1 00:00:00,000 --> 00:00:02,846 My team has two designers. When the growth team wants to 2 00:00:02,846 --> 00:00:05,294 create some pages, the designers take that design, use Locofy, 3 00:00:05,294 --> 00:00:08,134 get the code, pull it into Cursor, just prompt, explain, 4 00:00:08,134 --> 00:00:10,548 and then publish it. And maybe an engineer needs 10 5 00:00:10,548 --> 00:00:13,186 minutes to just review. Locofy converts your existing 6 00:00:13,186 --> 00:00:16,140 Figma designs into whichever sort of front-end framework. 7 00:00:16,230 --> 00:00:19,901 React, Angular, Vue, HTML, CSS, Flutter, React Native, pretty 8 00:00:19,901 --> 00:00:22,345 much anything. Cut the time by 70-80%. 9 00:00:22,365 --> 00:00:25,720 Honey Mittal built his own large design models before ChatGPT 10 00:00:25,720 --> 00:00:27,975 existed. Spent four years on AI that 11 00:00:27,975 --> 00:00:29,805 understands visual design, not just text. 12 00:00:29,865 --> 00:00:33,471 His company, Locofy, now helps engineers build front-end 2 - 3 13 00:00:33,471 --> 00:00:35,597 times faster. As we were launching this, 14 00:00:35,597 --> 00:00:37,747 ChatGPT launched. And we started hearing this term 15 00:00:37,747 --> 00:00:40,675 LLMs. But the more we looked at LLMs, the more we realized the 16 00:00:40,675 --> 00:00:43,972 LLMs architecture itself makes it super hard, if not impossible 17 00:00:43,972 --> 00:00:47,318 to understand Figma design. We spent about four years 18 00:00:47,318 --> 00:00:49,280 building our design models from scratch. 19 00:00:49,400 --> 00:00:50,720 We call it the large design models. 20 00:00:50,750 --> 00:00:53,630 You mentioned about maybe 70%, 80% reduction. 21 00:00:53,660 --> 00:00:55,280 How long does it take for you to generate? 22 00:00:55,440 --> 00:00:57,860 The code itself? Not more than 20 to 30 seconds. 23 00:00:57,890 --> 00:01:01,850 An engineer who was able to do 10 designs or 10 screens per 24 00:01:01,850 --> 00:01:03,710 month is now able to do 25 to 30. 25 00:01:04,160 --> 00:01:07,679 We have customers who have built large projects of 1,000+ 26 00:01:07,679 --> 00:01:09,605 screens. In a matter of months using 27 00:01:09,605 --> 00:01:11,290 Locofy. And I do believe that designers 28 00:01:11,290 --> 00:01:13,679 can do the front-end development and they have absolutely every 29 00:01:13,679 --> 00:01:16,179 reason and all the tools available today to do it. 30 00:01:16,300 --> 00:01:20,640 No more Business PM, Technical PM, Lead UX Researcher, UI 31 00:01:20,640 --> 00:01:23,332 Designer, UX Designer. Could just be a Product Manager 32 00:01:23,332 --> 00:01:40,937 and a Designer. Hello. 33 00:01:40,937 --> 00:01:43,697 Welcome back to another new episode of the Tech Lead Journal 34 00:01:43,697 --> 00:01:45,197 podcast. Another in-person recording. 35 00:01:45,197 --> 00:01:47,807 This time, I have Honey Mittal with me today. 36 00:01:47,927 --> 00:01:51,477 He's the CEO and co-founder of Locofy.ai. 37 00:01:52,149 --> 00:01:56,213 Interestingly, it's a tool that can convert like design, maybe 38 00:01:56,213 --> 00:01:59,229 Figma design, into working front-end code. 39 00:01:59,791 --> 00:02:02,789 I think it's always like a dream of many developers, you know, 40 00:02:02,789 --> 00:02:05,349 not having to build front-end UI and all that. 41 00:02:05,349 --> 00:02:07,659 So I'm very pleased to have you in the show today. 42 00:02:07,659 --> 00:02:09,859 And I hope to learn a lot about Locofy today. 43 00:02:10,139 --> 00:02:11,649 So welcome, Honey. Thanks for having me, Henry. 44 00:02:11,649 --> 00:02:13,445 Yeah. Good to see this happening in 45 00:02:13,445 --> 00:02:14,319 Singapore. Right. 46 00:02:14,769 --> 00:02:18,489 So Honey, I love to maybe understand a bit more about your 47 00:02:18,489 --> 00:02:20,760 background. So maybe looking back from your 48 00:02:20,760 --> 00:02:24,144 career, from your past history until now, would you share maybe 49 00:02:24,144 --> 00:02:27,240 any kind of learnings or key turning points that you had in 50 00:02:27,240 --> 00:02:29,088 your career that we can learn from you as well? 51 00:02:29,132 --> 00:02:31,778 Yeah. absolutely. So I originally come from, uh, 52 00:02:31,778 --> 00:02:35,160 the north of India, near the Himalayas, but I've been now in 53 00:02:35,160 --> 00:02:37,573 Singapore 20 years. I came here to study at NUS. 54 00:02:38,176 --> 00:02:40,952 And I think, since you talk about turning points, I think 55 00:02:40,952 --> 00:02:44,837 for me the biggest turning point was the app store launch, 56 00:02:44,837 --> 00:02:48,348 because, uh, I was essentially building Windows mobile apps 57 00:02:48,348 --> 00:02:52,120 before the app store. I was an intern in Microsoft and 58 00:02:52,120 --> 00:02:54,945 I picked it up, .NET and everything, and I was building 59 00:02:54,945 --> 00:02:57,893 these apps to find location and post it on social media. 60 00:02:58,703 --> 00:03:00,929 And I think when the app store was launched, I think it was a 61 00:03:00,929 --> 00:03:03,717 big aha moment. That said, I was still, as a 62 00:03:03,717 --> 00:03:06,842 graduate, there were not many startups back in 2009 and 2010. 63 00:03:07,552 --> 00:03:10,426 So I was just building things for, yeah, just as a hobby. 64 00:03:11,146 --> 00:03:14,116 But then I think when the Singapore startup scene kind of 65 00:03:14,116 --> 00:03:17,500 started getting a little bit more interesting around 2012-13, 66 00:03:17,500 --> 00:03:21,450 that's where I think just like today, people are looking for 67 00:03:21,450 --> 00:03:23,866 AI, uh, you know, native engineers, product managers. 68 00:03:24,226 --> 00:03:27,357 Back then was more about mobile. And I was one of the few people 69 00:03:27,357 --> 00:03:30,062 who had actually built mobile apps in Singapore, no matter how 70 00:03:30,062 --> 00:03:33,751 scrappy and looking back at it, how not so good, you know, those 71 00:03:33,751 --> 00:03:36,205 apps were. But essentially I joined 72 00:03:36,205 --> 00:03:40,121 Wego.com and, uh, I think my career took off, I guess. 73 00:03:40,258 --> 00:03:43,645 Because, yeah, I was able to build my own mobile apps for a 74 00:03:43,645 --> 00:03:45,923 few thousand bucks, right? So joining a Series C startup, I 75 00:03:45,923 --> 00:03:49,480 had the budgets and I, but I brought that sort of early stage 76 00:03:49,480 --> 00:03:53,408 mentality into the startup. The first engineer I hired in 77 00:03:53,408 --> 00:03:57,351 Wego is today my co-founder. So I think that was also a big 78 00:03:57,351 --> 00:04:00,480 turning point, just realizing that I got lucky in terms of 79 00:04:00,480 --> 00:04:01,966 hiring. And I started also realizing I 80 00:04:01,966 --> 00:04:04,990 was good at hiring maybe and attracting the right talent and 81 00:04:04,990 --> 00:04:08,831 seeing just what a 10x engineer meant and can do. 82 00:04:09,491 --> 00:04:12,105 And that kind of shaped what we are doing today as well. 83 00:04:12,105 --> 00:04:15,329 To your point, we were building, we built the first Editor's 84 00:04:15,329 --> 00:04:18,940 Choice app on Google Play from Asia in 2014. 85 00:04:19,541 --> 00:04:21,557 Uh, and even the Google Play guys came down to Singapore and 86 00:04:21,557 --> 00:04:24,389 they were like, we can't believe, I mean, you must have a 87 00:04:24,389 --> 00:04:27,318 hundred people engineering team. And I said, I have one engineer, 88 00:04:27,318 --> 00:04:28,944 who has never done mobile before. 89 00:04:29,654 --> 00:04:33,517 So I think through not just my learnings from my own sort of 90 00:04:33,517 --> 00:04:37,411 career, but also I leaned a lot on Sohaib, who's my co-founder's 91 00:04:37,411 --> 00:04:40,281 experience as well because he just went from gaming to mobile 92 00:04:40,281 --> 00:04:42,641 apps and just killed, kicked it out of the park. 93 00:04:43,841 --> 00:04:47,044 Then we took over web of which we had no idea how to do and 94 00:04:47,044 --> 00:04:50,141 that became interesting. We launched the world's first 95 00:04:50,141 --> 00:04:52,047 progressive web app, for travel especially. 96 00:04:52,047 --> 00:04:54,102 It was showcased at Google I/O, left, right, and center. 97 00:04:54,882 --> 00:04:58,572 And I think, um, that's where my co-founder and I, our careers 98 00:04:58,572 --> 00:05:01,712 kind of grew really quickly. And we also started realizing 99 00:05:01,712 --> 00:05:04,013 that we are very passionate about building amazing 100 00:05:04,013 --> 00:05:06,734 experiences like, you know, irrespective of whether it's 101 00:05:06,734 --> 00:05:08,454 mobile, tablet, which framework it is. 102 00:05:08,844 --> 00:05:11,996 And that led to where we are today, uh, starting Locofy about 103 00:05:11,996 --> 00:05:15,924 four and a half years ago, was based on the last decade of 104 00:05:15,924 --> 00:05:18,699 experience and realizing a lot of companies and engineers 105 00:05:18,699 --> 00:05:21,834 either struggle with it or they see it as grunt work. 106 00:05:22,314 --> 00:05:25,777 Yet very few people can actually achieve amazing results. 107 00:05:26,806 --> 00:05:28,576 And that's where Locofy was born. 108 00:05:28,832 --> 00:05:30,968 Right. So thanks for sharing, uh, some 109 00:05:30,968 --> 00:05:33,085 of the achievements, the highlights, your past 110 00:05:33,085 --> 00:05:35,136 experience. I think it was very cool when 111 00:05:35,136 --> 00:05:37,987 you mentioned that you published the first Google Play apps in 112 00:05:37,987 --> 00:05:40,562 the store, so I think, that's, uh, really great. 113 00:05:40,862 --> 00:05:43,766 So I think you mentioned about Wego and I know that in your 114 00:05:43,766 --> 00:05:46,509 career you also went to FinAccel, right? 115 00:05:46,553 --> 00:05:47,679 Yeah. Yeah. 116 00:05:47,679 --> 00:05:49,749 And also the other thing is, uh, Homage, right? 