00:18:19,753 --> 00:18:21,411 system? It will help you come up with 360 00:18:21,411 --> 00:18:24,189 more requirements. Because the more that you can 361 00:18:24,189 --> 00:18:27,483 kind of specify upfront, the less you'll be having to change 362 00:18:27,483 --> 00:18:29,511 when you're actually building your application. 363 00:18:29,511 --> 00:18:32,921 So that is really why you wanna go through this whole funnel. 364 00:18:33,677 --> 00:18:36,527 But there's other features also of the BMad method. 365 00:18:36,567 --> 00:18:41,578 One of the things that people really seem to love, and this 366 00:18:41,578 --> 00:18:45,017 surprised me, but we have a brainstorming mode at the very 367 00:18:45,017 --> 00:18:47,773 top of the funnel. So if you don't even know 368 00:18:47,773 --> 00:18:51,030 exactly what you wanna do, this will do brainstorming with 369 00:18:51,030 --> 00:18:55,590 you, acting as a facilitator. And people might have done this, 370 00:18:55,590 --> 00:18:59,339 you know, if you go into like ChatGPT or some of these other 371 00:18:59,339 --> 00:19:02,478 ones, there are brainstorming modes in there, but I've 372 00:19:02,478 --> 00:19:04,485 developed something that I feel is really special because 373 00:19:04,485 --> 00:19:07,480 people have told me sometimes that's all they use. 374 00:19:07,660 --> 00:19:10,845 And people have told me they've spent six hours just using the 375 00:19:10,845 --> 00:19:14,505 brainstorming agent, using it for everything across their 376 00:19:14,505 --> 00:19:16,930 lives, which really surprised me. 377 00:19:16,930 --> 00:19:18,758 You know, it's funny when you release a piece of software and 378 00:19:18,758 --> 00:19:21,410 you start learning how other people are using it, it gives 379 00:19:21,410 --> 00:19:23,788 you new ideas. And so that's what happened with 380 00:19:23,788 --> 00:19:25,510 here. So the BMad method has evolved. 381 00:19:25,720 --> 00:19:27,977 I know this is a very long answer of what is the BMad 382 00:19:27,977 --> 00:19:29,410 method. I hope this is making sense, 383 00:19:29,410 --> 00:19:31,666 Henry. But it really is this platform 384 00:19:31,666 --> 00:19:34,846 with all of these different skills and tools, but they're 385 00:19:34,846 --> 00:19:39,323 all meant to work together to kind of eventually produce 386 00:19:39,323 --> 00:19:41,580 software. But yeah, the brainstorming is 387 00:19:41,580 --> 00:19:43,480 special because it has all these techniques in it. 388 00:19:43,890 --> 00:19:46,880 It's working with you. I am horrible at brainstorming. 389 00:19:47,240 --> 00:19:50,204 If you give me a whiteboard like Miro or something and they 390 00:19:50,204 --> 00:19:54,076 have these brainstorming methods like built into it, I stare at 391 00:19:54,076 --> 00:19:57,570 it and it gives me the rules, and I still, I don't know, I 392 00:19:57,570 --> 00:20:00,129 just draw a blank. And I used to always think I'm 393 00:20:00,137 --> 00:20:02,360 not a creative person, and so many people have told me this. 394 00:20:02,360 --> 00:20:04,470 I never realized it was, I was a creative person. 395 00:20:05,110 --> 00:20:09,704 When you treat the AI as a facilitator, and it is not 396 00:20:09,704 --> 00:20:12,498 doing the brainstorming for you, it's perf, it's specifically 397 00:20:12,506 --> 00:20:15,553 set up not to do the brainstorming for you but to 398 00:20:15,553 --> 00:20:18,423 give you prompts and ideas and work through these different 399 00:20:18,423 --> 00:20:20,870 interesting techniques, all of a sudden you're coming up with 400 00:20:20,870 --> 00:20:23,968 stuff and then it helps you kind of distill down what you came up 401 00:20:23,968 --> 00:20:26,548 with. I believe I'm a very creative 402 00:20:26,548 --> 00:20:28,370 person now. And that is what so many 403 00:20:28,378 --> 00:20:31,396 people have told me. So it's just so, it's so cool to 404 00:20:31,396 --> 00:20:33,460 see people that think they are not creative. 405 00:20:33,460 --> 00:20:36,904 Now all of a sudden, you know, coming up with creative ideas 406 00:20:36,904 --> 00:20:39,824 through the help of AI, but it is not like the AI coming up 407 00:20:39,824 --> 00:20:41,150 with the ideas. It really is them. 408 00:20:41,150 --> 00:20:45,206 So that is really, I think, the special sauce of all the BMad 409 00:20:45,206 --> 00:20:48,516 method is it treats you as somebody with ideas and it's 410 00:20:48,516 --> 00:20:50,690 not just doing all of it for you. 411 00:20:50,720 --> 00:20:52,860 So let me pause there cause I know that's a lot. 412 00:20:53,273 --> 00:20:55,290 Yeah. Speaking about creativity, I 413 00:20:55,290 --> 00:20:58,480 think that kind of like just reminds me that to be creative, 414 00:20:58,480 --> 00:21:01,008 it comes back to the questions that you ask, right? 415 00:21:01,008 --> 00:21:04,158 Like if you naturally don't think in questions, sometimes 416 00:21:04,218 --> 00:21:07,543 you can get stuck, right? But if you are always like 417 00:21:07,543 --> 00:21:09,595 curious, asking different questions, different ways, 418 00:21:09,595 --> 00:21:12,048 different perspectives, I think you can be creative. 419 00:21:12,053 --> 00:21:14,614 Yeah. And that's the cool thing about 420 00:21:14,614 --> 00:21:17,799 the brainstorming method also. Like if anybody wants to try out 421 00:21:17,799 --> 00:21:21,370 the BMad method, if you try nothing out, just try out the 422 00:21:21,370 --> 00:21:24,092 BMad method brainstorming. It's different than everything 423 00:21:24,100 --> 00:21:26,626 I've seen out there. It's about 50 different 424 00:21:26,634 --> 00:21:30,307 techniques built into it. Depending on what you're working 425 00:21:30,307 --> 00:21:33,010 on, you'll tell it what you wanna do and it'll suggest some 426 00:21:33,010 --> 00:21:34,742 of the techniques. But there's some crazy ones in 427 00:21:34,742 --> 00:21:36,525 there. There's like alien visitors, you 428 00:21:36,525 --> 00:21:39,792 know, and it kind of guides you through if like there were 429 00:21:39,792 --> 00:21:42,444 aliens seeing this for the first time, what would their 430 00:21:42,444 --> 00:21:45,012 perspective be? There's, just, there's idea 431 00:21:45,012 --> 00:21:49,773 gardening, there's some of the classic ones like the thinking 432 00:21:49,773 --> 00:21:53,836 hats and, you know, the seven ways of, you know, dissecting a 433 00:21:53,836 --> 00:21:56,099 problem. But they're all just very 434 00:21:56,099 --> 00:21:57,798 creative. And again it will, like you 435 00:21:57,806 --> 00:22:01,644 said, you hit it on the nail. It's the questions it asks you, 436 00:22:01,644 --> 00:22:04,704 and you'll come up with some very interesting things. 437 00:22:04,944 --> 00:22:06,264 It's a lot of fun. I love doing it. 438 00:22:06,264 --> 00:22:09,152 I, and I even use it for things outside of software 439 00:22:09,152 --> 00:22:10,414 engineering. It's a lot of fun. 440 00:22:11,618 --> 00:22:14,834 So maybe since I'm not like a very advanced BMad user, in 441 00:22:14,834 --> 00:22:18,008 fact, I have never used it in my project before. 442 00:22:18,188 --> 00:22:20,933 So if I can understand it further, first, it's like you 443 00:22:20,933 --> 00:22:23,768 try to solve this kind of like context engineering problem, 444 00:22:23,776 --> 00:22:26,294 especially when you build applications that evolves over 445 00:22:26,294 --> 00:22:28,740 time, right? Especially complex software, you 446 00:22:28,740 --> 00:22:32,180 kind of want to build some kind of context knowledge base 447 00:22:32,180 --> 00:22:35,616 so that the agent will be able to refer to it and always use 448 00:22:35,616 --> 00:22:37,896 that to produce the next output, right? 449 00:22:38,136 --> 00:22:42,232 And the other thing is that you have these different kind of 450 00:22:42,232 --> 00:22:45,720 like personas built in. Maybe some people is, like you 451 00:22:45,720 --> 00:22:49,002 can associate that as like agent skills these days, like 452 00:22:49,002 --> 00:22:51,336 where you have different skills for different purpose. 453 00:22:51,726 --> 00:22:54,641 And, yeah. And then the third thing that I 454 00:22:54,641 --> 00:22:57,456 can associate BMad is actually with spec-driven development. 455 00:22:57,666 --> 00:22:58,926 Maybe let's start from there first. 456 00:22:58,926 --> 00:23:03,174 Is this a different flavor of spec-driven development or is 457 00:23:03,174 --> 00:23:07,017 it something different? Yeah, Henry, that's a great 458 00:23:07,017 --> 00:23:08,602 question. You know, when I kind of gotta 459 00:23:08,610 --> 00:23:12,498 go back to when I started BMad, there was no name for these 460 00:23:12,498 --> 00:23:14,480 things. Like even when I started it, 461 00:23:14,488 --> 00:23:17,312 there wasn't really a name for vibe coding and then like, vibe 462 00:23:17,312 --> 00:23:19,120 coding became a very popular term. 463 00:23:19,190 --> 00:23:21,950 And I don't consider BMad vibe coding. 464 00:23:21,950 --> 00:23:25,135 I think of it as the antithesis of vibe coding because you're 465 00:23:25,135 --> 00:23:27,731 actually putting some thought and planning upfront and 466 00:23:27,731 --> 00:23:31,329 working with a plan. Context engineering was 467 00:23:31,329 --> 00:23:33,800 something I didn't have a name for. 468 00:23:33,980 --> 00:23:36,507 It's funny. I must be like many engineers, 469 00:23:36,515 --> 00:23:37,970 you know, they say naming is hard. 470 00:23:37,970 --> 00:23:40,070 I'm not good at naming and branding things, I guess. 471 00:23:40,070 --> 00:23:45,361 But we were doing context engineering before that was a 472 00:23:45,361 --> 00:23:47,758 term. And if you think about what is 473 00:23:47,758 --> 00:23:50,188 context engineering, again, it's taking those ideas, getting 474 00:23:50,188 --> 00:23:54,088 it down to something. There wasn't really a name for 475 00:23:54,088 --> 00:23:56,202 it yet, but this thing, I was calling it the user story 476 00:23:56,202 --> 00:23:59,745 because that's what I had for my own vocabulary of working in 477 00:23:59,745 --> 00:24:02,314 Agile. So, you know, so long, with 478 00:24:02,322 --> 00:24:06,118 acceptance criteria in it. So a very good user story is not 479 00:24:06,118 --> 00:24:10,531 just, you know, as an X, I wanna do Y so that Z, but it's 480 00:24:10,531 --> 00:24:14,454 also putting in good BDD or some kind of style of acceptance 481 00:24:14,454 --> 00:24:16,952 criteria. Just in the quote 'real world' 482 00:24:16,952 --> 00:24:20,507 before AI, if you do your user stories that way when your 483 00:24:20,507 --> 00:24:23,106 developers pick them up, they really know what needs to be 484 00:24:23,106 --> 00:24:25,460 done for the story. And you can specify things. 485 00:24:26,210 --> 00:24:31,216 By building the stories in that way with the BDD, I was 486 00:24:31,216 --> 00:24:35,078 noticing these huge improvements in the agent to actually be 487 00:24:35,078 --> 00:24:38,414 able to build these complex software applications 100% 488 00:24:38,422 --> 00:24:41,600 without me having to type the code. 489 00:24:41,600 --> 00:24:44,660 And this is even a year ago when the models were nowhere 490 00:24:44,660 --> 00:24:48,760 near what they are now. So that was context engineering. 491 00:24:49,190 --> 00:24:52,820 And what we are really producing are specs. 492 00:24:52,850 --> 00:24:54,590 So they call it spec engineering now. 493 00:24:54,920 --> 00:24:57,110 And that really is what BMad is producing for you. 494 00:24:57,110 --> 00:24:59,540 And now we're starting to lean into the same language and we're 495 00:24:59,540 --> 00:25:03,407 calling things skills and context engineering and spec 496 00:25:03,407 --> 00:25:06,974 engineering. But really the stories with the 497 00:25:06,974 --> 00:25:12,180 AC in them, you know, organized into these epics, are 498 00:25:12,180 --> 00:25:14,716 specifications. And, you know, just like what 499 00:25:14,724 --> 00:25:18,980 we now call spec engineering, it's basically the same thing. 