117 00:05:49,989 --> 00:05:52,288 So three startups. But, um, you mentioned in the 118 00:05:52,288 --> 00:05:54,873 beginning, you were building mobile apps, but then you turned 119 00:05:54,873 --> 00:05:57,639 into CPO actually, like product management and all that. 120 00:05:57,639 --> 00:06:01,370 So tell us this transition, what makes you transition to product 121 00:06:01,370 --> 00:06:03,237 management? Yeah, so I actually started my 122 00:06:03,237 --> 00:06:06,711 career in Credit Suisse. And I realized very quickly that 123 00:06:06,711 --> 00:06:09,451 I'm not cut out to work in a large bank. 124 00:06:10,624 --> 00:06:13,337 But the good thing over there was I had free time and I was, I 125 00:06:13,337 --> 00:06:15,922 used to hustle a lot, right? So I was building my own apps 126 00:06:15,922 --> 00:06:17,344 and some people noticed at Credit Suisse. 127 00:06:17,974 --> 00:06:20,246 And they offered me to build something for Credit Suisse 128 00:06:20,246 --> 00:06:22,497 outta my free time. But I just realized how slow it 129 00:06:22,497 --> 00:06:24,013 was, and then I was like, okay, you know what? 130 00:06:24,013 --> 00:06:25,993 I'm gonna be a, I'm gonna work in a startup. 131 00:06:25,993 --> 00:06:28,843 So that was, I think, a very big turning point. 132 00:06:29,680 --> 00:06:33,433 But yeah, also I also said, you know, if I go to a startup, I'm 133 00:06:33,433 --> 00:06:35,183 not a 10 out of 10 or a 10x engineer. 134 00:06:35,183 --> 00:06:37,433 I'm maybe a seven or eight, and that was a long time ago. 135 00:06:38,153 --> 00:06:41,851 And I just realized that what I was really naturally maybe good 136 00:06:41,851 --> 00:06:44,737 at, a lot of people started telling me about this, and I 137 00:06:44,737 --> 00:06:47,021 didn't even know about it, was that, hey, you're a really good 138 00:06:47,021 --> 00:06:48,881 product manager. And I was like, well, what's a 139 00:06:48,881 --> 00:06:52,320 product manager? And the more I read about it, 140 00:06:52,320 --> 00:06:54,669 the more I started realizing that's essentially what I want 141 00:06:54,669 --> 00:06:56,798 to do. And I set myself a goal that I, 142 00:06:56,798 --> 00:06:58,569 by the age of 30, I wanna be a product manager. 143 00:06:58,959 --> 00:07:02,154 I mean, of course back then, product managers would not just 144 00:07:02,154 --> 00:07:03,624 come outta straight out of college. 145 00:07:04,104 --> 00:07:07,464 It was expected for you to be an engineer for five, six years and 146 00:07:07,464 --> 00:07:10,572 then have some business sort of sense and spend time with 147 00:07:10,572 --> 00:07:13,438 designs. So I set myself that goal and I 148 00:07:13,438 --> 00:07:16,714 got a job as a product manager with Wego, purely because of my 149 00:07:16,714 --> 00:07:19,610 experience with building mobile apps, being one of the early 150 00:07:19,610 --> 00:07:23,775 adopters of building apps. And, uh, because I think I was, 151 00:07:23,775 --> 00:07:26,954 I led that transformation from a .com to a mobile-first company 152 00:07:26,954 --> 00:07:30,630 for Wego, my career grew and I became a Chief Product Officer 153 00:07:30,720 --> 00:07:33,780 very, very quickly before the age of 30. 154 00:07:34,179 --> 00:07:36,256 And I just started realizing that's where I can have the most 155 00:07:36,256 --> 00:07:39,019 impact. And I was good at, I think, 156 00:07:39,019 --> 00:07:41,727 identifying problems. And I've gotten better, um, as 157 00:07:41,727 --> 00:07:44,349 I've taken more and more challenging problems through 158 00:07:44,349 --> 00:07:47,269 Wego, FinAccel, Homage, and today Locofy. 159 00:07:47,829 --> 00:07:50,193 So yeah, it was just product management happened because 160 00:07:50,193 --> 00:07:51,939 someone told me you're a good product manager. 161 00:07:52,569 --> 00:07:55,669 And the more I read about it, yeah, it was just like, yeah, 162 00:07:55,669 --> 00:07:58,132 this is who I am. Uh, I come from a business 163 00:07:58,132 --> 00:08:01,119 family. So I naturally like ownership 164 00:08:01,119 --> 00:08:04,644 and solving problems. Yeah, I just started realizing 165 00:08:04,644 --> 00:08:07,580 if I can attract the best engineers, which I was able to 166 00:08:07,580 --> 00:08:10,516 do in the early in my career, I started realizing that I can 167 00:08:10,516 --> 00:08:14,251 have big impact in companies owning decisions and owning 168 00:08:14,251 --> 00:08:17,106 that, you know, engineering execution essentially. 169 00:08:17,627 --> 00:08:19,097 Right. Very inspiring. 170 00:08:19,217 --> 00:08:22,392 I think for people who sometimes felt that, I dunno. 171 00:08:22,392 --> 00:08:25,087 For example, if you are an engineer and you feel that 172 00:08:25,087 --> 00:08:26,642 engineering is not your world, right? 173 00:08:27,031 --> 00:08:29,949 It's worth to take a plunge and try out, maybe in this case, 174 00:08:29,949 --> 00:08:32,135 it's product management. And especially if you have 175 00:08:32,135 --> 00:08:34,765 feedback from people telling you, hey, you are good at in 176 00:08:34,765 --> 00:08:37,495 this area as well, right? So I think it's very inspiring 177 00:08:37,495 --> 00:08:39,881 to see that someone can transform their career. 178 00:08:40,871 --> 00:08:44,198 Even setting aside a goal, right, to achieve the role. 179 00:08:44,227 --> 00:08:47,302 I think that's where that comes through family upbringing, I 180 00:08:47,302 --> 00:08:50,387 think basically, right? Like the way I grew up was 181 00:08:50,387 --> 00:08:53,827 basically just say yes and then figure it out, right? 182 00:08:54,487 --> 00:08:59,066 I think that helped me adapt. And I still feel that, yeah, for 183 00:08:59,066 --> 00:09:02,937 any young listeners who feel they've done a degree and they 184 00:09:02,937 --> 00:09:05,983 don't necessarily enjoy it. If you're young, just try 185 00:09:05,983 --> 00:09:07,454 everything. And I've talked to a lot of 186 00:09:07,454 --> 00:09:10,631 folks who say, should I do an MBA and then get into product 187 00:09:10,631 --> 00:09:11,477 management? And I'm like, no. 188 00:09:11,477 --> 00:09:13,511 Just build something. Even if it doesn't work out. 189 00:09:13,511 --> 00:09:16,607 Even if it's, you know, that doesn't go beyond a few hundred 190 00:09:16,607 --> 00:09:19,055 users, it'll teach you what product management really is 191 00:09:19,055 --> 00:09:21,711 that you cannot learn from a book or from just lectures. 192 00:09:22,519 --> 00:09:25,969 So yeah, absolutely, any young people or even, uh, mid-career, 193 00:09:25,969 --> 00:09:30,429 I would say in this AI age today, it's becoming easier to 194 00:09:30,429 --> 00:09:35,461 find the resources and the tools to pick up new skills and new 195 00:09:35,461 --> 00:09:37,951 job descriptions. And job descriptions are getting 196 00:09:37,951 --> 00:09:39,661 transformed as you obviously know. 197 00:09:40,361 --> 00:09:42,073 So yeah. That's something I still stand 198 00:09:42,073 --> 00:09:43,697 by. Nice. 199 00:09:43,757 --> 00:09:46,304 So yeah, I think trying, experimenting, failing, I think 200 00:09:46,304 --> 00:09:48,933 it's part of the learning journey and the part of the 201 00:09:48,933 --> 00:09:50,807 growth, right? I mean, yeah, it could fail, I 202 00:09:50,807 --> 00:09:53,807 mean, but it could also succeed just like what you had. 203 00:09:53,807 --> 00:09:56,919 So you mentioned that you have worked in these three, I would 204 00:09:56,919 --> 00:10:00,317 say they are like big major startups, right? 205 00:10:00,317 --> 00:10:03,167 Maybe more scale ups rather than, you know, tiny startups. 206 00:10:03,467 --> 00:10:07,037 So what are the key lessons, if you can pick, you know, as a 207 00:10:07,037 --> 00:10:10,672 product, like a CPO or product manager, what are key lessons 208 00:10:10,672 --> 00:10:13,667 that you learn about building product, like great products 209 00:10:13,667 --> 00:10:16,187 from these various three companies that you have? 210 00:10:16,856 --> 00:10:21,301 I think product management... In the last five to seven years, 211 00:10:21,301 --> 00:10:24,161 I've seen a lot of young PMs saying product management is 212 00:10:24,161 --> 00:10:26,981 about gathering requirements, writing PRDs. 213 00:10:27,631 --> 00:10:29,436 Uh, it's not. It's just not. 214 00:10:29,896 --> 00:10:33,635 So I think for me, product management is about solving 215 00:10:33,635 --> 00:10:35,984 tough problems. Uh, and in this case, because we 216 00:10:35,984 --> 00:10:39,439 are a part of the tech industry, it's about solving, you know, 217 00:10:39,439 --> 00:10:43,723 tough problems that can be solved very efficiently and have 218 00:10:43,723 --> 00:10:47,899 massive impact through tech. So I think for me, the biggest 219 00:10:47,899 --> 00:10:51,989 learning has been throughout my career just how to identify the 220 00:10:51,989 --> 00:10:55,103 right problems. Because when you have good 221 00:10:55,103 --> 00:10:58,829 engineers, there is no shortage of a, for a company to, you 222 00:10:58,829 --> 00:11:02,453 know, like in a company to build applications and automations, 223 00:11:02,663 --> 00:11:05,335 but not everything is worth it. So I think you always have to 224 00:11:05,335 --> 00:11:08,854 keep an eye on a) understanding what impact means, b) I guess 225 00:11:08,854 --> 00:11:11,886 identifying the biggest problems, which is like one and 226 00:11:11,886 --> 00:11:16,522 two are loosely related. And then just building as fast 227 00:11:16,522 --> 00:11:19,549 as possible. I've never believed in 228 00:11:19,549 --> 00:11:23,543 documentation, which has, uh, maybe pissed off a few people 229 00:11:23,543 --> 00:11:24,663 that I've worked with in the past. 230 00:11:25,213 --> 00:11:28,827 But I always believe, I think just once you know a problem is 231 00:11:28,827 --> 00:11:31,873 worth solving, just go build it as quickly as possible. 232 00:11:32,173 --> 00:11:34,333 Take some shortcuts if you have to because it's still a 233 00:11:34,333 --> 00:11:36,153 hypothesis at that point of time. 234 00:11:36,403 --> 00:11:38,923 And you want to learn whether you're on the right path as 235 00:11:38,923 --> 00:11:41,310 quickly as possible. And then kind of course correct 236 00:11:41,310 --> 00:11:44,538 before going too big into like, here's my problem and I'm gonna 237 00:11:44,538 --> 00:11:47,082 spend two years before I get any feedback from the, from users. 238 00:11:47,952 --> 00:11:52,038 And number two for me. Another learning was I think 239 00:11:52,038 --> 00:11:57,117 asking your users for what they want and surveying them and 240 00:11:57,117 --> 00:11:58,782 feedbacking is actually overrated. 241 00:11:59,562 --> 00:12:02,230 I think sometimes people don't know what they... well, they 242 00:12:02,230 --> 00:12:06,372 might know what they want, but they may not necessarily know 243 00:12:06,372 --> 00:12:10,148 that beyond their, you know, needs, there are certain things 244 00:12:10,148 --> 00:12:12,508 that can actually have a much bigger impact. 245 00:12:12,508 --> 00:12:14,038 And the only way to find out is if you build. 246 00:12:14,098 --> 00:12:17,818 Like for example, in 2007-8, if you ask mobile phone users, what 247 00:12:17,818 --> 00:12:21,133 do you want? They would've said a slightly 248 00:12:21,133 --> 00:12:24,088 faster, slightly slimmer, more memory, more songs. 249 00:12:24,298 --> 00:12:26,098 And then no one would've said, I want an iPhone. 250 00:12:26,368 --> 00:12:31,738 You know, like a smartphone and like no keyboard on it and 251 00:12:32,608 --> 00:12:36,298 there's an app store around it. So I think that's what I've 252 00:12:36,298 --> 00:12:39,336 realized that spending time asking users for feedback and 253 00:12:39,336 --> 00:12:42,178 hoping that will give you the answer is a bit overrated. 254 00:12:42,178 --> 00:12:45,358 People sometimes just say yes because there's no cost to 255 00:12:45,358 --> 00:12:47,388 saying, hey, would you, if I build something like this, would 256 00:12:47,388 --> 00:12:48,813 you use it? People are like, yeah, sure. 257 00:12:49,653 --> 00:12:51,839 And then you spend six, nine months building it and no one 258 00:12:51,839 --> 00:12:54,825 uses it. So I think you just have to 259 00:12:54,825 --> 00:12:58,474 become better at reading all the data points around you and using 260 00:12:58,474 --> 00:13:01,773 your gut instinct as well. And the more you build, your gut 261 00:13:01,773 --> 00:13:04,605 gets stronger, obviously. But choosing the problem is, I 262 00:13:04,605 --> 00:13:08,347 would say, the more important part of product management than, 263 00:13:08,347 --> 00:13:11,427 you know, gathering requirements and documenting it. 264 00:13:12,267 --> 00:13:16,742 And I think in the AI age as writing PRDs and, you know, 265 00:13:16,742 --> 00:13:21,132 doing services becoming easier, a product manager's job is now 266 00:13:21,132 --> 00:13:24,674 even more about identifying the right problems where you may not 267 00:13:24,674 --> 00:13:27,609 even have the data, or you have the data, but it doesn't 268 00:13:27,609 --> 00:13:29,850 necessarily point you towards, hey, go and do this. 269 00:13:30,300 --> 00:13:32,992 So I think that's where a product manager should be first 270 00:13:32,992 --> 00:13:36,776 a problem solver, and problem solving requires to you choose 271 00:13:36,776 --> 00:13:39,000 the right problems and then solve it the right way. 272 00:13:39,470 --> 00:13:41,310 So, yeah, go down to the basics, basically. 273 00:13:41,611 --> 00:13:43,636 Right. I think very good key points 274 00:13:43,636 --> 00:13:45,779 that you mentioned. So first, of course, you need to 275 00:13:45,779 --> 00:13:47,971 identify the problems that you wanna solve, understand the 276 00:13:47,971 --> 00:13:50,053 impact, right? So problems and impact kind of 277 00:13:50,053 --> 00:13:53,122 like correlate each other. So you want to solve bigger 278 00:13:53,122 --> 00:13:56,236 impact problem, I would say, and not like small impact problems. 279 00:13:56,536 --> 00:13:59,884 And then, uh, obviously build fast, uh, as much as necessarily 280 00:13:59,884 --> 00:14:01,726 validated with users. Sometimes use your gut. 281 00:14:01,876 --> 00:14:03,886 I think I remember when you mentioned about this, I remember 282 00:14:03,886 --> 00:14:06,668 Henry Ford's quote, right? So if I ask people what they 283 00:14:06,668 --> 00:14:09,748 wanted back then, they wanted like a faster horse, not a car. 284 00:14:10,498 --> 00:14:12,338 So I think that, uh, reminded me of that quote. 285 00:14:12,748 --> 00:14:17,481 So throughout those three key experiences, do you have an 286 00:14:17,481 --> 00:14:19,274 example where you actually have to pivot? 287 00:14:19,454 --> 00:14:22,526 So for example, from, okay, I identify this is the problem I 288 00:14:22,526 --> 00:14:25,481 wanna build, but over the time after you release it maybe or 289 00:14:25,481 --> 00:14:28,520 get feedback from users, you actually have to pivot solving 290 00:14:28,520 --> 00:14:29,892 slightly more tangential problems. 291 00:14:30,192 --> 00:14:32,392 So is there such examples that you can share maybe? 292 00:14:32,662 --> 00:14:36,613 Uh, yeah. I think the most recent one I 293 00:14:36,613 --> 00:14:39,917 would say was uh, around Locofy itself. 294 00:14:40,387 --> 00:14:43,687 So when we started Locofy, I think we were very clear that 295 00:14:43,687 --> 00:14:46,394 design-to-code is a big problem. And I think, look, today, it's 296 00:14:46,394 --> 00:14:48,904 quite hot, I would say, and everyone agrees with it. 297 00:14:49,384 --> 00:14:54,494 But back in 2021, early pre-ChatGPT, pre-Stable 298 00:14:54,494 --> 00:14:58,730 Diffusion, pre-Copilot, to think you can go from design to code 299 00:14:58,730 --> 00:15:00,263 was considered a little bit crazy. 300 00:15:01,043 --> 00:15:04,312 And my co-founder and I were also a little bit, uh, naive in 301 00:15:04,312 --> 00:15:07,511 thinking we can do it for any design tool to any code 302 00:15:07,511 --> 00:15:09,768 framework very quickly. Doing the first step, building 303 00:15:09,768 --> 00:15:13,533 the MVP is easier. Getting it into production is 304 00:15:13,533 --> 00:15:16,744 super difficult. Uh, and it takes a lot of 305 00:15:16,744 --> 00:15:18,301 investment obviously, and a lot of focus. 306 00:15:18,571 --> 00:15:22,722 So I think for us, the pivot started from... we were app 307 00:15:22,722 --> 00:15:24,791 builders, right? So we were like, let's do 308 00:15:24,791 --> 00:15:27,992 building iOS apps and Android apps can be easy from designs or 309 00:15:27,992 --> 00:15:31,896 from design tools. And we had to almost like say, 310 00:15:31,896 --> 00:15:35,322 hey, do we build our own design tool in this case? 311 00:15:35,322 --> 00:15:36,852 No, let's not like reinvent the wheel. 312 00:15:37,002 --> 00:15:39,032 We will go with a existing design tool. 313 00:15:39,862 --> 00:15:43,968 And we did a simple website where people could sign up and 314 00:15:43,968 --> 00:15:48,146 we just started realizing just how big, not only web was, but 315 00:15:48,146 --> 00:15:52,081 also I think it was a turning point of, you know, web-based 316 00:15:52,081 --> 00:15:54,128 mobile app frameworks. Like, you know, obviously 317 00:15:54,128 --> 00:15:56,312 Flutter, React Native becoming a lot more popular. 318 00:15:56,522 --> 00:16:00,092 And we said, ah, it's okay. Let's just do web first and 319 00:16:00,092 --> 00:16:03,002 maybe in six months, we'll do Native iOS and Android. 320 00:16:03,572 --> 00:16:06,487 Four and a half years later today, we still haven't started 321 00:16:06,487 --> 00:16:09,346 working on it. Just because I think once you 322 00:16:09,346 --> 00:16:12,922 launch something, so to my point earlier about like, you know, 323 00:16:12,922 --> 00:16:17,086 surveys and asking users, that was more about like doing 324 00:16:17,086 --> 00:16:20,301 something before you build. But once you solve, once you 325 00:16:20,301 --> 00:16:23,914 kind of like launch your core problems first prototype or MVP 326 00:16:23,914 --> 00:16:28,574 after that I think we just got so much feedback, that our path 327 00:16:28,574 --> 00:16:31,318 became a lot more clear that we are going to build a... we are 328 00:16:31,318 --> 00:16:34,975 going to be a web first, uh, you know, design-to-code platform, 329 00:16:34,975 --> 00:16:37,943 front-end development platform, rather than just like app 330 00:16:37,943 --> 00:16:40,018 building that we initially thought about. 331 00:16:40,018 --> 00:16:42,085 And of course we will go and build that. 332 00:16:42,505 --> 00:16:45,745 But that was the first major sort of pivot. 333 00:16:45,745 --> 00:16:50,197 Second one was we, even though we were clear AI will play a big 334 00:16:50,197 --> 00:16:53,437 role. The early few prototypes in the 335 00:16:53,437 --> 00:16:56,955 first few MVPs and all was a lot of heuristics. 336 00:16:57,635 --> 00:17:00,505 And there were engineers in the teams who believed that you can 337 00:17:00,505 --> 00:17:01,795 solve this problem with heuristics. 338 00:17:02,425 --> 00:17:04,585 But I think we started seeing patterns and then we started 339 00:17:04,585 --> 00:17:06,714 building our own, uh, design models. 