500 00:25:18,980 --> 00:25:20,990 So it's a spec engineering framework. 501 00:25:21,350 --> 00:25:24,474 But the difference with BMad, again, is that you can start 502 00:25:24,474 --> 00:25:28,526 at the top of this thought funnel and it really helps you 503 00:25:28,526 --> 00:25:32,084 plan and think about what you're gonna need before you produce 504 00:25:32,084 --> 00:25:35,624 all of your specs and have it start working so you're not 505 00:25:35,624 --> 00:25:38,480 hopefully changing this much halfway through or forgetting 506 00:25:38,480 --> 00:25:42,436 major features or, oh, I forgot I need a database or I need 507 00:25:42,436 --> 00:25:45,377 authentication. So it will guide you and help 508 00:25:45,377 --> 00:25:47,620 you through that. Now you asked about the agents. 509 00:25:48,270 --> 00:25:51,359 Henry, when I first made the BMad method, and maybe we can 510 00:25:51,359 --> 00:25:56,140 talk a little bit about why I even made it, but it started 511 00:25:56,140 --> 00:26:00,081 out as just six files. So I knew that I wanted to 512 00:26:00,081 --> 00:26:03,431 create a PRD, and then I knew I wanted to create this like 513 00:26:03,431 --> 00:26:06,300 tech stack list, which I called the architecture, and I knew 514 00:26:06,300 --> 00:26:09,710 if I was working on UI, I wanted some UX information there. 515 00:26:10,190 --> 00:26:14,138 So I just, at the time, you know, all of the prompting 516 00:26:14,138 --> 00:26:19,743 guidance was give it a persona, you know, tell it like you are 517 00:26:19,743 --> 00:26:22,925 a professional product manager and you have all these 518 00:26:22,925 --> 00:26:26,372 attributes. And in the single file, it was 519 00:26:26,380 --> 00:26:30,260 talking about here's the PRD and this is the information that we 520 00:26:30,260 --> 00:26:32,420 need, and do this in a facilitative way. 521 00:26:32,420 --> 00:26:34,493 So, you know, you're getting this information out of the 522 00:26:34,493 --> 00:26:37,619 user. It grew over time to be multiple 523 00:26:37,619 --> 00:26:40,619 files in a separate agent, because I wanted to give people 524 00:26:40,619 --> 00:26:42,155 the ability to customize these things. 525 00:26:42,635 --> 00:26:45,759 But the reason the agent is there is because it started out 526 00:26:45,759 --> 00:26:48,468 as kind of tied to the PRD. It was kind of a thought 527 00:26:48,468 --> 00:26:50,273 process. In the real world, when you're 528 00:26:50,273 --> 00:26:53,690 working in Agile on a lot of teams, the product manager is 529 00:26:53,690 --> 00:26:57,413 producing the PRD. Or the lead engineer, the tech 530 00:26:57,413 --> 00:27:00,065 leader, the architect is producing the solution design. 531 00:27:00,065 --> 00:27:02,465 So it's kind of just how it evolved. 532 00:27:02,825 --> 00:27:07,865 But now with agents in the BMad method, the agents have the 533 00:27:07,865 --> 00:27:11,375 ability to have memory and remember things, and can 534 00:27:11,375 --> 00:27:14,297 evolve over time. So it has become much more 535 00:27:14,297 --> 00:27:18,367 than as it started. There's another thing that I 536 00:27:18,367 --> 00:27:21,891 should also mention about the BMad method is it is part of a 537 00:27:21,891 --> 00:27:23,705 platform now. So there is something called 538 00:27:23,705 --> 00:27:26,705 the BMad Builder. It's going through some rework. 539 00:27:26,735 --> 00:27:31,082 It will be up really soon. But it's basically a skill 540 00:27:31,090 --> 00:27:34,919 builder for workflows. So you can create your own 541 00:27:34,927 --> 00:27:37,325 custom workflows and work them into the BMad method. 542 00:27:37,745 --> 00:27:41,759 Unlike most skill builders though and skills in general 543 00:27:41,759 --> 00:27:47,177 is, in the BMad method, they all register into a help system and 544 00:27:47,177 --> 00:27:49,565 they all have sequencing and information. 545 00:27:49,565 --> 00:27:53,735 So as you install different modules, which are like apps for 546 00:27:53,735 --> 00:27:58,172 the BMad method or plugins, you know, is what Anthropic calls 547 00:27:58,172 --> 00:28:01,185 them. The interesting thing is they 548 00:28:01,193 --> 00:28:04,811 all have information about how they're related to each other 549 00:28:04,811 --> 00:28:06,725 and it dynamically populates the help. 550 00:28:06,995 --> 00:28:09,530 So if you don't even know what to do with the BMad method 551 00:28:09,530 --> 00:28:12,341 'cause it can be overwhelming at first, you can just ask the 552 00:28:12,341 --> 00:28:14,585 help and describe to it what you wanna do. 553 00:28:14,975 --> 00:28:17,969 And it will suggest, oh, you don't really know what your 554 00:28:17,969 --> 00:28:20,598 idea is yet. You might wanna start with the 555 00:28:20,598 --> 00:28:23,010 brainstorming. Oh, you installed the Creative 556 00:28:23,018 --> 00:28:25,475 Intelligence Suite and you need to work on a pitch. 557 00:28:25,475 --> 00:28:29,848 Well, maybe you want to use that module and use its 558 00:28:29,848 --> 00:28:33,647 storytelling for business workflow to help you kind of 559 00:28:33,647 --> 00:28:36,858 think of these ideas. So the BMad method has also 560 00:28:36,858 --> 00:28:38,810 grown to be a lot of different modules. 561 00:28:38,810 --> 00:28:41,090 The community can also contribute modules. 562 00:28:42,290 --> 00:28:46,070 And it evolves to help you with many different things. 563 00:28:46,070 --> 00:28:48,880 So it's more than a skill collection because it all ends 564 00:28:48,880 --> 00:28:52,480 up being tied together. But it can do now a lot more 565 00:28:52,488 --> 00:28:56,870 than just helping you get to the end and do software development. 566 00:28:58,709 --> 00:29:00,705 Yeah, sounds like... I know that was a very long 567 00:29:00,705 --> 00:29:04,063 answer. Yeah, sounds like it has grown 568 00:29:04,063 --> 00:29:06,799 since you started it like a year ago or so, right? 569 00:29:06,799 --> 00:29:10,872 So it has grown into like more this framework, modules, 570 00:29:10,872 --> 00:29:12,987 plugins. And I'm quite curious when you 571 00:29:12,987 --> 00:29:14,623 mentioned just now, right, why ... 572 00:29:14,623 --> 00:29:18,049 Like the story why you actually started BMad method. 573 00:29:18,049 --> 00:29:21,325 Maybe if you can share with me, how did it started in the 574 00:29:21,325 --> 00:29:23,045 first place? Like what kind of problems do 575 00:29:23,045 --> 00:29:24,890 you see? Yeah. 576 00:29:25,550 --> 00:29:28,255 That's a good question. Thanks for asking that, because 577 00:29:28,255 --> 00:29:32,215 I think it helps explain where it is also now. 578 00:29:33,015 --> 00:29:37,519 Henry, a little over a year ago, I work at a company called 579 00:29:37,519 --> 00:29:39,995 Extend. It's a startup in San Francisco. 580 00:29:40,475 --> 00:29:42,375 It's a, you know, smallish company. 581 00:29:42,425 --> 00:29:46,325 And I started out as a staff engineer and now, you know, I 582 00:29:46,325 --> 00:29:48,575 manage teams across a product engineering work. 583 00:29:49,085 --> 00:29:53,788 And we were using Copilot, and if you remember, you know, 584 00:29:53,788 --> 00:29:58,215 Copilot like two years ago, it was basically a very good 585 00:29:58,215 --> 00:30:00,774 autocomplete, right? Like at the time it seemed like 586 00:30:00,774 --> 00:30:03,353 magic. But a little over a year ago, 587 00:30:03,353 --> 00:30:04,655 you know, Cursor's started getting very big. 588 00:30:04,655 --> 00:30:09,885 I assume you're familiar with a lot of these tools, Henry? 589 00:30:10,749 --> 00:30:12,009 Yep. Yeah. 590 00:30:12,009 --> 00:30:14,034 So Cursor came out. I was working in this 591 00:30:14,034 --> 00:30:16,255 organization and we were looking at other tools to 592 00:30:16,255 --> 00:30:19,119 replace Copilot. People loved Copilot. 593 00:30:19,179 --> 00:30:20,949 You know, we'd been using it about a year. 594 00:30:21,819 --> 00:30:23,619 And it was great for autocomplete. 595 00:30:23,619 --> 00:30:26,007 You could like show it a function and it would help you 596 00:30:26,007 --> 00:30:28,919 like rework the function. And it did seem like magic at 597 00:30:28,919 --> 00:30:32,459 the time. When ChatGPT like really started 598 00:30:32,459 --> 00:30:36,654 getting big though, like when that 3.1 came out, we started 599 00:30:36,654 --> 00:30:40,854 seeing engineers using ChatGPT instead of Stack Overflow, 600 00:30:40,854 --> 00:30:43,383 right? And it was getting good at 601 00:30:43,383 --> 00:30:46,479 answering questions. And then Cursor came out. 602 00:30:46,479 --> 00:30:49,681 And so we were talking about it in the organization and a few 603 00:30:49,681 --> 00:30:52,804 of us started playing with Cursor on the side and we were 604 00:30:52,804 --> 00:30:55,908 enamored with the agent mode. It was amazing! 605 00:30:55,918 --> 00:30:58,566 Even for the time, it was nowhere near as powerful as 606 00:30:58,566 --> 00:31:02,406 any of the tools we have now. It seems like generations ago. 607 00:31:02,406 --> 00:31:04,746 It's hard to believe it was only a little over a year ago. 608 00:31:05,416 --> 00:31:07,926 But we rolled it out across the company. 609 00:31:08,886 --> 00:31:13,475 Now as I was managing my teams, I saw that they picked up 610 00:31:13,475 --> 00:31:18,206 Cursor and they liked it because it had like a very good way to 611 00:31:18,206 --> 00:31:19,866 do autocomplete on multiple lines. 612 00:31:19,866 --> 00:31:22,706 But people were still typing code. 613 00:31:23,116 --> 00:31:27,572 But a few of us in the company, mostly working on like smaller 614 00:31:27,572 --> 00:31:31,606 tools or utilities, we're using this agent mode. 615 00:31:32,046 --> 00:31:35,626 And I was seeing this disconnect between the engineers on the 616 00:31:35,626 --> 00:31:39,166 teams over the months, the few months of us rolling out Cursor. 617 00:31:39,876 --> 00:31:42,606 Nobody was really using the agent mode to talk to it and 618 00:31:42,606 --> 00:31:44,106 actually write the code for them. 619 00:31:44,856 --> 00:31:47,642 But I was seeing other people in the company, like a few people 620 00:31:47,642 --> 00:31:50,922 like myself and a few others, we were able to get it to do a 621 00:31:50,922 --> 00:31:54,630 little bit more. But nothing big or grand that it 622 00:31:54,630 --> 00:31:56,796 wasn't working very well in brownfield. 623 00:31:56,796 --> 00:31:59,452 It struggled a lot. It would run out of context 624 00:31:59,452 --> 00:32:04,135 really quick. The disconnect from what I was 625 00:32:04,135 --> 00:32:07,884 seeing on YouTube, Henry, though, was these channels 626 00:32:07,884 --> 00:32:13,716 showing Cursor and these tools and these amazing claims. 627 00:32:14,216 --> 00:32:17,696 You can have no experience, you don't even need to be a 628 00:32:17,696 --> 00:32:21,281 software developer, and overnight you can build your 629 00:32:21,289 --> 00:32:25,731 million dollar SaaS application. And I kept seeing these videos 630 00:32:25,739 --> 00:32:29,071 and they were impressive when you would watch them, but I'm 631 00:32:29,071 --> 00:32:32,061 like that seems like a little bit farfetched. 632 00:32:32,061 --> 00:32:35,245 I don't know about that. But there was this wide 633 00:32:35,245 --> 00:32:37,261 disconnect and I kept seeing those videos. 634 00:32:37,891 --> 00:32:40,467 And then at the other end of the spectrum, I was seeing 635 00:32:40,467 --> 00:32:43,335 engineers in my company not really wanting anything to do 636 00:32:43,335 --> 00:32:45,939 with the agent mode. It just seemed like a fad or 637 00:32:45,947 --> 00:32:48,711 maybe you could use it to ask it a question like ChatGPT. 