340 00:17:07,325 --> 00:17:11,589 The way and how quickly that has happened and now how much of 341 00:17:11,589 --> 00:17:14,689 our, not just revenue, but the core product sort of depends on 342 00:17:14,689 --> 00:17:18,612 it, I would not have anticipated or predicted that. 343 00:17:19,192 --> 00:17:21,694 Four and a half years back, we thought AI would play a support 344 00:17:21,694 --> 00:17:23,883 role. You know, like some basic tasks 345 00:17:23,883 --> 00:17:27,249 of going from design to code, picking up assets, making them 346 00:17:27,249 --> 00:17:30,522 work across different browsers, making them work across 347 00:17:30,522 --> 00:17:34,531 different sort of form factors. Maybe identifying some buttons 348 00:17:34,531 --> 00:17:37,290 and inputs. But today, it's like one click 349 00:17:37,290 --> 00:17:39,551 basically to go from design to code. 350 00:17:39,551 --> 00:17:43,185 And that has, that looks completely different from what 351 00:17:43,185 --> 00:17:44,831 we initially thought in the early days. 352 00:17:44,831 --> 00:17:47,801 And I think that's where that feedback of having something out 353 00:17:47,801 --> 00:17:51,291 in the open is a lot more valuable than asking people 354 00:17:51,291 --> 00:17:52,811 through like a form itself. Yeah. 355 00:17:53,744 --> 00:17:56,409 Again, like, uh, I think everyone who has built product 356 00:17:56,409 --> 00:17:58,833 would have the same opinion, I would say, right? 357 00:17:58,833 --> 00:18:01,483 Because once you throw something out in the market, right, the 358 00:18:01,483 --> 00:18:03,811 market will kind of determine exactly what products, and I 359 00:18:03,811 --> 00:18:05,643 think you will pivot along the way, right? 360 00:18:05,643 --> 00:18:09,348 So I think I've heard a lot of product companies that turn to 361 00:18:09,348 --> 00:18:11,185 successful, right? The initial story is kind of 362 00:18:11,185 --> 00:18:12,627 like different from what they're doing now. 363 00:18:12,652 --> 00:18:14,265 Oh, yeah. I mean, one more thing that I 364 00:18:14,265 --> 00:18:16,357 forgot to mention is when we started the company we said 365 00:18:16,357 --> 00:18:19,357 we'll never do enterprise. Like we will just be a bunch of 366 00:18:19,357 --> 00:18:22,202 engineers, product managers, you know, like have a $20 367 00:18:22,202 --> 00:18:24,363 subscription. And you know how every everyone 368 00:18:24,363 --> 00:18:26,312 starts, right? Like I just need a million 369 00:18:26,312 --> 00:18:28,977 people buying a $20 subscription and never have a sales person. 370 00:18:29,817 --> 00:18:31,167 But today, we are fully enterprise. 371 00:18:31,677 --> 00:18:33,747 Like 90% of our business is enterprise. 372 00:18:33,747 --> 00:18:37,296 And, yeah, we just started realizing, I think I would say 373 00:18:37,296 --> 00:18:40,684 early to mid last year that enterprises or enterprise 374 00:18:40,684 --> 00:18:43,624 engineers were using Locofy in free beta. 375 00:18:44,789 --> 00:18:47,937 They started asking us, hey, we love Locofy, but we can't use it 376 00:18:47,937 --> 00:18:50,829 in production in the company because we need a security audit 377 00:18:50,829 --> 00:18:52,769 and like, you need to have a proper pricing. 378 00:18:52,769 --> 00:18:55,834 We can't use free tools. And initially, we kept saying, 379 00:18:55,834 --> 00:18:57,839 sorry, no, we don't even have a salesperson, right? 380 00:18:57,839 --> 00:19:00,659 Like, whatever, you can use a tool for free. 381 00:19:01,139 --> 00:19:05,414 If not, no problems. And as we started benchmarking 382 00:19:05,414 --> 00:19:08,289 things across individual developers and enterprises, it 383 00:19:08,289 --> 00:19:11,849 just hit us on our face and we are like, we are an enterprise 384 00:19:11,909 --> 00:19:14,189 company whether we like it or not. 385 00:19:14,249 --> 00:19:17,369 The data suggests so, qualitative and quantitative. 386 00:19:17,999 --> 00:19:20,819 And today we are like a fully enterprise company. 387 00:19:21,499 --> 00:19:23,950 One and a half years ago, I actually told some investors 388 00:19:23,950 --> 00:19:26,270 I'll never do enterprise. And today, they're like, wait, 389 00:19:26,270 --> 00:19:29,296 what happened? So which is just part and parcel 390 00:19:29,296 --> 00:19:30,748 of obviously building a startup, right? 391 00:19:30,748 --> 00:19:33,805 Like you can have an end-goal in mind, but the, like the path you 392 00:19:33,805 --> 00:19:35,649 take to get there will never be a straight line. 393 00:19:36,890 --> 00:19:39,873 So I think this is also a good segue for people who haven't 394 00:19:39,873 --> 00:19:42,080 heard about Locofy. I think we have mentioned some 395 00:19:42,080 --> 00:19:43,830 interesting pivots that you've done with Locofy. 396 00:19:43,850 --> 00:19:47,924 But in the first place, maybe if you can explain what Locofy is. 397 00:19:48,135 --> 00:19:51,402 Like what kind of proposition that it has or what is the cool 398 00:19:51,402 --> 00:19:52,589 thing about Locofy? Yeah. 399 00:19:52,589 --> 00:19:55,718 I mean, in very simple terms, part of the product development 400 00:19:55,718 --> 00:19:58,854 journey, when you build a website or mobile app or 401 00:19:58,854 --> 00:20:02,563 dashboards or anything internal, external is, uh, you know, you 402 00:20:02,563 --> 00:20:06,006 spec out something, you design it in simple terms, and then 403 00:20:06,006 --> 00:20:08,470 those designs are obviously done on tools like Figma, which is 404 00:20:08,470 --> 00:20:10,861 very popular. There are other examples like 405 00:20:10,861 --> 00:20:14,836 Sketch, Adobe XD from a decade ago, Penpot for open source 406 00:20:14,836 --> 00:20:17,338 community. And these designs are done by 407 00:20:17,338 --> 00:20:19,960 designers. And then once, let's say, the 408 00:20:19,960 --> 00:20:22,870 stakeholders have kind of agreed to the design, it goes to 409 00:20:22,870 --> 00:20:25,108 engineers. Engineers, very, very, like 410 00:20:25,108 --> 00:20:27,940 simplifying, I guess it's just front-end, back-end engineers. 411 00:20:27,940 --> 00:20:31,063 Let's say front-end guys are more focused on bringing your 412 00:20:31,063 --> 00:20:34,481 designs to life through code and back-end engineers build 413 00:20:34,481 --> 00:20:35,991 everything. Infrastructure, back-end, 414 00:20:35,991 --> 00:20:38,992 servers, and security. And the list is obviously very 415 00:20:38,992 --> 00:20:42,669 long. We only focus on the automation 416 00:20:42,669 --> 00:20:47,694 of front-end itself, because we've seen in like a decade plus 417 00:20:47,694 --> 00:20:51,767 of our experience that as much as front-end development seems 418 00:20:51,767 --> 00:20:54,595 easy or designs ready, why not just have it ready in a couple 419 00:20:54,595 --> 00:20:59,067 of days, it's where 60 to 70% of our time and our best engineer's 420 00:20:59,067 --> 00:21:02,189 time was going. And yeah, so we just thought 421 00:21:02,189 --> 00:21:04,769 let's automate as much of that as possible directly from the 422 00:21:04,769 --> 00:21:07,173 design. So in short, Locofy converts 423 00:21:07,173 --> 00:21:11,591 your existing Figma designs and Penpot designs and Adobe XD 424 00:21:11,591 --> 00:21:14,777 designs into whichever sort of front-end framework. 425 00:21:15,035 --> 00:21:19,574 React, Angular, Vue, HTML, CSS, Flutter, React Native, pretty 426 00:21:19,574 --> 00:21:23,556 much anything. And uh, how it does that is we 427 00:21:23,556 --> 00:21:25,026 build our own proprietary technology. 428 00:21:25,386 --> 00:21:26,766 We call it the large design models. 429 00:21:27,459 --> 00:21:29,259 Because LLMs actually don't solve this problem. 430 00:21:29,319 --> 00:21:32,097 So we kind of trained our own models, deep learning techniques 431 00:21:32,097 --> 00:21:34,599 to go from designs to front-end code. 432 00:21:35,607 --> 00:21:38,802 We are not a no-code app. A lot of people think we are a 433 00:21:38,802 --> 00:21:39,867 no-code app builder. We are not. 434 00:21:40,077 --> 00:21:43,501 We are a low-code builder. We give you the code, but that's 435 00:21:43,501 --> 00:21:45,397 not complete. It's not the full app, 436 00:21:45,397 --> 00:21:47,668 obviously. You have to then go and refine 437 00:21:47,668 --> 00:21:50,687 it further and potentially like add, you know, your APIs and 438 00:21:50,687 --> 00:21:53,457 back-end logic. So we, Locofy is essentially a 439 00:21:53,457 --> 00:21:56,209 tool for front-end engineers, full stack engineers, anyone 440 00:21:56,209 --> 00:21:59,677 who's actually involved in building the front-end to cut 441 00:21:59,677 --> 00:22:05,473 the time by 60, 70, 80% by going from designs to a pixel perfect, 442 00:22:05,473 --> 00:22:09,142 developer friendly, you know, uh, responsive and interactive, 443 00:22:09,142 --> 00:22:12,981 uh, replica of that design rather than having to build it 444 00:22:12,981 --> 00:22:15,930 from scratch. Once you get that code, we have 445 00:22:15,930 --> 00:22:19,207 integrations into, you know, the Cursor of the world into GitHub, 446 00:22:19,207 --> 00:22:23,058 VS Code, uh, so that you can easily pull that code and keep 447 00:22:23,058 --> 00:22:25,915 merging it and integrate it continuously, uh, beyond that 448 00:22:25,915 --> 00:22:26,742 point. Yeah. 449 00:22:26,742 --> 00:22:30,895 So what we stand for is building amazing experiences with less 450 00:22:30,895 --> 00:22:34,019 effort and freeing up designers from, engineers especially, from 451 00:22:34,019 --> 00:22:36,252 grunt work of writing UI code from scratch. 452 00:22:36,782 --> 00:22:39,166 And letting them focus on the tougher front-end and back-end 453 00:22:39,166 --> 00:22:40,238 problems, yeah. Right. 454 00:22:40,928 --> 00:22:45,373 Maybe a bit of trivia, I would assume Locofy here stands for 455 00:22:45,373 --> 00:22:46,393 low-code-ify? Is that correct? 456 00:22:46,513 --> 00:22:50,100 Yes. we were, when we started the company, I think low-code, 457 00:22:50,100 --> 00:22:52,602 no-code was still a kind of nascent industry. 458 00:22:53,142 --> 00:22:55,572 And when I said we are gonna do design-to-code, uh, to one of my 459 00:22:55,572 --> 00:22:56,932 investors, he is like, oh, low-code. 460 00:22:56,952 --> 00:22:57,942 And I was like, what is that term? 461 00:22:57,942 --> 00:23:00,680 I've never heard of it. So I started researching and I 462 00:23:00,680 --> 00:23:03,771 think low-code became like, for a couple of weeks we were just 463 00:23:03,771 --> 00:23:06,838 saying low code so much that loco started becoming a little 464 00:23:06,838 --> 00:23:10,107 bit of a possibility. And like loco.ai would be nice. 465 00:23:10,107 --> 00:23:12,851 And it's a kind of crazy that we are doing a startup, and loco 466 00:23:12,851 --> 00:23:14,577 obviously means crazy. So it's... 467 00:23:14,577 --> 00:23:16,377 We look for loco.ai, but someone had it. 468 00:23:16,933 --> 00:23:18,883 And then we were like, hey, Shopify is pretty big. 469 00:23:19,393 --> 00:23:22,263 And there was this trend of like adding -ify in the end of 470 00:23:22,263 --> 00:23:23,983 everything. So we just call it Locofy. 471 00:23:24,414 --> 00:23:25,464 I see. Okay. 472 00:23:25,964 --> 00:23:29,226 I like that thinking process. So obviously, it's very 473 00:23:29,226 --> 00:23:31,458 promising, right? Everyone who has seen Locofy 474 00:23:31,458 --> 00:23:33,968 website, I assume, they will be thrilled, right? 475 00:23:33,998 --> 00:23:36,878 Oh, from sketches you can turn it into code, right? 476 00:23:37,178 --> 00:23:40,247 And you mentioned pixel perfect. But how much pixel perfect it 477 00:23:40,247 --> 00:23:42,183 is? Because sketch can be anything, 478 00:23:42,183 --> 00:23:44,332 right. From your visuals, colors, you 479 00:23:44,332 --> 00:23:47,228 know, maybe even animation, I dunno, into code. 480 00:23:47,228 --> 00:23:50,004 How much? Like is it 100% accurate, or is 481 00:23:50,004 --> 00:23:51,967 it like, how much? That's an amazing observation, 482 00:23:51,967 --> 00:23:57,454 you are obviously an engineer. So currently, about 80 to 85% of 483 00:23:57,454 --> 00:24:03,304 screens that are converted using Locofy do a 95% plus 484 00:24:03,304 --> 00:24:07,686 pixel-perfect representation. We basically look at every node, 485 00:24:07,686 --> 00:24:11,430 every layer, uh, once we render the code that Locofy produces 486 00:24:11,430 --> 00:24:15,120 and then we match it against the design to see how many of those 487 00:24:15,120 --> 00:24:17,040 elements have moved or don't look the same. 488 00:24:17,759 --> 00:24:21,940 But yeah, how easy is it is a different story because that 489 00:24:21,940 --> 00:24:25,212 depends a lot on how your designs have been structured. 490 00:24:25,572 --> 00:24:28,132 And to your point, right, like designs can be anything. 491 00:24:28,132 --> 00:24:30,322 And Figma became popular because of freeform. 492 00:24:30,382 --> 00:24:34,081 Like, you know, people can coordinate and like work 493 00:24:34,081 --> 00:24:36,945 together, multi collaborate. Sorry, multiple people can 494 00:24:36,945 --> 00:24:38,853 collaborate and build a design together. 495 00:24:39,523 --> 00:24:42,132 But there was no real importance, I think, in the past 496 00:24:42,132 --> 00:24:45,933 about how you structure those Figma layers and nodes. 497 00:24:46,796 --> 00:24:50,696 Whether you belong to the 80, 85% of accurate code, 498 00:24:50,696 --> 00:24:54,020 pixel-perfect code or not depends on not just how 499 00:24:54,020 --> 00:24:56,836 beautiful your designs are, more how they're structured. 500 00:24:57,326 --> 00:25:00,305 And especially to do with like, you know, removing unnecessary 501 00:25:00,305 --> 00:25:03,311 layers and nodes. Grouping things properly so that 502 00:25:03,311 --> 00:25:06,586 there's a relationship between them using best practices like 503 00:25:06,586 --> 00:25:09,026 auto layout. Naming your layers properly. 504 00:25:10,106 --> 00:25:13,756 If you do that, then the chances are you'll save 70 to 80% of 505 00:25:13,756 --> 00:25:16,696 your time. As we continuously see with our 506 00:25:16,696 --> 00:25:19,932 enterprise customers, when we go and train them, before we even 507 00:25:19,932 --> 00:25:23,976 train them on how to use Locofy, we first train them on how to 508 00:25:23,976 --> 00:25:26,496 structure your design, setting up your design files. 509 00:25:27,246 --> 00:25:29,728 Just like any AI, right? Like the input and output are 510 00:25:29,728 --> 00:25:32,745 related to each other. And if garbage goes in, then 511 00:25:32,745 --> 00:25:36,015 garbage comes out. So I think that's where Locofy 512 00:25:36,015 --> 00:25:39,479 does a tremendous job with people who are either already 513 00:25:39,479 --> 00:25:43,510 following the best practices on their Figma designs or Sketch 514 00:25:43,510 --> 00:25:46,522 designs. Or for the folks who say, this 515 00:25:46,522 --> 00:25:49,444 is a big enough problem, I can invest a little bit time 516 00:25:49,444 --> 00:25:51,817 upfront. Maybe half a day of my 517 00:25:51,817 --> 00:25:54,457 designer's time can save me four days of my engineer's time. 518 00:25:55,027 --> 00:25:58,701 That obviously equation is, quite hard to say no to. 519 00:25:59,391 --> 00:26:01,691 So it really depends on the input itself. 520 00:26:01,691 --> 00:26:04,826 And then the second part of it depends on how our models have 521 00:26:04,826 --> 00:26:07,208 been created. We spent about four years 522 00:26:07,208 --> 00:26:09,342 building our design models from scratch. 523 00:26:10,062 --> 00:26:14,382 And they're a combination of, people mistake them for, so 524 00:26:14,382 --> 00:26:17,270 which LLMs are you using? The answer is it's a completely 525 00:26:17,270 --> 00:26:19,478 different architecture. It's not based on language. 526 00:26:19,478 --> 00:26:21,888 It's based on visual understanding of designs. 527 00:26:22,398 --> 00:26:24,947 So whether it's to do with the design structure, whether it's 528 00:26:24,947 --> 00:26:28,917 to do with identifying, you know, HTML elements, uh, 529 00:26:28,917 --> 00:26:32,241 interactive elements, buttons, input tags, hundreds of them. 530 00:26:32,521 --> 00:26:34,834 Identifying the libraries, right, because you could be 531 00:26:34,834 --> 00:26:37,951 using Material and Bootstrap or your own design system. 532 00:26:39,073 --> 00:26:41,711 Then responsiveness, which is one of the toughest problems 533 00:26:41,711 --> 00:26:44,581 that again, goes back to how well your designs are 534 00:26:44,581 --> 00:26:46,831 structured. How it behaves on mobile, 535 00:26:46,831 --> 00:26:50,922 desktop, and tablet. Relationships have to be formed 536 00:26:50,922 --> 00:26:54,666 and identified between each element on your desktop website 537 00:26:54,666 --> 00:26:57,732 for it to work. So it really comes down to the 538 00:26:57,732 --> 00:27:00,682 quality of input and that's why we are doing really well with 539 00:27:00,682 --> 00:27:04,597 enterprises, I would say, more than the individual engineers 540 00:27:04,597 --> 00:27:06,103 today. Because individual engineers 541 00:27:06,103 --> 00:27:07,972 like things to just happen very fast. 542 00:27:08,752 --> 00:27:12,052 And then they also don't wanna go into designs and structure 543 00:27:12,052 --> 00:27:14,326 those designs. Whereas in enterprises, we talk 544 00:27:14,326 --> 00:27:17,332 to teams together and we organize the trainings and 545 00:27:17,332 --> 00:27:20,122 workshops, which enterprises are used to. 546 00:27:20,722 --> 00:27:24,892 They see massive benefits of investing some time into 547 00:27:24,892 --> 00:27:27,498 designs, because their roadblock is never design. 548 00:27:27,768 --> 00:27:31,408 It's always engineering. Whereas in young startups, I 549 00:27:31,408 --> 00:27:34,593 think, engineers, like I said, don't wanna go into Figma design 550 00:27:34,593 --> 00:27:37,806 and change things. And, uh, yeah, until they do 551 00:27:37,806 --> 00:27:40,398 that, they won't just, they won't save any time. 552 00:27:40,398 --> 00:27:43,366 And, most of them will say, I can do this page in two days 553 00:27:43,366 --> 00:27:46,656 myself. If I have to then invest an hour 554 00:27:46,656 --> 00:27:49,638 finding who the designer is, convince them to make the 555 00:27:49,638 --> 00:27:52,853 changes, then it doesn't work. Whereas in enterprises, we talk 556 00:27:52,853 --> 00:27:56,231 to teams where it works out because there's a common 557 00:27:56,231 --> 00:27:58,260 denominator. Generally, the VP of Design, VP 558 00:27:58,260 --> 00:28:00,628 of Engineering, who sees the benefit of doing this. 559 00:28:01,524 --> 00:28:03,409 Wow, I think that's very a great insight, right? 560 00:28:03,409 --> 00:28:06,216 Because yeah, like, now that you mentioned it, of course, if you 561 00:28:06,216 --> 00:28:09,401 are a few people startup, right? You don't have proper Figma 562 00:28:09,401 --> 00:28:11,869 design and all that. So yeah, I think that perfectly 563 00:28:11,869 --> 00:28:13,741 makes sense why you go into enterprise, right? 564 00:28:13,741 --> 00:28:16,551 Or like a bigger technologies organization set up, I would 565 00:28:16,551 --> 00:28:19,321 say, where you have designers and engineers, right? 566 00:28:19,330 --> 00:28:24,188 That one learning was, I would say, the biggest reason for our 567 00:28:24,188 --> 00:28:26,855 pivot towards enterprises. Because initially, we were like, 568 00:28:26,855 --> 00:28:28,395 yeah, people will just do designs properly. 