638 00:32:49,101 --> 00:32:53,181 But there was this chasm. So I became obsessed. 639 00:32:53,885 --> 00:32:56,105 I'm like there has to be more to it. 640 00:32:56,105 --> 00:33:00,935 So as a leader of my team, I wanted to figure out is there 641 00:33:00,935 --> 00:33:03,905 something to this agent mode because Cursor's making these 642 00:33:03,905 --> 00:33:06,581 claims. Sonnet 3.5 is supposed, at the 643 00:33:06,581 --> 00:33:08,975 time, was supposed to be this amazing new thing. 644 00:33:09,769 --> 00:33:11,079 But everybody was using it like autocomplete. 645 00:33:11,509 --> 00:33:15,859 So I started, before there was a term for vibe coding, I was 646 00:33:15,859 --> 00:33:20,939 basically vibe coding at night. I would watch Netflix and just 647 00:33:20,939 --> 00:33:24,113 vibe and it was fun. And Cursor at the time was very 648 00:33:24,113 --> 00:33:25,717 slow. You would hit this thing, they 649 00:33:25,717 --> 00:33:27,889 would call Slow Mode. I don't know if everybody knows 650 00:33:27,889 --> 00:33:30,289 Slow Mode, but after you would use up your tokens. 651 00:33:30,939 --> 00:33:33,905 It would like make you wait for 15 minutes before it would 652 00:33:33,905 --> 00:33:35,641 start processing. So it was great for watching 653 00:33:35,641 --> 00:33:37,946 TV on the couch. Every once in a while you would 654 00:33:37,946 --> 00:33:39,419 look back and give it another command. 655 00:33:40,099 --> 00:33:43,519 But that's when I started seeing the vibe coding curse, right? 656 00:33:43,559 --> 00:33:47,029 I had an idea for a pretty complex application at the time. 657 00:33:48,014 --> 00:33:51,542 And I gave myself the challenge, I'm gonna figure out how to 658 00:33:51,542 --> 00:33:54,686 get this agent to build this complicated application, 659 00:33:54,686 --> 00:33:57,479 multiple stacks, front-end, back-end, database. 660 00:33:58,359 --> 00:34:01,699 And get to do it without me typing any code. 661 00:34:03,139 --> 00:34:05,299 And I would do this over and over again, Henry. 662 00:34:05,869 --> 00:34:11,539 And I started to kind of realize like, oh, it gets this far and 663 00:34:11,539 --> 00:34:14,920 it starts breaking. Or it starts randomly doing a 664 00:34:14,920 --> 00:34:18,694 test, it fails, and it says, well, maybe if I just throw out 665 00:34:18,694 --> 00:34:21,244 the database and put in another one, it'll work better. 666 00:34:21,244 --> 00:34:22,987 And if you're not paying attention, it'll do these 667 00:34:22,987 --> 00:34:26,692 crazy things. So I said, well, if I have this 668 00:34:26,692 --> 00:34:30,592 file that says we are using these libraries and frameworks 669 00:34:30,592 --> 00:34:34,414 and only use those, I started to notice that, oh, now I can 670 00:34:34,414 --> 00:34:36,922 get a little bit farther where it doesn't start doing those 671 00:34:36,922 --> 00:34:40,505 still silly things. And then I started to realize 672 00:34:40,505 --> 00:34:45,360 like, oh, if I actually create a prompt file, I don't have to 673 00:34:45,360 --> 00:34:48,994 keep repeating myself, and it gets a little bit farther. 674 00:34:50,164 --> 00:34:54,293 Oh, what if I take a PRD, and have it lean so it knows all 675 00:34:54,293 --> 00:34:57,094 the requirements upfront, and it gets a little bit farther. 676 00:34:57,094 --> 00:35:00,207 And it's just, it kind of evolved out of this loop that 677 00:35:00,207 --> 00:35:03,242 I was doing. And I got to a point where I 678 00:35:03,242 --> 00:35:05,944 could actually build this pretty complex application. 679 00:35:06,094 --> 00:35:09,550 It was exciting. And again, this is like a year 680 00:35:09,550 --> 00:35:13,016 and a half ago now with old Sonnet 3.5 in Cursor with this 681 00:35:13,016 --> 00:35:15,424 limited context window and it was doing this thing. 682 00:35:15,424 --> 00:35:18,961 And that's another thing is with Cursor, you start to realize, 683 00:35:18,961 --> 00:35:21,964 wow, this thing starts hallucinating really quick. 684 00:35:21,964 --> 00:35:24,652 And this is not like a knock on Cursor right now, but this is 685 00:35:24,652 --> 00:35:27,064 just the state of things at the time. 686 00:35:27,364 --> 00:35:30,624 So you had to start coming up with your own ways of like, 687 00:35:30,624 --> 00:35:34,599 oh, if we have files and we break it down into smaller 688 00:35:34,607 --> 00:35:36,454 segments, it actually does better. 689 00:35:37,384 --> 00:35:38,944 So why don't we call that a user story? 690 00:35:39,064 --> 00:35:43,243 And so thing kind of started forming where basically I'm 691 00:35:43,243 --> 00:35:47,274 taking what I've done for the last 15 years with Agile and 692 00:35:47,274 --> 00:35:51,078 thinking in this Agile fashion. I said it actually works really 693 00:35:51,078 --> 00:35:54,574 well to use that as a thought framework, as a model to kind of 694 00:35:54,574 --> 00:35:56,104 apply it. Here we're doing the same thing. 695 00:35:56,344 --> 00:35:59,054 We're talking about a problem, we're doing a product inception. 696 00:35:59,584 --> 00:36:02,284 We're breaking it down, we're doing a solution design, and 697 00:36:02,284 --> 00:36:04,255 we're coming up with stories. Again, context engineering, 698 00:36:04,263 --> 00:36:07,162 there wasn't a name for it yet. Spec engineering, there wasn't 699 00:36:07,162 --> 00:36:09,744 a name for it yet. So I was so excited. 700 00:36:09,744 --> 00:36:12,304 I'm like, this is amazing. I can take this back to my team. 701 00:36:13,849 --> 00:36:15,529 It is gonna revolutionize everything. 702 00:36:15,869 --> 00:36:18,529 I brought it back and nothing changed. 703 00:36:18,919 --> 00:36:20,609 So that was the surprising thing. 704 00:36:20,939 --> 00:36:24,439 But I put it online and it did start to, you know, catch fire 705 00:36:24,439 --> 00:36:26,809 there with that first one. I just put it up there to be 706 00:36:26,809 --> 00:36:28,747 this thought framework. I never thought it would become 707 00:36:28,747 --> 00:36:30,139 anything it is like today, Henry. 708 00:36:30,589 --> 00:36:35,359 And it started resonating a lot with people online. 709 00:36:35,749 --> 00:36:39,049 Product managers started seeing this and saying, wow, this can 710 00:36:39,049 --> 00:36:42,559 actually help me produce better PRDs than I ever had before. 711 00:36:42,559 --> 00:36:46,360 Because as a professional product manager, it's 712 00:36:46,360 --> 00:36:49,689 reminding me I forget these things and it's actually 713 00:36:49,697 --> 00:36:52,877 helping ask me the questions of. So it's funny because product 714 00:36:52,885 --> 00:36:55,339 managers started reaching out to me saying they were using it. 715 00:36:55,639 --> 00:36:58,449 And then people that have never written code before were 716 00:36:58,449 --> 00:37:01,174 starting to use it. And it's like they're learning 717 00:37:01,174 --> 00:37:04,179 like how Agile works and how industry works and people 718 00:37:04,179 --> 00:37:07,757 started learning from it. But I still wasn't seeing it 719 00:37:07,765 --> 00:37:11,451 catch fire in my own organization which was the whole 720 00:37:11,451 --> 00:37:15,296 point. So I did this technique, Henry. 721 00:37:15,416 --> 00:37:17,909 If I can go into maybe another little story, 'cause I think 722 00:37:17,909 --> 00:37:20,996 this is useful for a lot of tech leaders right now because you 723 00:37:20,996 --> 00:37:23,546 might be seeing this in your own organizations. 724 00:37:24,386 --> 00:37:27,002 And since I've done this, I've actually talked to a lot of 725 00:37:27,002 --> 00:37:30,356 companies and helped transform my own organization and others. 726 00:37:30,356 --> 00:37:33,056 So there is a success story here at the end. 727 00:37:34,166 --> 00:37:39,746 But if you just give engineers this thing and say, hey guys, 728 00:37:39,746 --> 00:37:43,552 I figured it out. If you just do a PRD, and do 729 00:37:43,552 --> 00:37:45,566 this, and do this, it's gonna give you this magic. 730 00:37:45,566 --> 00:37:49,946 It's gonna work. Engineers, I think, have egos. 731 00:37:49,946 --> 00:37:52,586 We all have egos, and it's like, no, I'll figure it out myself. 732 00:37:52,826 --> 00:37:56,857 And they don't really come to the same conclusions. 733 00:37:57,905 --> 00:38:00,605 And so they kind of will still feel like, yeah, the agent mode 734 00:38:00,605 --> 00:38:02,750 is good, but I'm still gonna go back to doing it this other 735 00:38:02,750 --> 00:38:03,845 way. I'm faster. 736 00:38:05,220 --> 00:38:08,417 What I did with every — and by the way, it's not for lack of 737 00:38:08,417 --> 00:38:09,763 trying. Like I would tell people to try 738 00:38:09,763 --> 00:38:12,260 it. They would try the agent mode, 739 00:38:12,268 --> 00:38:14,970 especially in legacy projects where you have big code bases. 740 00:38:15,790 --> 00:38:17,810 It's like, nah, it's still not figuring it out. 741 00:38:17,870 --> 00:38:19,850 It's getting it wrong. It's messing up the code. 742 00:38:20,090 --> 00:38:23,066 I'll just go in and fix it. And you kind of fall back 743 00:38:23,066 --> 00:38:25,280 because you're under the demands of your current sprint. 744 00:38:25,280 --> 00:38:27,500 You know, your team probably wants you to get things done. 745 00:38:28,160 --> 00:38:31,355 And a lot of times you don't feel like you have the space 746 00:38:31,355 --> 00:38:33,338 to really experiment. So you just kind of fall back on 747 00:38:33,338 --> 00:38:34,430 what you're, what you need to do. 748 00:38:35,527 --> 00:38:41,403 We had done AI sprints before in our company where the goal was 749 00:38:41,403 --> 00:38:45,453 let's try to come up with an agent or let's try to come up 750 00:38:45,453 --> 00:38:48,354 with some way to put AI or everybody come up with a 751 00:38:48,354 --> 00:38:51,056 creative use of AI. And those were pretty 752 00:38:51,056 --> 00:38:53,144 successful. And we did some hackathons and 753 00:38:53,144 --> 00:38:56,328 it led to some good ideas within our company, even, you 754 00:38:56,328 --> 00:38:59,689 know, for our products. But I said, what if instead of 755 00:38:59,689 --> 00:39:03,658 an AI hackathon where the goal is to come up with some new AI 756 00:39:03,658 --> 00:39:07,584 feature for the product, I had everybody do what I did for 757 00:39:07,584 --> 00:39:12,682 months in a two week sprint. So with my team, I took my team, 758 00:39:12,682 --> 00:39:15,653 Henry, and I said, we work in two week sprints. 759 00:39:15,653 --> 00:39:17,761 If people don't know what a sprint is, it's like a 760 00:39:17,761 --> 00:39:21,239 two-week block of work where the team commits to so much work 761 00:39:21,239 --> 00:39:24,203 at the beginning. I assume by now everybody knows 762 00:39:24,203 --> 00:39:27,335 by Agile, but just if people don't, right, you commit to a 763 00:39:27,335 --> 00:39:29,977 group of work and by the end of the two weeks, the team 764 00:39:29,977 --> 00:39:33,213 really tries to push and get that two week work done. 765 00:39:33,823 --> 00:39:36,383 I said for this sprint, what we're gonna do is everybody's 766 00:39:36,383 --> 00:39:40,103 only gonna have one story. Each will be assigned a story. 767 00:39:40,103 --> 00:39:41,963 That's not something we normally do in Agile. 768 00:39:41,963 --> 00:39:44,220 We want it to be more organic, but I said, in this case, 769 00:39:44,220 --> 00:39:45,763 everybody will have one story to do. 770 00:39:46,363 --> 00:39:49,195 This is a piece of work that would normally take you two to 771 00:39:49,195 --> 00:39:53,093 three days to do. But the only caveat is everybody 772 00:39:53,093 --> 00:39:57,113 has to use the agent mode. You cannot type any code. 773 00:39:57,563 --> 00:40:02,123 And if you get it done, I want you to do it over again and keep 774 00:40:02,123 --> 00:40:07,169 doing it and refining it. This was the most transformative 775 00:40:07,169 --> 00:40:11,295 sprint, because first of all, it gave everybody the permission 776 00:40:11,295 --> 00:40:15,639 to not have to worry about getting the thing done and 777 00:40:15,639 --> 00:40:20,606 moving on to the next thing. And even in a normal sprint, if 778 00:40:20,606 --> 00:40:24,482 you tell your developers and your teams, don't worry, like 779 00:40:24,482 --> 00:40:28,226 learn how to use these tools, take some time to do it, a lot 780 00:40:28,226 --> 00:40:31,839 of developers will still self internalize that, yeah, well I 781 00:40:31,839 --> 00:40:34,607 can take the time. I really still need to get the 782 00:40:34,607 --> 00:40:36,533 work done. And that takes the priority. 783 00:40:36,533 --> 00:40:41,293 So by taking the pressure off for two weeks, it gave everybody 784 00:40:41,538 --> 00:40:44,868 this permission to fail. The permission to fail, right? 785 00:40:44,868 --> 00:40:47,538 The permission to take a chance and see what happens. 786 00:40:48,318 --> 00:40:50,552 Henry, when I tell you this was the most transformative thing 787 00:40:50,552 --> 00:40:53,298 I've ever seen, like it's not an understatement. 788 00:40:54,108 --> 00:40:56,848 People started coming to the same conclusions. 789 00:40:58,408 --> 00:41:00,978 And basically inventing spec-driven development. 790 00:41:01,128 --> 00:41:03,021 You know, that's why it's funny because all these frameworks 791 00:41:03,021 --> 00:41:05,630 are coming out and it probably seems across the industry that 792 00:41:05,630 --> 00:41:08,088 people are always kind of coming up with the same ideas. 793 00:41:08,538 --> 00:41:10,793 Because if you actually go through this and do this, you 794 00:41:10,793 --> 00:41:12,528 start coming to your own conclusions. 795 00:41:12,588 --> 00:41:17,148 Oh, I don't have to repeat myself to the agent every time. 796 00:41:17,153 --> 00:41:20,778 I can just put it in a file and give it that information, and I 797 00:41:20,778 --> 00:41:24,054 don't have to explain to it, hey, I'm a X, Y, Z, and we're 798 00:41:24,054 --> 00:41:25,848 doing this thing and we have to do this. 799 00:41:26,628 --> 00:41:29,262 You might not have the words for it, but you'll come to all 800 00:41:29,262 --> 00:41:32,653 these same conclusions. Not only did people start 801 00:41:32,653 --> 00:41:35,474 figuring this out, but they started figuring out like, oh, 802 00:41:35,474 --> 00:41:38,648 we can save these prompts and it can help me do the same thing 803 00:41:38,648 --> 00:41:41,595 across all these code bases. And I can create an agent 804 00:41:41,595 --> 00:41:44,213 persona and they, it can help me do these things. 805 00:41:44,873 --> 00:41:49,132 Oh, I can give it a task list. So whether you call it a PRD or 806 00:41:49,132 --> 00:41:53,073 a story breakdown or a list of things to do, it'll do better. 807 00:41:53,343 --> 00:41:55,791 And oh, it failed for this reason, but when I start over, 808 00:41:55,791 --> 00:41:58,758 if it has this rule or this thing, it will now get a 809 00:41:58,758 --> 00:42:01,309 little bit farther. Not only did people figure out 810 00:42:01,317 --> 00:42:04,675 how to get their stories and use agent modes, it was 811 00:42:04,675 --> 00:42:08,013 transformative in ways I did not even expect. 812 00:42:08,463 --> 00:42:13,632 All of a sudden people started saying, why do we need to do 813 00:42:13,632 --> 00:42:16,893 error triage manually every morning and look at the logs? 814 00:42:17,333 --> 00:42:20,645 We can now create a prompt that does that for us and pulls 815 00:42:20,645 --> 00:42:22,843 this information. Or, oh, there's this thing 816 00:42:22,843 --> 00:42:24,703 called MCP. Why don't we put an MCP in 817 00:42:24,703 --> 00:42:27,869 front of this tool? Before that, for months I was 818 00:42:27,869 --> 00:42:30,953 trying to get people to like, let's be creative guys and just 819 00:42:30,953 --> 00:42:33,575 think of these things. Whatever it was about this 820 00:42:33,575 --> 00:42:36,453 exercise, not only did people start thinking about how to work 821 00:42:36,453 --> 00:42:38,915 with agent mode, but they started looking at all these 822 00:42:38,915 --> 00:42:41,909 opportunities to start automating their job away and 823 00:42:41,909 --> 00:42:45,471 it became like wildfire. And this is even before like 824 00:42:45,471 --> 00:42:49,165 there was a skills explosion or before the tools were good as 825 00:42:49,165 --> 00:42:52,625 they are now. So if I could give one piece of 826 00:42:52,625 --> 00:42:55,785 advice to every company, if you wanna transform your engineering 827 00:42:55,785 --> 00:42:59,339 team and get them to start using these tools, give them 828 00:42:59,339 --> 00:43:05,178 that space to do it. After that — this is no 829 00:43:05,178 --> 00:43:09,694 exaggeration — a few engineers went from never using the agent 830 00:43:09,694 --> 00:43:14,241 mode to using it 100% of the time, where they've taken it 831 00:43:14,241 --> 00:43:16,935 as their challenge to do everything with the agent mode. 832 00:43:17,415 --> 00:43:20,985 And this is not simple greenfield hobby projects. 833 00:43:21,225 --> 00:43:24,009 Like a lot of people think this only works with a greenfield 834 00:43:24,009 --> 00:43:27,323 or simple projects. This was people working in these 835 00:43:27,323 --> 00:43:32,070 legacy, you know, services that have existed for 5, 6, 7 836 00:43:32,070 --> 00:43:37,095 years using the agent mode. Now since then, this has spread 837 00:43:37,095 --> 00:43:40,515 across the organization and it's been transformative across 838 00:43:40,515 --> 00:43:43,815 product engineering and architecture and design. 839 00:43:44,315 --> 00:43:49,590 And, yeah, it's just been magic. Like I can't explain how useful 840 00:43:49,590 --> 00:43:52,136 that was. So this all evolved from, you 841 00:43:52,136 --> 00:43:55,112 know, trying to do this myself, coming up with this 842 00:43:55,112 --> 00:43:57,982 thought framework, which was the BMad method, and then using 843 00:43:57,982 --> 00:44:01,910 it as a teaching tool to kind of have people do the same 844 00:44:01,910 --> 00:44:05,031 exercise, come to their own conclusions which funny enough 845 00:44:05,031 --> 00:44:07,316 were basically the same conclusions, right? 846 00:44:07,316 --> 00:44:11,571 Having a plan, giving it small bite-sized things, giving it 847 00:44:11,571 --> 00:44:13,278 context. Again, we didn't have the name 848 00:44:13,278 --> 00:44:16,156 for it at the time but it became context engineering. 849 00:44:16,526 --> 00:44:20,376 So let me pause there, Henry, 'cause that's kind of a long 850 00:44:20,376 --> 00:44:24,256 story. But it's just, it was an amazing 851 00:44:24,256 --> 00:44:26,796 experience and it's transformed things. 852 00:44:28,025 --> 00:44:29,417 Yeah. Thanks for sharing such a 853 00:44:29,417 --> 00:44:31,519 beautiful story. I think I can see when you 854 00:44:31,519 --> 00:44:33,853 look back, right? So it's kind of like the 855 00:44:33,861 --> 00:44:36,081 transformational impact that I think you must be proud of, 856 00:44:36,081 --> 00:44:38,389 right? But like which is also a 857 00:44:38,389 --> 00:44:40,501 reminder for, you know, leaders here, listeners here, 858 00:44:40,501 --> 00:44:42,681 right? So I think organizations that 859 00:44:42,681 --> 00:44:45,745 thrive is organization that learns a lot, right? 860 00:44:45,775 --> 00:44:49,240 And give the space for people, you know, to fail, experiment 861 00:44:49,240 --> 00:44:51,827 without kind of like the pressure of always delivering 862 00:44:51,827 --> 00:44:54,805 all the time, right? So I think that's a actually a 863 00:44:54,805 --> 00:44:57,130 very good story. Speaking about, you know, this, 864 00:44:57,130 --> 00:44:59,881 the thing that you mentioned, you know, you can work with 865 00:44:59,881 --> 00:45:03,199 legacy code base, you can work, you know, or purely 100% 866 00:45:03,199 --> 00:45:04,999 agentic. What kind of impact, maybe if 867 00:45:04,999 --> 00:45:07,360 you can quantify a little bit to tease, you know, the 868 00:45:07,360 --> 00:45:10,061 listeners here to give it a try. Like what kind of impact have 869 00:45:10,061 --> 00:45:12,115 you seen? Is it like productivity impact, 870 00:45:12,115 --> 00:45:14,125 the number of people becomes lesser? 871 00:45:14,215 --> 00:45:15,895 Or what kind of impact do you see? 872 00:45:16,135 --> 00:45:18,985 Maybe not just from your team, but other users as well? 873 00:45:20,176 --> 00:45:23,370 Yeah, well let me tie it to the organization first and then 874 00:45:23,370 --> 00:45:25,766 I'll talk about users of the BMad method. 875 00:45:25,766 --> 00:45:29,290 But in the organization since we did that, you know, that was 876 00:45:29,290 --> 00:45:32,013 probably six, six to eight months ago that we did it. 877 00:45:32,023 --> 00:45:33,893 Maybe six months ago, nine months ago. 878 00:45:35,603 --> 00:45:38,393 A lot has changed since then. The tooling has gotten better. 879 00:45:39,413 --> 00:45:43,018 Skills have exploded. We started to realize that this 880 00:45:43,018 --> 00:45:48,283 is not only great for the engineers, but can we help 881 00:45:48,283 --> 00:45:51,827 accelerate product? And you know, again, the BMad 882 00:45:51,827 --> 00:45:54,452 method I told you about, how like product managers reach out 883 00:45:54,452 --> 00:45:57,083 about how it can help them produce better PRDs. 884 00:45:57,353 --> 00:45:59,846 So, you know, training and realizing these tools, like 885 00:45:59,846 --> 00:46:03,403 now this is common sense, but 6-8 months ago we were 886 00:46:03,411 --> 00:46:06,377 realizing that these tools are actually really good for 887 00:46:06,377 --> 00:46:07,963 everybody. So helping share this 888 00:46:07,963 --> 00:46:10,778 information and teaching everybody to start working this 889 00:46:10,778 --> 00:46:13,154 way. And you know what is now a 890 00:46:13,154 --> 00:46:15,687 pretty common term, but we started calling it like thinking 891 00:46:15,687 --> 00:46:19,507 AI native, it doesn't mean like what is the next AI product we 892 00:46:19,507 --> 00:46:22,832 can build or bolt onto our product but how can we start 893 00:46:22,832 --> 00:46:24,743 using AI in all of our workflows? 894 00:46:24,743 --> 00:46:29,711 As a tech leader or a manager, you can use AI to help you 895 00:46:29,711 --> 00:46:33,598 look at data and create feedback loops to better understand 896 00:46:33,598 --> 00:46:37,481 where your team is operating. So, you know, you can use it for 897 00:46:37,481 --> 00:46:39,203 so much more than coding and engineering. 898 00:46:39,203 --> 00:46:42,616 So we started realizing all of these things and there's been 899 00:46:42,616 --> 00:46:47,465 this huge acceleration. There's been a lot of studies, 900 00:46:47,465 --> 00:46:50,785 you know, and I think the general consensus is you can see 901 00:46:50,785 --> 00:46:53,243 productivity gains from anywhere from, some people say 902 00:46:53,253 --> 00:46:56,423 10-20%, some people say even more. 903 00:46:57,518 --> 00:47:00,668 As software engineers, we bring a lot more to the table and not 904 00:47:00,668 --> 00:47:03,776 everything is pounding out code. When you're, you know, a senior 905 00:47:03,776 --> 00:47:07,194 engineer or a junior engineer or anywhere in there, your job 906 00:47:07,194 --> 00:47:10,093 is not to type code all day. And sometimes we lose sight of 907 00:47:10,093 --> 00:47:12,864 that. So AI will help you speed that 908 00:47:12,864 --> 00:47:15,510 aspect up. But it can also help you speed 909 00:47:15,518 --> 00:47:20,128 up the other aspects of the job. The other 60-70% of the time 910 00:47:20,136 --> 00:47:23,572 as an engineer, you're having discussions, you're thinking, 911 00:47:23,580 --> 00:47:26,288 you're communicating, you're having confidence in explaining 912 00:47:26,288 --> 00:47:29,228 ideas to different people in different ways. 913 00:47:29,618 --> 00:47:33,745 Those things don't go away, and those things are still the very 914 00:47:33,745 --> 00:47:36,008 human aspect of what we bring to the table. 