569 00:28:29,175 --> 00:28:31,119 And we do a demo, and people are like, oh, this is great, but 570 00:28:31,119 --> 00:28:33,105 when I tried on my design, it doesn't work. 571 00:28:33,795 --> 00:28:36,165 And we say, oh, when we investigate, your design files 572 00:28:36,165 --> 00:28:40,130 are not done in a certain way. And then we would do videos and 573 00:28:40,130 --> 00:28:42,950 we do trainings and workshops and everything, but individual 574 00:28:42,950 --> 00:28:45,950 engineers, they don't like to do all of this, maybe. 575 00:28:46,430 --> 00:28:49,024 But with enterprises, we started realizing that when we were 576 00:28:49,024 --> 00:28:52,885 running an open beta, the larger teams would take interests, 577 00:28:52,885 --> 00:28:57,320 would come in like groups of 10. We would run like a one hour 578 00:28:57,320 --> 00:28:59,344 workshop and then they would start reporting that they're 579 00:28:59,344 --> 00:29:02,360 saving a lot of time. And we started just realizing 580 00:29:02,360 --> 00:29:05,864 there's a clear sign over here that larger enterprises are okay 581 00:29:05,864 --> 00:29:09,763 investing their time upfront, because they have the designer, 582 00:29:09,763 --> 00:29:13,492 engineers all together. And we, for a long time, we 583 00:29:13,492 --> 00:29:15,903 thought Figma itself would, you know, launch tools to optimize 584 00:29:15,903 --> 00:29:18,209 designs and they did in pieces here and there. 585 00:29:18,209 --> 00:29:20,249 We partnered with them, we gave them that feedback as well. 586 00:29:21,329 --> 00:29:25,199 But yeah, it still remains to be one of the biggest problems with 587 00:29:25,199 --> 00:29:28,279 design-to-code, right? People who expect it to work 588 00:29:28,279 --> 00:29:30,409 like magic will very likely be disappointed. 589 00:29:31,219 --> 00:29:36,109 But folks who've invested the time into the prerequisites have 590 00:29:36,114 --> 00:29:38,778 seen massive improvements and massive like time saving 591 00:29:38,778 --> 00:29:41,421 overall. Right. so that means like, uh, 592 00:29:41,421 --> 00:29:43,561 after hearing what you explained, right, definitely, it 593 00:29:43,561 --> 00:29:46,559 doesn't just use like a visual analysis, right? 594 00:29:46,619 --> 00:29:49,679 It also analyze like the Figma design itself, the metadata, the 595 00:29:49,679 --> 00:29:52,792 structure and all that. And then, uh, you also mentioned 596 00:29:52,792 --> 00:29:56,471 that the team most likely has a designer and engineering setup, 597 00:29:56,471 --> 00:29:58,288 right? So that typical setup. 598 00:29:58,730 --> 00:30:02,465 But one question, uh, remains. You mentioned about, you know, 599 00:30:02,465 --> 00:30:06,713 maybe 70%, 80% reduction. How long does it take, like for 600 00:30:06,713 --> 00:30:10,115 me to submit Figma files, or click one button? 601 00:30:10,205 --> 00:30:11,885 How long does it take for you to generate? 602 00:30:12,304 --> 00:30:13,119 The code itself? Yeah. 603 00:30:13,119 --> 00:30:16,419 I mean, not more than 20 to 30 seconds. 604 00:30:16,659 --> 00:30:19,822 Wow, okay. And you can select even three or 605 00:30:19,822 --> 00:30:25,577 four screens in the same go. We have customers who have built 606 00:30:25,577 --> 00:30:30,581 large projects of thousand plus screens in a matter of months 607 00:30:30,581 --> 00:30:33,433 using Locofy. And I think, um, in the 608 00:30:33,433 --> 00:30:36,361 background, there's just maybe like seven to 10 different 609 00:30:36,361 --> 00:30:39,193 models running. One is to optimize a design. 610 00:30:39,193 --> 00:30:41,083 One goes and finds those HTML elements. 611 00:30:41,083 --> 00:30:42,913 The other one goes and looks for responsiveness. 612 00:30:42,913 --> 00:30:46,123 And another one says, hey, let me find all the libraries and 613 00:30:46,123 --> 00:30:49,169 custom components you have. The other one says, hey, there's 614 00:30:49,169 --> 00:30:51,401 some repetition. Let me break your design into 615 00:30:51,401 --> 00:30:54,558 smaller, reusable components. The other one goes and breaks it 616 00:30:54,558 --> 00:30:56,203 into what are the props, for example. 617 00:30:56,846 --> 00:31:00,421 But all of these happen in parallel and they never take 618 00:31:00,421 --> 00:31:02,266 more than a few seconds, actually. 619 00:31:03,006 --> 00:31:06,998 Yeah. The time is actually spent on 620 00:31:06,998 --> 00:31:09,540 the pre-Locofy. So like I said again, right? 621 00:31:09,600 --> 00:31:12,394 You invest half an hour into cleaning up your design the 622 00:31:12,394 --> 00:31:14,699 right way. Locofy itself doesn't take much 623 00:31:14,699 --> 00:31:16,968 time. And of course, like the beyond 624 00:31:16,968 --> 00:31:19,668 the initial code generation, it'll make some mistakes. 625 00:31:19,848 --> 00:31:22,678 So we provide editor tools where you probably spend another half 626 00:31:22,678 --> 00:31:24,892 a day or whatever. But yeah. 627 00:31:25,222 --> 00:31:27,934 The equation that we have seen with some of our customers 628 00:31:27,934 --> 00:31:31,791 reporting back numbers is that an engineer who was able to do, 629 00:31:31,791 --> 00:31:35,914 you know, 10 designs or 10 screens per month is now able to 630 00:31:35,914 --> 00:31:38,889 do 25 to 30 after the initial sort of learning curve. 631 00:31:39,595 --> 00:31:42,648 But yeah, we wanna bring it down, obviously, even, uh, even 632 00:31:42,648 --> 00:31:44,263 more. I don't think people complain 633 00:31:44,263 --> 00:31:45,685 about the 20-30 seconds it takes. 634 00:31:46,320 --> 00:31:49,730 It really comes down to the initial code that gets 635 00:31:49,730 --> 00:31:51,920 generated. If it's too far away from what 636 00:31:51,920 --> 00:31:55,079 you have in mind, you're not gonna invest the 20% time to 637 00:31:55,079 --> 00:31:58,567 make it better. So it really comes down to doing 638 00:31:58,567 --> 00:31:59,970 all the pre-checks and everything. 639 00:31:59,970 --> 00:32:02,700 And that's why we are innovating a lot more right now, because we 640 00:32:02,700 --> 00:32:06,585 are realizing that it's still 9 out 10 organizations are not 641 00:32:06,585 --> 00:32:08,893 really doing designs a certain way. 642 00:32:09,652 --> 00:32:12,466 And we are teaching them, but we can also automate that part as 643 00:32:12,466 --> 00:32:13,162 well. Yeah. 644 00:32:13,388 --> 00:32:16,114 Nice. So you mentioned in the 645 00:32:16,114 --> 00:32:17,872 beginning about building this model, right? 646 00:32:18,323 --> 00:32:21,272 And I think four and a half years ago, LLM was not the craze 647 00:32:21,272 --> 00:32:24,188 as we have now. So basically your path is 648 00:32:24,188 --> 00:32:27,460 actually really like building the traditional ML, right? 649 00:32:27,460 --> 00:32:28,970 Like using neural network and all that. 650 00:32:28,970 --> 00:32:32,484 So tell us about, the decision, the rationale, why you need to 651 00:32:32,484 --> 00:32:35,502 build your own model. And after the LLM comes in, do 652 00:32:35,502 --> 00:32:38,558 you even try, you know, using the LLM that whether it can 653 00:32:38,558 --> 00:32:39,790 solve the problem? The evolution. 654 00:32:40,435 --> 00:32:45,848 So we launched our first beta product in 2022 Jan. 655 00:32:45,848 --> 00:32:48,520 And in there, we had a very simple model. 656 00:32:49,153 --> 00:32:51,913 Well, okay, the first initial alpha version is you would have 657 00:32:51,913 --> 00:32:55,248 to select, let's say a button in the design and say this is a 658 00:32:55,248 --> 00:32:56,653 button, and then add its properties. 659 00:32:56,653 --> 00:32:59,797 And then select one other one. Because your human eyes can look 660 00:32:59,797 --> 00:33:01,603 at a rectangle that says "Buy Now". 661 00:33:01,933 --> 00:33:04,848 And anyone can say that's a button, but it's just a 662 00:33:04,848 --> 00:33:07,003 designer's vector that constitutes a button. 663 00:33:07,573 --> 00:33:09,283 The engineer would still have to write the code for it. 664 00:33:09,853 --> 00:33:11,503 So we said, okay, let's make it this way. 665 00:33:11,503 --> 00:33:13,923 The designer or the engineer just selects that rectangle and 666 00:33:13,923 --> 00:33:16,513 say, this is a button and we will automate the code for that. 667 00:33:17,563 --> 00:33:19,603 We started doing that and we realized, hey, you know what? 668 00:33:19,633 --> 00:33:24,937 We have a lot of data. Can we train maybe like an 669 00:33:24,937 --> 00:33:28,430 image-based model or a CNN to understand? 670 00:33:28,430 --> 00:33:31,454 Because I mean, if it's a rectangle, it can be a button, 671 00:33:31,454 --> 00:33:34,510 input, daytime picker. All of them have a text and an 672 00:33:34,510 --> 00:33:37,991 icon and a rectangle. But if we have enough data and 673 00:33:37,991 --> 00:33:40,986 we can kind of like train certain of, these CNNs and 674 00:33:40,986 --> 00:33:43,759 transformers were kind of new back then as well, maybe we can 675 00:33:43,759 --> 00:33:46,558 get some results. And in the initial days, we were 676 00:33:46,558 --> 00:33:48,382 like, ah, it's kind of doing the job. 677 00:33:48,652 --> 00:33:52,252 It spits out and says it's a button or input or daytime 678 00:33:52,252 --> 00:33:54,166 picker. But you know, that's already 679 00:33:54,166 --> 00:33:56,092 better than asking a user to do it manually. 680 00:33:56,242 --> 00:34:00,082 So it was initially in our v1, it was more of a auto suggest. 681 00:34:00,942 --> 00:34:03,684 You select a rectangle or a Figma node and we'll say, hey, 682 00:34:03,684 --> 00:34:07,036 we think this is a button, 80% sure about it, but it could also 683 00:34:07,036 --> 00:34:09,012 be a datetime picker, which is like 60%. 684 00:34:09,612 --> 00:34:11,141 And they just had to say yes or no, right? 