915 00:47:37,118 --> 00:47:39,590 It's funny, I just had this conversation a week or two ago 916 00:47:39,590 --> 00:47:43,918 because we are having such acceleration with AI in our 917 00:47:43,918 --> 00:47:47,290 organization and people are getting better and better using 918 00:47:47,290 --> 00:47:50,085 at it. Now everybody is using agent 919 00:47:50,085 --> 00:47:51,698 mode. Everybody is using skills. 920 00:47:51,698 --> 00:47:56,798 Everybody's producing these amazing skills to do everything. 921 00:47:56,798 --> 00:48:00,358 You know, you can spit up a new dashboard to monitor something 922 00:48:00,358 --> 00:48:03,478 in a day now, what used to, you know, take months of planning 923 00:48:03,486 --> 00:48:05,318 and, you know, what tool are we gonna use. 924 00:48:06,008 --> 00:48:09,803 And some developers on my team have started to ask the 925 00:48:09,803 --> 00:48:13,120 question, well, it's fun and it's exciting, but I'm also 926 00:48:13,128 --> 00:48:17,491 starting to lose my, not lose my identity a little bit, but a lot 927 00:48:17,491 --> 00:48:20,740 of developers, their identity is tied to how good or how fast 928 00:48:20,740 --> 00:48:24,041 they can type the code. But what I really, you know, try 929 00:48:24,041 --> 00:48:26,320 to share with people, and I strongly believe this, and I 930 00:48:26,320 --> 00:48:29,900 think a lot of them are also kind of feeling this is, it's 931 00:48:29,900 --> 00:48:33,456 just another abstraction, right? It's like the switch from 932 00:48:33,464 --> 00:48:36,158 Assembly language to, I'll just say to TypeScript, right? 933 00:48:36,158 --> 00:48:38,618 But this feels like almost that much of a gap. 934 00:48:39,098 --> 00:48:40,718 You're able to step away a little bit. 935 00:48:40,718 --> 00:48:44,898 You're almost becoming more of a leader of AI. 936 00:48:44,928 --> 00:48:48,375 You're looking at it, you're planning, you are thinking 937 00:48:48,375 --> 00:48:53,145 about a bigger picture. And I think it really frees you 938 00:48:53,145 --> 00:48:55,763 up to focus on more interesting challenges. 939 00:48:55,763 --> 00:48:58,441 You know, like how to do the for loop or how, you know, like 940 00:48:58,441 --> 00:49:00,383 what is the best way to structure this code. 941 00:49:00,743 --> 00:49:04,073 It's still important, and as an experienced engineer it's still 942 00:49:04,073 --> 00:49:05,593 important to keep those things in mind. 943 00:49:06,183 --> 00:49:09,108 But this also frees you up to think about it in a different 944 00:49:09,108 --> 00:49:12,543 way and plan. And as the unit of work changes 945 00:49:12,543 --> 00:49:15,588 for developers, 'cause I will say this is something that will 946 00:49:15,588 --> 00:49:20,316 happen and is happening now. If the unit of work used to be 947 00:49:20,316 --> 00:49:24,107 the user story, I really believe very soon the unit of 948 00:49:24,107 --> 00:49:27,963 work is become, is gonna become the feature or the Epic. 949 00:49:29,343 --> 00:49:32,543 Different teams use different phrases, but what I mean by an 950 00:49:32,543 --> 00:49:36,733 epic is a collection of related user stories to build a whole 951 00:49:36,733 --> 00:49:40,807 feature. We're at a point now where a 952 00:49:40,815 --> 00:49:44,483 single developer can definitely finish a user story much faster. 953 00:49:44,483 --> 00:49:48,133 A user story that used to take a week, maybe it takes hours or 954 00:49:48,133 --> 00:49:52,258 a day now. And that is realistically true, 955 00:49:52,258 --> 00:49:57,359 that there is faster velocity. When the work is defined, the 956 00:49:57,367 --> 00:50:01,061 output is gonna be quicker. And as we develop better tools, 957 00:50:01,061 --> 00:50:04,721 it will get faster and faster. We can automate code reviews, we 958 00:50:04,721 --> 00:50:09,185 can, you know, I mean, that's a whole other topic, but what I 959 00:50:09,185 --> 00:50:13,547 wanna say is now the developer can be responsible for a whole 960 00:50:13,547 --> 00:50:16,253 epic and maybe even deliver that epic themselves. 961 00:50:16,703 --> 00:50:20,163 So what I see is this transformation coming where 962 00:50:20,163 --> 00:50:23,353 now the developer can be more responsible, or almost 963 00:50:23,353 --> 00:50:26,831 everybody, you know, almost everybody can be a tech lead at 964 00:50:26,831 --> 00:50:30,527 this point in the sense that we call them epic champions or 965 00:50:30,527 --> 00:50:33,243 feature champions. And that's a good way to develop 966 00:50:33,243 --> 00:50:35,343 all the people on your team, right? 967 00:50:35,343 --> 00:50:38,907 Regardless of their level, give them responsibility for an 968 00:50:38,907 --> 00:50:41,428 epic. But now by giving them 969 00:50:41,428 --> 00:50:44,475 responsibility for an epic, they can actually work closer 970 00:50:44,475 --> 00:50:47,823 with the product manager or whoever's defining the epic. 971 00:50:48,453 --> 00:50:52,233 They should really be more involved in that conversation, 972 00:50:52,233 --> 00:50:56,013 understanding it, being able to explain it, and really 973 00:50:56,013 --> 00:50:58,713 implementing the whole thing almost themselves now. 974 00:50:58,713 --> 00:51:03,173 So we are in a transitional period where this larger body 975 00:51:03,173 --> 00:51:06,273 of work will become the unit of work. 976 00:51:06,753 --> 00:51:09,723 But I think it's exciting because it really moves you up 977 00:51:09,723 --> 00:51:12,161 the funnel. Instead of just being handed a 978 00:51:12,161 --> 00:51:16,188 single task where maybe you're not really fully aware of how it 979 00:51:16,188 --> 00:51:19,682 fits into the whole bigger picture, we are being freed up 980 00:51:19,682 --> 00:51:23,052 as engineers now to actually be able to start focusing on 981 00:51:23,052 --> 00:51:24,498 that. I think that's exciting. 982 00:51:24,498 --> 00:51:30,513 Because remember, our goal as engineers is not to, you know, 983 00:51:30,513 --> 00:51:33,249 create the most — I mean, sometimes optimization 984 00:51:33,249 --> 00:51:35,523 matters, right? But it's not the elegant for 985 00:51:35,531 --> 00:51:38,037 loop that you wrote, right? It's not gonna matter six months 986 00:51:38,037 --> 00:51:39,888 from now. It's the features that you're 987 00:51:39,896 --> 00:51:42,663 building to deliver value for the company that you work for. 988 00:51:43,113 --> 00:51:46,206 And so you actually now have the ability as an engineer to step 989 00:51:46,206 --> 00:51:49,819 back a little bit, orchestrate these things to do the work 990 00:51:49,819 --> 00:51:52,260 with you. Still treating it as a 991 00:51:52,260 --> 00:51:54,753 facilitator. Still, you know, keeping your 992 00:51:54,753 --> 00:51:58,316 own discernment and taste in the picture, and bringing your 993 00:51:58,316 --> 00:52:01,018 experience to the table to guide the AI, but you get to 994 00:52:01,018 --> 00:52:04,201 focus on this other stuff. So framing it that way, I think 995 00:52:04,201 --> 00:52:06,941 people are very excited and starting to see that, yeah, this 996 00:52:06,941 --> 00:52:09,387 is good. It does free me up to kind of 997 00:52:09,387 --> 00:52:11,958 look at these bigger picture things and feel like you're 998 00:52:11,958 --> 00:52:13,913 actually having a bigger impact on the company. 999 00:52:13,913 --> 00:52:17,963 Because now you're delivering whole features at a time that 1000 00:52:17,963 --> 00:52:21,617 are delivering value, instead of delivering small pieces that 1001 00:52:21,617 --> 00:52:24,758 might take three months to actually release to production 1002 00:52:24,758 --> 00:52:27,097 and give value. So it's an exciting time to be 1003 00:52:27,097 --> 00:52:31,073 an engineer. I wanna very briefly though, say 1004 00:52:31,073 --> 00:52:35,838 also, for the BMad method itself, I have heard from many 1005 00:52:35,838 --> 00:52:38,268 people that have never done software engineering, so go 1006 00:52:38,268 --> 00:52:42,138 back to what I said was the scammy thing I've seen before. 1007 00:52:42,828 --> 00:52:45,918 I really did also develop and keep evolving the BMad method 1008 00:52:45,918 --> 00:52:48,645 with the community because people were giving their own 1009 00:52:48,645 --> 00:52:51,188 feedback. They really do appreciate that 1010 00:52:51,188 --> 00:52:54,740 if you've never built software before, if you've never been an 1011 00:52:54,740 --> 00:52:58,640 engineer, we take for granted as engineers, one of our 1012 00:52:58,640 --> 00:53:01,964 greatest skills is the ability to take a problem and 1013 00:53:01,964 --> 00:53:05,138 decompose it into small things, which is why we're gonna work 1014 00:53:05,138 --> 00:53:08,965 with AI so well. That is not something people 1015 00:53:08,965 --> 00:53:11,061 take for granted. And if you've been a tech, if 1016 00:53:11,061 --> 00:53:13,462 you're a tech lead or if you've been in tech tech even 1017 00:53:13,462 --> 00:53:16,174 more than a year or two and you've been out of university 1018 00:53:16,174 --> 00:53:18,590 for a while, you start to forget that. 1019 00:53:18,590 --> 00:53:20,690 It just becomes just an innate skill you have. 1020 00:53:21,080 --> 00:53:22,760 You know, you might even do it in the world. 1021 00:53:22,940 --> 00:53:25,333 You're thinking about how to cook and you're thinking it in 1022 00:53:25,333 --> 00:53:27,170 these small, granular steps, right? 1023 00:53:28,220 --> 00:53:30,920 That is not a natural thing for most people. 1024 00:53:31,280 --> 00:53:34,370 So that is really one of the engineering superpowers. 1025 00:53:34,670 --> 00:53:37,589 So when people that have never developed something and they 1026 00:53:37,589 --> 00:53:40,707 first just try vibe coding, they're not thinking about 1027 00:53:40,707 --> 00:53:42,600 decomposing it into these smaller problems. 1028 00:53:43,230 --> 00:53:46,618 And that's where I think people really realize the magic of 1029 00:53:46,618 --> 00:53:50,375 the BMad method is, oh, I've, when it's working with you and 1030 00:53:50,375 --> 00:53:54,245 you're not a software engineer and it's forcing you to think 1031 00:53:54,245 --> 00:53:57,425 about breaking it down into these smaller pieces, and 1032 00:53:57,425 --> 00:54:00,975 actually thinking about like, oh, I never thought I might 1033 00:54:00,975 --> 00:54:03,546 need authentication. Or yeah, I guess I do need a UI 1034 00:54:03,546 --> 00:54:04,925 to go in and administer these things. 