685 00:34:11,502 --> 00:34:12,672 So that's how we started doing it. 686 00:34:12,672 --> 00:34:15,688 And the more our customers did it, and in free beta, we told 687 00:34:15,688 --> 00:34:18,512 our users as well that we are going to use this data 688 00:34:18,512 --> 00:34:21,495 anonymized to train our models. The more we started doing that, 689 00:34:21,495 --> 00:34:24,510 the more we are like, hey, if we are hitting 90% accuracy, do we 690 00:34:24,510 --> 00:34:27,110 even need to suggest? Can we just not go and do it? 691 00:34:27,801 --> 00:34:31,257 So that was the first sort of AI that we started building on. 692 00:34:31,257 --> 00:34:34,063 It was purely for tagging. Understanding buttons input, 693 00:34:34,063 --> 00:34:36,628 datetime pickers, switches, you know, headers, footers. 694 00:34:37,422 --> 00:34:40,956 The second one came around design optimizations where we 695 00:34:40,956 --> 00:34:43,965 started realizing that we have to remove these unnecessary 696 00:34:43,965 --> 00:34:47,565 layers and groups and nodes. We started seeing some patterns 697 00:34:47,565 --> 00:34:50,246 and we said, hey, there are certain graph neural networks 698 00:34:50,246 --> 00:34:51,938 that can potentially help over here. 699 00:34:52,869 --> 00:34:55,663 Another one was, uh, around... These were just like three or 700 00:34:55,663 --> 00:34:58,869 four different models for very specific things. 701 00:34:59,348 --> 00:35:02,179 We saw patterns with our customer's data and talking to 702 00:35:02,179 --> 00:35:05,296 them as well. Another one was in the design, 703 00:35:05,296 --> 00:35:07,558 the designer says frame one, frame two, frame three. 704 00:35:07,858 --> 00:35:10,588 But in the actual design, it's maybe like a carousel. 705 00:35:11,398 --> 00:35:13,972 So in code, you don't wanna call it frame one, frame two, frame 706 00:35:13,972 --> 00:35:16,168 three, you wanna call it something else more useful. 707 00:35:16,498 --> 00:35:20,008 So for that, before ChatGPT was launched, actually, we were 708 00:35:20,008 --> 00:35:22,108 using OpenAI, right? So we'd heard of transformer 709 00:35:22,108 --> 00:35:24,764 models and we were, there were a bunch of APIs we could call at 710 00:35:24,764 --> 00:35:26,933 that point of time. So we started sending this 711 00:35:26,933 --> 00:35:28,940 information and saying, frame one, frame two, frame three, and 712 00:35:28,940 --> 00:35:31,198 this is the image. Tell me what this could be. 713 00:35:31,858 --> 00:35:34,449 And they come back and say, this could be a, actually a city 714 00:35:34,449 --> 00:35:37,338 card. Or it could be like a country 715 00:35:37,338 --> 00:35:40,293 carousel, whatever. We started naming the layers 716 00:35:40,293 --> 00:35:43,282 properly using that. And the more we started like 717 00:35:43,282 --> 00:35:45,332 building our own, sort of training our own models, 718 00:35:45,332 --> 00:35:48,410 different techniques, like we use a combination of transformer 719 00:35:48,410 --> 00:35:52,768 models, obviously, uh, sequence to sequence, YOLO, XGBoost, tons 720 00:35:52,768 --> 00:35:55,566 of them. And it was all experimental. 721 00:35:55,566 --> 00:35:57,854 There was a hypothesis that, look, this is all more image 722 00:35:57,854 --> 00:36:00,276 based or maybe give it like a CNN sort of a thing. 723 00:36:01,056 --> 00:36:03,570 And the more we started automating these smaller tasks, 724 00:36:03,570 --> 00:36:07,046 we started realizing that a user who clicks hundreds of times to 725 00:36:07,046 --> 00:36:10,791 go from design to code, maybe 80 of them, we can just go ahead 726 00:36:10,791 --> 00:36:13,950 and do it for them. And then instead of asking them 727 00:36:13,950 --> 00:36:17,982 to do it manually and learn from it only, we just go ahead and do 728 00:36:17,982 --> 00:36:21,491 it and then give them editor tools and see which ones they 729 00:36:21,491 --> 00:36:23,136 change. Which ones are not right. 730 00:36:23,526 --> 00:36:26,181 And that helped us also learn slightly faster and made the 731 00:36:26,181 --> 00:36:30,215 experience tremendously better. As we were launching this, 732 00:36:30,215 --> 00:36:32,953 ChatGPT launched. And we started hearing this term 733 00:36:32,953 --> 00:36:36,591 LLMs. And we are like, great, like if it can help us with the 734 00:36:36,591 --> 00:36:39,841 design to code, great, because we are not like a, we are not in 735 00:36:39,841 --> 00:36:41,751 the business of building a model and competing. 736 00:36:42,441 --> 00:36:44,910 We wanna solve front-end engineering, design-to-code 737 00:36:44,910 --> 00:36:48,144 problem. But the more we looked at LLMs, 738 00:36:48,144 --> 00:36:50,901 the more we realized the LLM architecture itself makes it 739 00:36:50,901 --> 00:36:56,291 super hard, if not impossible, to fine tune it or to prompt it 740 00:36:56,291 --> 00:37:01,131 to understand Figma designs. Because it's not built, I mean, 741 00:37:01,131 --> 00:37:03,396 it's built on understanding language and trying to predict 742 00:37:03,396 --> 00:37:04,980 the next word. Yeah, there's a certain 743 00:37:04,980 --> 00:37:07,191 structure to that. For designs, it's completely 744 00:37:07,191 --> 00:37:09,381 free form. Just because there's a button 745 00:37:09,381 --> 00:37:13,101 doesn't mean necessarily it'll always be next to a carousel or 746 00:37:13,101 --> 00:37:17,427 next to a header. Uh, LLMs struggled to understand 747 00:37:17,427 --> 00:37:20,377 that, but LLM solved some parts of the problem. 748 00:37:20,917 --> 00:37:23,427 So we are like, okay, we, the majority of understanding 749 00:37:23,427 --> 00:37:26,545 designs, we have to build our own design models, but we'll use 750 00:37:26,545 --> 00:37:30,097 LLMs as and when we can use them and we continue doing so even 751 00:37:30,097 --> 00:37:31,885 today. But in our case, going from 752 00:37:31,885 --> 00:37:36,459 design to code is about, I would say 80 to 85% of it, is how well 753 00:37:36,459 --> 00:37:40,369 we build these specialized models, uh, using, you know, any 754 00:37:40,369 --> 00:37:44,346 techniques really, and combine them together with LLMs. So for 755 00:37:44,346 --> 00:37:47,119 example, you go from design to code, LLMs may not understand 756 00:37:47,119 --> 00:37:50,516 designs very well. But once we spit out the code, 757 00:37:50,516 --> 00:37:52,136 LLMs actually understand code really well. 758 00:37:52,956 --> 00:37:58,038 So if I give, if Locofy gives out the code for a form that 759 00:37:58,038 --> 00:38:01,637 looks like a flight search form, we could actually ask an LLM to 760 00:38:01,637 --> 00:38:04,652 write the logic for it. And that's where we started 761 00:38:04,652 --> 00:38:07,668 saying, okay, you know what? Design to code is not just a 762 00:38:07,668 --> 00:38:10,182 matter of like click a button, get the code and go on with it. 763 00:38:10,602 --> 00:38:15,402 It's more about reducing 80% of the more, you know, things that 764 00:38:15,402 --> 00:38:19,080 can be automated where we can see a pattern, get it to code, 765 00:38:19,080 --> 00:38:21,012 and then see what LLMs can do with it. 766 00:38:21,642 --> 00:38:24,192 Potentially build our own vibe coding functionality. 767 00:38:24,732 --> 00:38:27,312 But then we started saying, hey, there's a MCP protocol. 768 00:38:27,312 --> 00:38:29,162 Like, you know, it's kind of becoming bigger. 769 00:38:29,622 --> 00:38:31,782 And the likes of Cursor, Windsurf are already investing 770 00:38:31,782 --> 00:38:35,718 so much with amazing minds to add functionality, code reviews, 771 00:38:35,718 --> 00:38:39,811 all the non-UI part. Then maybe that's the best 772 00:38:39,811 --> 00:38:43,251 marriage possible, right? Design to code with Locofy, and 773 00:38:43,251 --> 00:38:47,364 from code to more functionality and everything with LLMs. So the 774 00:38:47,364 --> 00:38:50,770 way we look at it is, rather than say LLMs or not, we think 775 00:38:50,770 --> 00:38:52,846 about what would an engineer do with at this step? 776 00:38:53,536 --> 00:38:55,306 Can we automate some parts of that problem? 777 00:38:55,966 --> 00:38:58,551 To automate that part of the problem, is it an LLM that can 778 00:38:58,551 --> 00:39:01,815 do it or our own models? And we go and build that rather 779 00:39:01,815 --> 00:39:04,868 than just say we are just gonna build an LLM, vibe coding tool 780 00:39:04,868 --> 00:39:08,366 because it's hot right now. So it's about breaking the 781 00:39:08,366 --> 00:39:11,266 problem into sub problems and seeing where, what fits. 782 00:39:11,826 --> 00:39:13,681 Wow. So thanks for sharing all this. 783 00:39:13,681 --> 00:39:16,901 I dunno how much is proprietary, but I think we get a peek into 784 00:39:16,901 --> 00:39:18,241 internal.. We actually have written a 785 00:39:18,241 --> 00:39:20,896 research paper about it. It's public, so yeah. 786 00:39:21,106 --> 00:39:23,554 So the design model part is proprietary, but we are very 787 00:39:23,554 --> 00:39:24,976 open about sharing how we build them, yeah. 788 00:39:25,336 --> 00:39:28,048 Nice. So maybe a little bit, uh, 789 00:39:28,048 --> 00:39:30,826 diving slightly deeper, right? So when you mentioned you help 790 00:39:30,826 --> 00:39:33,490 front-end engineering to come up with, uh, I don't know, like the 791 00:39:33,490 --> 00:39:35,026 first draft of the code, I assume, right? 792 00:39:35,356 --> 00:39:39,859 So, what was the stop? Like you produce a code, you 793 00:39:39,859 --> 00:39:42,588 have the visual elements, UI elements and all that, I assume. 794 00:39:42,948 --> 00:39:45,858 Then you have a hook points, I guess, like where the developers 795 00:39:45,858 --> 00:39:49,394 now come in to put it in logic. Or do you also fill in the logic 796 00:39:49,394 --> 00:39:51,338 and, you know, integrate with something, you know? 797 00:39:51,407 --> 00:39:53,977 No, current- First version was just we give you the code, you 798 00:39:53,977 --> 00:39:58,073 export it, and gone. Then we build a web app where 799 00:39:58,073 --> 00:40:01,032 you could view the code and you can kind of added some editor 800 00:40:01,032 --> 00:40:02,842 tools for you to change a few things. 