1035 00:54:04,985 --> 00:54:07,121 Or whatever, right? Like we take those for granted 1036 00:54:07,121 --> 00:54:10,442 so much, but it's helping so many people just get their ideas 1037 00:54:10,442 --> 00:54:12,810 down on paper. Whether they build the actual 1038 00:54:12,810 --> 00:54:16,355 final application or it still gets difficult for them at some 1039 00:54:16,355 --> 00:54:19,355 point, it's been magic and then it's been helping people 1040 00:54:19,355 --> 00:54:24,001 plan out their ideas in their applications in a thoughtful, 1041 00:54:24,001 --> 00:54:26,573 methodical way. People have reached out to me 1042 00:54:26,573 --> 00:54:29,343 that it's helped them build pitch decks and, you know, 1043 00:54:29,343 --> 00:54:31,970 launch applications that they've been able to get VC 1044 00:54:31,970 --> 00:54:34,490 funding more, that they were having trouble explaining 1045 00:54:34,490 --> 00:54:36,633 before. And just working with AI, it 1046 00:54:36,641 --> 00:54:39,670 was never doing them for that. So as engineers, remember that 1047 00:54:39,670 --> 00:54:42,758 really is a superpower. The ability to break things down 1048 00:54:42,758 --> 00:54:46,135 into these small tasks, and that is how, that really is the 1049 00:54:46,135 --> 00:54:47,900 magic. And you know, that it really is 1050 00:54:47,900 --> 00:54:49,824 the lesson that everybody learned even from this two 1051 00:54:49,824 --> 00:54:51,530 weeks, you know, sprint that I did. 1052 00:54:51,740 --> 00:54:53,660 Now that I, think about it, I just thought of this now, 1053 00:54:53,660 --> 00:54:56,204 Henry, but it really did kind of remind them that just like 1054 00:54:56,212 --> 00:54:59,036 humans, oh yeah, if we break this down into small, granular 1055 00:54:59,036 --> 00:55:03,447 things and give it to the AI, it can even work well in the 1056 00:55:03,447 --> 00:55:07,117 nastiest old code bases and still figure it out because 1057 00:55:07,117 --> 00:55:10,435 we're giving it this bite-sized bit of instruction that it can 1058 00:55:10,435 --> 00:55:14,024 figure out and do so. Yeah, so that's how it's helped 1059 00:55:14,024 --> 00:55:17,066 out both, you know, in the organization and also, you know, 1060 00:55:17,066 --> 00:55:20,980 people from all walks of life kind of think methodically. 1061 00:55:22,208 --> 00:55:24,348 Cool! So I really like the reminders 1062 00:55:24,356 --> 00:55:25,968 that you mentioned just now, right? 1063 00:55:25,968 --> 00:55:28,986 So when engineers now kind of like lose their identity, you 1064 00:55:28,986 --> 00:55:32,208 know, because code is largely written by AI agents now. 1065 00:55:32,438 --> 00:55:34,993 So I think you reminded something about, you know, 1066 00:55:34,993 --> 00:55:38,663 moving up the abstraction layers and still using our problem 1067 00:55:38,663 --> 00:55:42,601 decomposition skills to actually solve a problem and can 1068 00:55:42,601 --> 00:55:43,938 become much more ambitious now, right? 1069 00:55:43,938 --> 00:55:46,518 You're talking about, you know, epic level instead of just user 1070 00:55:46,518 --> 00:55:49,806 story or individual tasks. So I think those are great 1071 00:55:49,806 --> 00:55:51,438 reminders for engineers to evolve yourself. 1072 00:55:51,738 --> 00:55:54,798 So speaking about, you know, evolve, evolution, right? 1073 00:55:54,798 --> 00:55:57,171 Because one of the challenges working with AI is actually 1074 00:55:57,171 --> 00:55:58,758 yeah, definitely, context engineering. 1075 00:55:58,818 --> 00:56:02,850 Because if you have a complex system, you know, with large 1076 00:56:02,850 --> 00:56:06,066 user stories, large number of features over the time with 1077 00:56:06,074 --> 00:56:08,688 large number of developers, you know, changing the code, 1078 00:56:08,688 --> 00:56:12,150 sometimes it can lose track and it can be out of sync from 1079 00:56:12,150 --> 00:56:15,762 the spec and the code as well. So tell us, how do you actually 1080 00:56:15,762 --> 00:56:17,298 solve this with your BMad method? 1081 00:56:19,223 --> 00:56:23,299 With the BMad method itself, let's just talk about a 1082 00:56:23,299 --> 00:56:25,553 greenfield project for a second, just to keep it simple. 1083 00:56:25,553 --> 00:56:27,769 And then I can talk about, you know, more of a legacy 1084 00:56:27,769 --> 00:56:30,089 application. But for a brand new application, 1085 00:56:30,089 --> 00:56:34,043 one of the places the BMad method does help is just the 1086 00:56:34,043 --> 00:56:36,578 better planning upfront. Again, it will help you think of 1087 00:56:36,578 --> 00:56:39,427 things that you might forget if you were just doing vibe 1088 00:56:39,427 --> 00:56:42,353 coding or even if you just create a quick, simple spec. 1089 00:56:43,245 --> 00:56:45,661 You might forget certain features and then you're 1090 00:56:45,661 --> 00:56:49,042 halfway through and it's harder to have the AI put those in 1091 00:56:49,050 --> 00:56:52,271 after the fact, if you can think of, you know, some of those 1092 00:56:52,271 --> 00:56:54,726 things up front. But, you know, there's a 1093 00:56:54,734 --> 00:56:57,316 saying, best laid plans, right? It's still, you're still gonna 1094 00:56:57,316 --> 00:56:59,256 forget things no matter how much planning. 1095 00:56:59,936 --> 00:57:04,392 You still wanna be Agile with it and you're still gonna have to 1096 00:57:04,392 --> 00:57:09,159 change things. The PRD, the architecture, and 1097 00:57:09,159 --> 00:57:13,886 the specs that you're gonna do, first of all, just like in 1098 00:57:13,886 --> 00:57:16,174 the real world, I always look at those as a snapshot. 1099 00:57:16,174 --> 00:57:18,784 In the real world, when you've been doing Agile or anything, 1100 00:57:18,784 --> 00:57:23,896 you create a solution design. The solution design is a 1101 00:57:23,896 --> 00:57:28,774 point-in-time document that I never that I remember keep 1102 00:57:28,774 --> 00:57:32,224 around as actual documentation long-term. 1103 00:57:33,744 --> 00:57:36,590 Even, you know, before AI, it's gonna evolve, it's gonna 1104 00:57:36,590 --> 00:57:38,200 change. It's meant to kind of convey a 1105 00:57:38,208 --> 00:57:39,334 picture of what we're gonna build. 1106 00:57:39,784 --> 00:57:43,664 But the best documentation is the code itself. 1107 00:57:44,544 --> 00:57:46,614 The code is a piece of documentation. 1108 00:57:46,744 --> 00:57:49,792 And that's the same with the AI. But you asked how does the 1109 00:57:49,792 --> 00:57:52,774 BMad help with that? We do have a few skills in the 1110 00:57:52,774 --> 00:57:54,304 BMad method. One is correct course. 1111 00:57:54,304 --> 00:57:58,814 So if you're halfway through your build and let's say you 1112 00:57:58,814 --> 00:58:02,282 forgot something major like, oh, this relational database is 1113 00:58:02,282 --> 00:58:05,964 not gonna work. I need to use NoSQL. 1114 00:58:07,099 --> 00:58:10,749 I forgot a major feature, I did not consider authentication. 1115 00:58:11,339 --> 00:58:14,219 That's a major change that you're going to use and I hope 1116 00:58:14,219 --> 00:58:16,727 the BMad method will help you actually come to those 1117 00:58:16,727 --> 00:58:19,292 conclusions before that. But you might wanna change it, 1118 00:58:19,292 --> 00:58:21,225 right? We do have a correct course 1119 00:58:21,225 --> 00:58:23,708 thing. The Scrum master will help you 1120 00:58:23,708 --> 00:58:26,744 do correct course, and it will actually talk you through what 1121 00:58:26,744 --> 00:58:31,118 is the impact of your change. Is it something that can just be 1122 00:58:31,118 --> 00:58:33,419 worked in with some new stories and kind of change the 1123 00:58:33,419 --> 00:58:35,581 flow? Or is it so drastic that you 1124 00:58:35,581 --> 00:58:37,902 need to do a replan? And this actually comes from 1125 00:58:37,910 --> 00:58:40,364 Agile ideas also, right? This happens in Agile. 1126 00:58:40,634 --> 00:58:44,642 You might be halfway through a project and you have to pull 1127 00:58:44,642 --> 00:58:48,890 the cord and do a, you know, a quick meeting and figure out, 1128 00:58:48,890 --> 00:58:51,644 are we throwing this all out? Are we gonna pivot? 1129 00:58:51,824 --> 00:58:56,190 It's no different with AI. So what it will do is it will 1130 00:58:56,190 --> 00:58:59,759 help you go through and update the documents as you need to and 1131 00:58:59,759 --> 00:59:02,078 replan the stories. And hopefully, you know, you can 1132 00:59:02,078 --> 00:59:05,908 just change the stories that are not done and work in your 1133 00:59:05,908 --> 00:59:09,224 changes there. So that works actually for both 1134 00:59:09,224 --> 00:59:10,784 greenfield and existing projects. 1135 00:59:10,784 --> 00:59:14,590 So that's one way that it helps. But I will say the bigger thing 1136 00:59:14,590 --> 00:59:18,139 is, the bigger way the BMad method helps, is it does really 1137 00:59:18,139 --> 00:59:21,914 coach you and work through trying to figure out most of 1138 00:59:21,914 --> 00:59:24,719 these things upfront. And that is the biggest 1139 00:59:24,719 --> 00:59:27,566 difference between vibe coding and some of the leaner spec 1140 00:59:27,566 --> 00:59:29,757 engineering kits. I mean, I think they're all 1141 00:59:29,757 --> 00:59:31,616 good. But some of them are not gonna 1142 00:59:31,624 --> 00:59:34,154 really push you to think a lot of these things upfront. 1143 00:59:34,634 --> 00:59:38,196 And sometimes that's fine. So in BMad, there is a quick 1144 00:59:38,196 --> 00:59:41,542 flow also because you don't always need a PRD in an 1145 00:59:41,542 --> 00:59:44,132 architecture up front. If you're working in an existing 1146 00:59:44,132 --> 00:59:47,141 project and you just want to add like a new feature onto an 1147 00:59:47,141 --> 00:59:50,108 existing, you already have a REST API and you want to add a 1148 00:59:50,108 --> 00:59:53,090 new route to it, you don't need the whole B..., you know, you 1149 00:59:53,090 --> 00:59:56,174 don't use the whole BMad method. Instead you use the quick flow 1150 00:59:56,174 --> 01:00:00,146 or you just make a quick spec, which basically amounts to just 1151 01:00:00,146 --> 01:00:03,104 understanding the current state of the project. 1152 01:00:03,404 --> 01:00:06,074 The AI will scan and understand your code base. 1153 01:00:06,344 --> 01:00:08,664 It will understand your documentation, and if you 1154 01:00:08,664 --> 01:00:11,339 don't have documentation, you want to generate good 1155 01:00:11,347 --> 01:00:13,504 documentation and context for the AI. 1156 01:00:14,254 --> 01:00:17,384 And it will help you generate just the story or the spec 1157 01:00:17,384 --> 01:00:19,354 that you need just for the one feature. 1158 01:00:19,764 --> 01:00:24,364 So brownfield in some ways believe it or not, is actually 1159 01:00:24,364 --> 01:00:28,854 easier with AI when you start to understand these techniques. 1160 01:00:29,094 --> 01:00:32,034 And we don't really have enough time to go into a lot of those 1161 01:00:32,034 --> 01:00:34,724 techniques. But again, it's giving the AI 1162 01:00:34,732 --> 01:00:39,214 the understanding of your code base, working in small 1163 01:00:39,214 --> 01:00:41,788 increments. And then you can start doing 1164 01:00:41,788 --> 01:00:43,449 magic things. Like even Claude agents can 1165 01:00:43,457 --> 01:00:46,150 start doing these things now. You can, or you can give it... 1166 01:00:46,158 --> 01:00:49,084 So, you know, Ralph loops are very popular or these things 1167 01:00:49,084 --> 01:00:52,044 that can autonomously, you know, work for you. 1168 01:00:52,044 --> 01:00:54,944 If you start using some of these techniques that will 1169 01:00:54,944 --> 01:00:57,694 actually work for you. And it's pretty amazing where 1170 01:00:57,694 --> 01:01:00,554 the tooling is getting to help you with it. 1171 01:01:01,704 --> 01:01:03,789 Wow, cool! So many different tips and 1172 01:01:03,789 --> 01:01:07,187 trick, I would say, and also embedded inside the BMad 1173 01:01:07,187 --> 01:01:09,227 workflow. It sounds really useful. 1174 01:01:09,407 --> 01:01:12,457 So maybe if, for people who are kind of like intrigued and 1175 01:01:12,457 --> 01:01:15,819 want to give a try, is there any like, I don't know, quick 1176 01:01:15,827 --> 01:01:18,872 start, advice, or tips that you wanna give to the listeners here 1177 01:01:18,872 --> 01:01:21,677 so that they can get up to speed with BMad method? 1178 01:01:21,677 --> 01:01:23,837 Because sometimes when you use a framework, it can be, you 1179 01:01:23,837 --> 01:01:26,159 know, super daunting, you know, there's so many different 1180 01:01:26,159 --> 01:01:28,111 things. But how do you do a quick start 1181 01:01:28,111 --> 01:01:31,693 for BMad? Well, a couple things I wanna 1182 01:01:31,693 --> 01:01:34,927 say is, first of all, if you're an engineer, continue to 1183 01:01:34,927 --> 01:01:38,569 be curious and just learn the tools themselves. 1184 01:01:38,809 --> 01:01:42,073 I would say before even using BMad, if you haven't used AI 1185 01:01:42,073 --> 01:01:45,974 or agent mode yet, first, just start with, you know, Claude 1186 01:01:45,974 --> 01:01:50,389 Code or, you know, Kimi or whatever tool you're gonna use. 1187 01:01:50,389 --> 01:01:54,905 Pick one and learn it well, and try to do the exercise I did 1188 01:01:54,905 --> 01:01:58,689 and start to come to some of the understanding of how the 1189 01:01:58,689 --> 01:02:02,569 context window works. Why it starts hallucinating. 1190 01:02:02,689 --> 01:02:05,869 Why it makes sense to, you know, start new conversations and 1191 01:02:05,869 --> 01:02:09,614 not keep the same one going. Because I have seen a lot of 1192 01:02:09,614 --> 01:02:12,559 people will come into BMad, and without developing some of 1193 01:02:12,559 --> 01:02:15,909 that foundational understanding of just how the LLMs work in 1194 01:02:15,909 --> 01:02:18,029 general, it can still be a little bit of a challenge. 1195 01:02:18,039 --> 01:02:22,129 So I wanna say any of these frameworks that you use, it is 1196 01:02:22,129 --> 01:02:24,516 still important to learn the tool, keep up with what the 1197 01:02:24,516 --> 01:02:27,259 tools are doing, and they're evolving so fast. 1198 01:02:27,349 --> 01:02:29,359 So learn to use a tool very well. 1199 01:02:30,319 --> 01:02:33,929 But then to get into using the BMad method specifically, you 1200 01:02:33,929 --> 01:02:36,709 can go to the GitHub. There's an installer, you 1201 01:02:36,709 --> 01:02:40,667 install it. Once you install it, you can do 1202 01:02:40,667 --> 01:02:45,424 BMad help, which is just a skills/bmad-help, and ask it 1203 01:02:45,424 --> 01:02:48,475 questions and it will give you advice of how to use the BMad 1204 01:02:48,475 --> 01:02:51,443 method. There is a website, 1205 01:02:51,443 --> 01:02:56,860 docs.bmad-method.org. You can go there and there's a 1206 01:02:56,860 --> 01:02:59,095 quickstart on there and it will kind of guide you also 1207 01:02:59,095 --> 01:03:02,365 through, first you do this skill, and then you do this 1208 01:03:02,365 --> 01:03:04,889 skill, and then you do this skill, and it's, and there's a 1209 01:03:04,889 --> 01:03:07,494 pretty simple workflow there. So that is how I would 1210 01:03:07,502 --> 01:03:11,505 recommend using the BMad method. But if somebody just wants to 1211 01:03:11,505 --> 01:03:14,749 experience the magic of treating the AI as a 1212 01:03:14,749 --> 01:03:18,425 facilitator, if nothing else, again, even if you don't think 1213 01:03:18,425 --> 01:03:22,408 you're creative, follow the tutorial, install the BMad 1214 01:03:22,416 --> 01:03:24,634 method for the tool of your choice. 1215 01:03:24,934 --> 01:03:28,024 I would recommend Claude Code, but whatever you're using. 1216 01:03:29,794 --> 01:03:33,574 Just work through — a GLM is very good with it also 'cause 1217 01:03:33,574 --> 01:03:35,194 they're both very creative models. 1218 01:03:36,454 --> 01:03:40,264 Just do the brainstorming and see how interesting it is. 1219 01:03:40,264 --> 01:03:41,560 And another thing I didn't mention about the 1220 01:03:41,560 --> 01:03:43,894 brainstorming is at the end of the brainstorming, it's gonna do 1221 01:03:43,894 --> 01:03:45,994 two or three different techniques with you. 1222 01:03:46,264 --> 01:03:49,104 It's gonna produce this report and a summary of the 1223 01:03:49,104 --> 01:03:51,526 brainstorming you did. And it's gonna also tell you 1224 01:03:51,534 --> 01:03:54,028 the nuggets and insights and highlights that you actually 1225 01:03:54,028 --> 01:03:56,044 came up with doing this brainstorming. 1226 01:03:56,944 --> 01:03:59,900 Just try that. You'll get a taste and a feeling 1227 01:03:59,900 --> 01:04:04,498 for what it means to treat the AI as a facilitator instead of 1228 01:04:04,506 --> 01:04:09,334 just a, you know, better Google. And that will I think unlock the 1229 01:04:09,334 --> 01:04:11,119 rest of the BMad method with you. 1230 01:04:11,299 --> 01:04:12,949 One of the thing is we do have a Discord. 1231 01:04:13,189 --> 01:04:15,979 It's totally free, so I would invite everybody to come there. 1232 01:04:16,459 --> 01:04:19,579 We have a community of 20-30,000 in there. 1233 01:04:20,029 --> 01:04:22,519 Very popular, but very helpful people. 1234 01:04:22,519 --> 01:04:25,389 There's just a lot of people. Also in organizations there's a 1235 01:04:25,389 --> 01:04:28,569 lot of tech leaders and people that own companies in there, all 1236 01:04:28,569 --> 01:04:32,197 trying to figure out how to apply AI in the workforce and 1237 01:04:32,197 --> 01:04:34,969 how to use these tools. It's a very welcoming community. 1238 01:04:34,969 --> 01:04:36,319 So just come in there and ask questions. 1239 01:04:36,319 --> 01:04:39,339 Also we're very happy to help you, you know, just how to use 1240 01:04:39,339 --> 01:04:41,814 the AI agents in general. We're in there exploring, you 1241 01:04:41,822 --> 01:04:44,869 know, what to do with these new models and what the new features 1242 01:04:44,869 --> 01:04:46,279 are and all these new techniques. 1243 01:04:46,519 --> 01:04:49,148 Because this is gonna change. What I'm saying today will not 1244 01:04:49,156 --> 01:04:52,849 be the same six months from now. So there's a lot of resources 1245 01:04:52,849 --> 01:04:54,619 out there. So I welcome everybody to come 1246 01:04:54,619 --> 01:04:57,741 in and try it and say hi. Reach out to me and say hello 1247 01:04:57,741 --> 01:05:00,399 also, if you heard about this on Henry's podcast here. 1248 01:05:00,622 --> 01:05:02,760 Yeah. So I'll definitely make sure to 1249 01:05:02,760 --> 01:05:04,552 put those things in the show notes. 1250 01:05:04,582 --> 01:05:05,992 So you mentioned about a couple of models. 1251 01:05:05,992 --> 01:05:10,042 So tell us like what models that these BMad tool support? 1252 01:05:10,552 --> 01:05:13,296 Does it have to be BYOK, or you know, what kind of, you know, 1253 01:05:13,296 --> 01:05:16,492 how do you run the agent with the AI model? 1254 01:05:17,229 --> 01:05:18,699 Sure. Very good question. 1255 01:05:19,009 --> 01:05:22,913 One thing I recommend to everybody is if you can afford 1256 01:05:22,913 --> 01:05:26,205 it — well this is much more affordable regardless, is go 1257 01:05:26,205 --> 01:05:28,779 with a subscription. Don't bring your own API key. 1258 01:05:29,049 --> 01:05:32,675 I will say some of the BMad method workflows right now, we 1259 01:05:32,675 --> 01:05:35,319 are working on improving them and making them leaner, but 1260 01:05:35,319 --> 01:05:38,351 they are context heavy. These are gonna use a lot of 1261 01:05:38,351 --> 01:05:40,883 tokens because it is working back and forth with you doing 1262 01:05:40,883 --> 01:05:43,476 all of this planning. It will read in documentation 1263 01:05:43,484 --> 01:05:47,877 and other things. So I highly recommend getting a 1264 01:05:47,877 --> 01:05:53,244 good subscription to Claude Code or Codex or, you know, 1265 01:05:53,244 --> 01:05:56,719 Kimi, or any of these. The nice thing about the BMad 1266 01:05:56,727 --> 01:05:59,694 method is we have built it to be model and tool agnostic. 1267 01:05:59,694 --> 01:06:03,048 So when you use the installer for the BMad method, I think 1268 01:06:03,048 --> 01:06:05,964 there's 12 or 15 tools that we support in there now. 1269 01:06:06,414 --> 01:06:09,204 Very soon we will be full skill compliant. 1270 01:06:09,204 --> 01:06:12,924 So this will work with anything that basically supports skills. 1271 01:06:13,074 --> 01:06:15,944 One of the goals with the BMad method is to actually get away 1272 01:06:15,944 --> 01:06:20,274 from the IDE and also support tools like Claude Cowork or 1273 01:06:20,282 --> 01:06:22,734 these platforms that allow non-technical. 1274 01:06:22,734 --> 01:06:26,012 Very soon the skills explosion is gonna lead to, you know, 1275 01:06:26,012 --> 01:06:28,974 web browsers and web tools will also support skills. 1276 01:06:28,974 --> 01:06:31,614 So very soon this is gonna work across platforms. 1277 01:06:31,614 --> 01:06:33,684 But find something, get a subscription, 'cause you're 1278 01:06:33,684 --> 01:06:38,659 gonna be using a lot of tokens. And that is my advice. 1279 01:06:38,869 --> 01:06:42,427 But it works across the models. Some models are more creative 1280 01:06:42,427 --> 01:06:44,659 than others. Some models will produce 1281 01:06:44,667 --> 01:06:48,297 better code than others. So if you use a tool like Cursor 1282 01:06:48,297 --> 01:06:51,317 or something that allows you to use different models, you'll 1283 01:06:51,317 --> 01:06:55,474 start to learn over time that the brainstorming is better with 1284 01:06:55,474 --> 01:06:59,263 Claude or GLM. Code review, we have code review 1285 01:06:59,263 --> 01:07:01,489 things in there that help improve the code. 1286 01:07:01,489 --> 01:07:05,459 Those might be better with Codex or, you know, the ChatGPT 1287 01:07:05,509 --> 01:07:08,299 model or these other things. But you don't have to worry 1288 01:07:08,307 --> 01:07:09,349 about all those details at first. 1289 01:07:09,349 --> 01:07:13,849 At first, just pick a tool, pick a model, and start learning 1290 01:07:13,849 --> 01:07:17,005 that one really well. And you'll start to learn maybe 1291 01:07:17,005 --> 01:07:20,123 where some models work better than the other ones but you can 1292 01:07:20,123 --> 01:07:22,486 learn to work with the model that you have also and make it 1293 01:07:22,486 --> 01:07:24,275 work for you. So those are just some of the 1294 01:07:24,283 --> 01:07:27,889 skills that you have to kind of develop a feel for over time. 1295 01:07:28,159 --> 01:07:30,749 But you start to pick it up, it's kind of like riding a bike. 1296 01:07:30,749 --> 01:07:34,408 You just start to kind of get a feel for it, and you'll start 1297 01:07:34,408 --> 01:07:35,479 working together with it really well. 1298 01:07:36,532 --> 01:07:39,280 Yeah. Always happy to whenever the 1299 01:07:39,280 --> 01:07:42,282 tool is agnostic, you know, especially all these AI models 1300 01:07:42,282 --> 01:07:43,762 that you can just pick and choose, right? 1301 01:07:43,762 --> 01:07:46,462 So definitely, it's a advantage, I would say. 1302 01:07:47,572 --> 01:07:50,124 We have covered a lot. As we go to the end of our 1303 01:07:50,124 --> 01:07:51,892 conversation, is there something else that you wanna 1304 01:07:51,892 --> 01:07:55,262 mention or shout out about BMad, maybe something that we 1305 01:07:55,262 --> 01:07:58,943 haven't covered today? You know, I think we covered a 1306 01:07:58,943 --> 01:08:01,524 lot of it, Henry. Again, I just suggest everybody 1307 01:08:01,524 --> 01:08:03,639 go out there and just try it out. 1308 01:08:03,849 --> 01:08:06,099 Have fun with it. And if you take nothing else 1309 01:08:06,099 --> 01:08:08,889 away from this, I'll say again, just treat the AI like a 1310 01:08:08,889 --> 01:08:11,849 facilitator. And it's honestly a magical 1311 01:08:11,849 --> 01:08:16,649 experience when you try it the first time, if you haven't 1312 01:08:16,649 --> 01:08:20,923 really worked with AI before. So that is kind of like my one 1313 01:08:20,923 --> 01:08:21,997 message. Right. 1314 01:08:22,506 --> 01:08:25,537 So Brian, as a tradition in my podcast, I always ask this 1315 01:08:25,537 --> 01:08:28,309 last question which I call the three technical leadership 1316 01:08:28,309 --> 01:08:30,283 wisdom. If you can think of it just like 1317 01:08:30,283 --> 01:08:32,402 advice you wanna give to the listeners, maybe if you can 1318 01:08:32,402 --> 01:08:34,162 share your version today that would be great. 1319 01:08:35,904 --> 01:08:38,754 Yeah. I really love this question. 1320 01:08:39,024 --> 01:08:41,163 I've listened to a few of your episodes, Henry, and I have 1321 01:08:41,163 --> 01:08:43,464 become a fan, and I like that you asked this for everybody. 1322 01:08:44,044 --> 01:08:47,881 I want to take this from the perspective of sharing 1323 01:08:47,881 --> 01:08:49,974 information for, with new tech leads. 1324 01:08:50,363 --> 01:08:52,776 Or people that are leaders or people that might be thinking 1325 01:08:52,776 --> 01:08:55,085 about leaders, because we can all be leaders, especially now 1326 01:08:55,085 --> 01:08:58,578 with AI, right? So, you know, the first one that 1327 01:08:58,578 --> 01:09:01,653 I've seen leaders really struggle with, especially when 1328 01:09:01,653 --> 01:09:03,504 they're a new leader is delegation. 1329 01:09:03,834 --> 01:09:08,502 So I just want you to remember that you as a tech lead or an 1330 01:09:08,502 --> 01:09:13,384 engineering manager or any kind of leader, your goal is really 1331 01:09:13,384 --> 01:09:16,703 to start becoming a force multiplier for your team. 1332 01:09:17,474 --> 01:09:20,464 And the best way you can do that is to delegate. 1333 01:09:20,974 --> 01:09:24,255 And at first you're gonna feel like, oh, I'm just pawning my 1334 01:09:24,255 --> 01:09:26,694 work off to people or this doesn't feel right. 1335 01:09:27,174 --> 01:09:30,585 But the more that you can learn to do that, A) the more you 1336 01:09:30,585 --> 01:09:34,180 can really focus and shepherding and helping your team, but 1337 01:09:34,180 --> 01:09:39,934 also you're going to give people on your team the opportunity to 1338 01:09:39,934 --> 01:09:41,934 go, to grow, and be responsible for things. 1339 01:09:42,413 --> 01:09:46,317 And you'll be able to take on more and more leadership 1340 01:09:46,317 --> 01:09:49,569 because you're starting to learn to trust people and delegate. 1341 01:09:50,049 --> 01:09:54,279 And even, you know, get away from knowing all of the details 1342 01:09:54,279 --> 01:09:58,124 and relying on other people. It's really, delegation is a 1343 01:09:58,124 --> 01:10:02,409 superpower and it's not like cheating, but I see people 1344 01:10:02,409 --> 01:10:05,111 struggle with that. So delegation, you know, is 1345 01:10:05,119 --> 01:10:08,954 really a force multiplier. The second one, and I already 1346 01:10:08,962 --> 01:10:12,574 talked about this so I'll be really brief on this, but giving 1347 01:10:12,574 --> 01:10:15,339 your team permission to experiment and fail. 1348 01:10:17,049 --> 01:10:19,479 And I already talked about the AI sprint. 1349 01:10:20,169 --> 01:10:22,907 But what that really means is sometimes you really gotta 1350 01:10:22,907 --> 01:10:25,721 give them the space to experiment and know that it's 1351 01:10:25,721 --> 01:10:28,299 okay if they come out with the wrong thing. 1352 01:10:28,859 --> 01:10:32,004 Or, you know, pull back the pressures from the sprint for 1353 01:10:32,004 --> 01:10:35,445 somebody to let them do a little bit of an experiment to come 1354 01:10:35,445 --> 01:10:38,589 to these conclusions. And you will see a flourishing 1355 01:10:38,589 --> 01:10:42,351 of new ideas and new ways of thinking if people really feel — 1356 01:10:42,351 --> 01:10:45,747 I think the term is psychological complexity or is 1357 01:10:45,747 --> 01:10:48,344 that psychological safety, excuse me, right? 1358 01:10:48,634 --> 01:10:52,695 But part of psychological safety — and complexity — is the 1359 01:10:52,695 --> 01:10:55,034 ability to feel comfortable in failing. 1360 01:10:55,034 --> 01:10:57,743 People forget that, and sometimes just think it means, 1361 01:10:57,743 --> 01:11:00,549 you know, feeling comfortable talking, but it is also 1362 01:11:00,549 --> 01:11:02,804 permission to fail or permission to experiment. 1363 01:11:04,764 --> 01:11:08,477 The third one, and this is a new one that I've come up with and 1364 01:11:08,477 --> 01:11:11,826 it's really recent, but with this AI and transformation, 1365 01:11:11,826 --> 01:11:15,742 remember that a lot of people are scared right now or just 1366 01:11:15,742 --> 01:11:17,514 don't know what the future holds. 1367 01:11:17,514 --> 01:11:19,224 I don't know if scared is the right word, but some people are 1368 01:11:19,224 --> 01:11:20,484 scared, some people are uncertain. 1369 01:11:21,234 --> 01:11:24,104 None of us have the answer of what is coming. 1370 01:11:25,394 --> 01:11:28,011 As a leader — and I said this at the beginning, I really picked 1371 01:11:28,011 --> 01:11:31,729 this up from my army training — sometimes it is putting a good 1372 01:11:31,729 --> 01:11:35,975 face on things because not only can your team recognize 1373 01:11:35,975 --> 01:11:40,573 confidence, but they can also recognize fear, and that will 1374 01:11:40,573 --> 01:11:42,279 have an impact on how they operate. 1375 01:11:42,789 --> 01:11:47,019 So sometimes you do have to put a good face on things to 1376 01:11:47,019 --> 01:11:51,147 lead people through adversity. And I'm not saying AI and this 1377 01:11:51,155 --> 01:11:56,124 transformation is adversity. But helping people learn to grow 1378 01:11:56,124 --> 01:11:59,934 through this new transition and recognize their own 1379 01:11:59,934 --> 01:12:02,134 self-worth. And people are gonna embrace 1380 01:12:02,134 --> 01:12:04,817 these tools and then they're gonna have these moments of, 1381 01:12:04,817 --> 01:12:09,441 wait, what is my purpose now? If you can shepherd them through 1382 01:12:09,441 --> 01:12:12,519 these times, that will make you a great leader. 1383 01:12:12,849 --> 01:12:16,734 But that also really shows that you care about people. 1384 01:12:16,734 --> 01:12:18,504 As a leader, you need to care about people. 1385 01:12:18,654 --> 01:12:22,527 And part of caring is just helping lead through adversity 1386 01:12:22,527 --> 01:12:26,994 or challenging times, or helping people realize what their 1387 01:12:26,994 --> 01:12:29,262 self-worth is. And I think we've shared a 1388 01:12:29,262 --> 01:12:31,471 little bit of that throughout the podcast of what your 1389 01:12:31,471 --> 01:12:35,508 self-worth is, as these AI tools comes, it is stepping back and 1390 01:12:35,508 --> 01:12:38,246 applying your experience and your discernment and knowing 1391 01:12:38,246 --> 01:12:43,244 that you can work with the AI as a partner, but you're giving it 1392 01:12:43,244 --> 01:12:47,856 something that the AI, I don't think has or ever will has, 1393 01:12:47,856 --> 01:12:51,654 which is your lived experience and knowledge and discernment. 1394 01:12:51,654 --> 01:12:55,194 So helping your team do that is really part of being a leader. 1395 01:12:56,904 --> 01:13:00,198 I have one more, Henry, though. I know you said three but I 1396 01:13:00,198 --> 01:13:03,988 wanna give one more. And this is a cliche old one 1397 01:13:03,996 --> 01:13:06,594 that I've always heard, but it applies now more than ever. 1398 01:13:07,194 --> 01:13:10,704 Leaders are readers and I really strongly believe that. 1399 01:13:11,034 --> 01:13:15,615 If you're new to tech leadership, it's fun to focus 1400 01:13:15,615 --> 01:13:19,119 on, you know, still learning engineering patterns. 1401 01:13:19,869 --> 01:13:24,077 There is a lot of fun in learning about leadership and 1402 01:13:24,077 --> 01:13:26,259 about management and organization. 1403 01:13:26,259 --> 01:13:29,853 Maybe this sounds boring. You had one of my favorite 1404 01:13:29,853 --> 01:13:32,389 people on your podcast a few weeks ago, Gene Kim. 1405 01:13:32,989 --> 01:13:36,201 I am a jump. I love Gene Kim's books, I've 1406 01:13:36,201 --> 01:13:39,379 read all of them. And I love business books and 1407 01:13:41,409 --> 01:13:43,779 management books that kind of have a story in 'em. 1408 01:13:43,869 --> 01:13:46,089 So I'm a sucker for those types of books. 1409 01:13:46,899 --> 01:13:49,722 But learn these techniques and you'll start to realize it's 1410 01:13:49,722 --> 01:13:52,604 just like engineering. You're starting to learn ways 1411 01:13:52,612 --> 01:13:55,479 of working with people or how to apply these different things. 1412 01:13:55,779 --> 01:13:59,010 And you might start to really enjoy the leadership and 1413 01:13:59,010 --> 01:14:01,749 management side of things and it will just help you grow. 1414 01:14:01,749 --> 01:14:05,691 So read about techniques and focus not only on the 1415 01:14:05,691 --> 01:14:07,829 technical, but how to become a better leader. 1416 01:14:08,099 --> 01:14:10,199 And you will, you'll become a better leader. 1417 01:14:10,259 --> 01:14:12,854 It's a lot of fun. So, sorry I had to squeeze one 1418 01:14:12,854 --> 01:14:15,469 more in there. But, you know, leaders are 1419 01:14:15,477 --> 01:14:20,014 readers and I believe that. Happy for more wisdom from my 1420 01:14:20,014 --> 01:14:22,243 guest to share. So thank you for sharing such 1421 01:14:22,243 --> 01:14:25,042 a lovely message. I think all those are really 1422 01:14:25,042 --> 01:14:27,172 critical, you know, skills for leaders. 1423 01:14:27,172 --> 01:14:29,860 You know, being able to delegate, creating 1424 01:14:29,860 --> 01:14:32,986 psychological safety, right? Caring about people going 1425 01:14:32,986 --> 01:14:35,632 through, you know, difficult periods, you know, adversity and 1426 01:14:35,632 --> 01:14:37,531 things like that. And definitely, inviting 1427 01:14:37,539 --> 01:14:41,518 people to read more. So BMad, if people want to 1428 01:14:41,518 --> 01:14:44,702 connect with you, asking you more questions maybe about BMad 1429 01:14:44,702 --> 01:14:47,335 maybe about other things, is there a place where they can 1430 01:14:47,335 --> 01:14:50,592 find you online? Yeah, the best thing, honestly, 1431 01:14:50,592 --> 01:14:52,426 it's easy. I mean I have a lot of email 1432 01:14:52,434 --> 01:14:53,574 addresses, but check me out on YouTube. 1433 01:14:53,574 --> 01:14:56,574 I do have a masterclass on there where you can also watch and 1434 01:14:56,574 --> 01:14:59,462 learn about the BMad method and my contact information is on 1435 01:14:59,462 --> 01:15:01,859 YouTube. But the easiest is come into 1436 01:15:01,859 --> 01:15:03,718 Discord. So if you go to our GitHub for 1437 01:15:03,718 --> 01:15:05,334 the BMad method, there is a Discord link. 1438 01:15:05,664 --> 01:15:07,644 You can join our community. It's free. 1439 01:15:07,644 --> 01:15:09,684 Come in there and say hello, ask questions. 1440 01:15:09,684 --> 01:15:12,582 Myself and the community will be happy to help you out and tell 1441 01:15:12,582 --> 01:15:16,562 you more about BMad or you know, how to apply engineering and 1442 01:15:16,562 --> 01:15:18,644 these changes across your own organization. 1443 01:15:18,654 --> 01:15:20,782 Happy to help. Right. 1444 01:15:21,232 --> 01:15:22,882 It's been a pleasant conversation. 1445 01:15:22,882 --> 01:15:24,622 Thank you so much for sharing all those things. 1446 01:15:24,622 --> 01:15:27,702 I really love your story as well, the background, how you 1447 01:15:27,702 --> 01:15:30,644 came up with this BMad method. I wish you good luck and great 1448 01:15:30,644 --> 01:15:32,072 success with this evolution of the BMad. 1449 01:15:33,714 --> 01:15:35,274 Thank you Henry. This was so much fun. 1450 01:15:35,274 --> 01:15:38,054 I really enjoyed talking to you and I love your show. 1451 01:15:38,054 --> 01:15:40,299 So, thank you so much. This was awesome.