801 00:40:03,652 --> 00:40:06,922 Then we started realizing that our customers, what do they do 802 00:40:06,922 --> 00:40:09,476 after they download the code? They figure out a way to open it 803 00:40:09,476 --> 00:40:10,622 in VS Code and add functionality. 804 00:40:10,622 --> 00:40:13,622 So we built a VS Code extension where with one command you can 805 00:40:13,622 --> 00:40:17,155 just pull that code and continue working on the code that Locofy 806 00:40:17,155 --> 00:40:19,880 produces. And lately, obviously with the 807 00:40:19,880 --> 00:40:23,269 Cursor and everything, we have an MCP where you can just 808 00:40:23,269 --> 00:40:27,123 convert a design to code, go to Cursor and say, hey, using my 809 00:40:27,123 --> 00:40:31,491 Locofy MCP, pull that code that I've generated for my flight 810 00:40:31,491 --> 00:40:35,021 homepage, add accessibility, clean it up, make the carousel 811 00:40:35,021 --> 00:40:37,433 functional. So that does beyond the 812 00:40:37,433 --> 00:40:39,820 front-end itself. And we have integrations to 813 00:40:39,820 --> 00:40:43,039 GitHub as well because you may be just building a small 814 00:40:43,039 --> 00:40:46,737 component or a section and push it directly to the right repo. 815 00:40:47,247 --> 00:40:50,157 And we then build a way for you to do continuous integration. 816 00:40:50,157 --> 00:40:53,289 So if you change something on the design, Locofy will identify 817 00:40:53,289 --> 00:40:56,817 that change and you will still make the choice of whether you 818 00:40:56,817 --> 00:41:00,192 want to just auto merge it or replace the older version of the 819 00:41:00,192 --> 00:41:04,196 code and kinda give you a three-way editor on GitHub to 820 00:41:04,196 --> 00:41:06,653 also select. So you can kind of keep going 821 00:41:06,653 --> 00:41:07,986 back because designs keep changing a lot. 822 00:41:09,148 --> 00:41:13,444 We are considering whether there are still some parts that we can 823 00:41:13,444 --> 00:41:17,252 do an LLM integration on the code that is generated before it 824 00:41:17,252 --> 00:41:21,126 goes into VS Code and other places where it's purely related 825 00:41:21,126 --> 00:41:26,211 to the UI that we can provide more tools, and maybe an LLM 826 00:41:26,211 --> 00:41:28,562 integration. But yeah, that's still something 827 00:41:28,562 --> 00:41:31,177 in the works. So we do only front-end and we 828 00:41:31,177 --> 00:41:34,252 make sure that that code is connected to your design system, 829 00:41:34,252 --> 00:41:37,370 connected to your designs, connected to your code repo and 830 00:41:37,370 --> 00:41:41,286 your developer tools, to provide like a very seamless way of not 831 00:41:41,286 --> 00:41:43,808 just doing it one time, but also coming back to it again and 832 00:41:43,808 --> 00:41:45,732 again. Yeah, I think the cycle, the 833 00:41:45,732 --> 00:41:48,502 loop here is very cool, right? Because, uh, many people would 834 00:41:48,502 --> 00:41:50,977 just assume like, okay, you turn into code and that's it. 835 00:41:50,977 --> 00:41:53,857 You just, you know, take it and, you know, transform things. 836 00:41:53,857 --> 00:41:56,772 But we know that design change, you know, sometimes, uh, user 837 00:41:56,772 --> 00:42:00,052 feedback gives, uh, like a certain, I dunno, changes to the 838 00:42:00,052 --> 00:42:01,592 design, right? And you have to iterate, right? 839 00:42:01,592 --> 00:42:04,425 I think it's really cool that you have all these integration 840 00:42:04,425 --> 00:42:05,116 points. Yeah. 841 00:42:05,206 --> 00:42:07,956 So for example, like, you know, merging code is something LLMs 842 00:42:07,956 --> 00:42:10,546 are amazing at. And that's why we use LLMs. 843 00:42:10,766 --> 00:42:13,985 Adding accessibility, we like, rather than we build a feature, 844 00:42:13,985 --> 00:42:16,048 actually LLMs do a really good job at it. 845 00:42:16,846 --> 00:42:18,946 Like we said earlier, right, like we see it as a end-to-end 846 00:42:18,946 --> 00:42:22,894 problem of automating front-end. Whether it requires us to build 847 00:42:22,894 --> 00:42:26,113 it using our proprietary models or LLMs, we don't necessarily 848 00:42:26,113 --> 00:42:28,350 care. So if LLMs get better at it, 849 00:42:28,350 --> 00:42:31,212 good for us, why build something that will be anyway, base model 850 00:42:31,212 --> 00:42:34,534 is already better than what we are spending six months on. 851 00:42:35,224 --> 00:42:38,843 But we also know LLMs cannot understand designs the same way 852 00:42:38,843 --> 00:42:41,901 it can understand legal jargon and, you know, create something 853 00:42:41,901 --> 00:42:42,999 from scratch itself. Yeah. 854 00:42:44,075 --> 00:42:46,665 So definitely I think people who have heard this will be very 855 00:42:46,665 --> 00:42:48,220 interested, especially front-end engineers, right? 856 00:42:48,220 --> 00:42:51,224 We are always very lazy translating, you know, design 857 00:42:51,224 --> 00:42:53,474 into code. You mentioned about, you know, 858 00:42:53,474 --> 00:42:56,054 what LLM is good at and what LLM is not good at. 859 00:42:56,054 --> 00:42:59,286 For other people who are also solving maybe a slightly similar 860 00:42:59,286 --> 00:43:02,489 problem where they think maybe LLM can solve the problem, but 861 00:43:02,489 --> 00:43:04,719 not quite right. What's the tipping point where 862 00:43:04,719 --> 00:43:07,589 you see like LLM is good at what, and LLM is not good at 863 00:43:07,589 --> 00:43:09,037 what? Maybe can give some suggestions 864 00:43:09,037 --> 00:43:12,719 based on your experience. Look, I won't claim to say I can 865 00:43:12,719 --> 00:43:16,439 make you a list of things it's not good at and things it is 866 00:43:16,439 --> 00:43:19,495 good at. But we understand obviously our 867 00:43:19,495 --> 00:43:23,170 space the most. And the answer lies to 868 00:43:23,170 --> 00:43:25,494 experimentation, right? Like if you think your problem 869 00:43:25,494 --> 00:43:29,145 can be solved with LLMs or can't be, for that matter, best to 870 00:43:29,145 --> 00:43:31,995 just run some experiments and see what it may be. 871 00:43:32,055 --> 00:43:34,811 You might get surprised by how good it is in some things that 872 00:43:34,811 --> 00:43:37,695 you never thought were possible and vice versa as well. 873 00:43:38,205 --> 00:43:42,105 And I think for us, the more we study LLMs in the underlying 874 00:43:42,105 --> 00:43:45,205 architecture itself and how it works, how does like this 875 00:43:45,205 --> 00:43:47,625 technology work. When I asked ChatGPT to do 876 00:43:47,625 --> 00:43:50,175 something, how does it even, what does it even follow? 877 00:43:50,415 --> 00:43:53,441 What technology it's built on? I think that's where we realized 878 00:43:53,441 --> 00:43:56,525 and we were like, there's absolutely no reason why a 879 00:43:56,525 --> 00:44:01,844 design would follow the same patterns as a research paper or 880 00:44:01,844 --> 00:44:06,052 code or books or tweets or posts, which is where LLMs shine 881 00:44:06,072 --> 00:44:07,452 and that's what it was built for. 882 00:44:08,370 --> 00:44:11,500 But I think just purely based on that knowledge itself, if you're 883 00:44:11,500 --> 00:44:15,449 building, let's say a legaltech company, LLMs will obviously do 884 00:44:15,449 --> 00:44:19,196 really well, wherever the, there's the, you know, language 885 00:44:19,196 --> 00:44:23,474 and text and sentence structure plays a big role, LLMs are 886 00:44:23,474 --> 00:44:27,438 obviously amazing. In the design context, they're 887 00:44:27,438 --> 00:44:31,249 still good in terms of producing patterns that have been built in 888 00:44:31,249 --> 00:44:33,305 the past. So asking an LLM to, for 889 00:44:33,305 --> 00:44:36,787 example, design a wallet. It can design a wallet for you, 890 00:44:36,787 --> 00:44:39,533 because there are hundreds of wallets in the world and it's 891 00:44:39,533 --> 00:44:43,265 been trained on all of them. But if you design that same 892 00:44:43,265 --> 00:44:46,073 wallet in Figma, the input changes completely. 893 00:44:46,133 --> 00:44:50,303 It's actually a json format. I mean, you obviously know how 894 00:44:50,303 --> 00:44:52,223 Figma nodes and layers structures work. 895 00:44:52,433 --> 00:44:55,950 That's where LLMs are just not made to understand that sort of 896 00:44:55,950 --> 00:44:59,015 parent child hierarchy that looks more like a graph than 897 00:44:59,015 --> 00:45:01,951 like a sentence itself. Or that's at least our 898 00:45:01,951 --> 00:45:04,433 understanding of it, right? So, yeah, I mean. 899 00:45:04,483 --> 00:45:08,808 But at the same time, like I've seen folks doing really, really 900 00:45:08,808 --> 00:45:11,193 well... Essentially, copywriting 901 00:45:11,193 --> 00:45:16,317 anything around marketing, creating images, LLMs are 902 00:45:16,317 --> 00:45:18,873 obviously amazing. Using the agentic sort of 903 00:45:18,873 --> 00:45:21,727 frameworks, you're starting to see a lot more of these menial 904 00:45:21,727 --> 00:45:24,637 tasks of going from one tab, selecting for something on your 905 00:45:24,637 --> 00:45:28,837 website, clicking it and buying it for you, but solving 906 00:45:28,837 --> 00:45:32,257 fundamental problems that require a different mindset or a 907 00:45:32,257 --> 00:45:34,867 different architecture than LLMs, of course, it can't. 908 00:45:35,309 --> 00:45:39,100 The newer models are starting to do specific things decently 909 00:45:39,100 --> 00:45:41,337 well. We can fine tune, for example, 910 00:45:41,337 --> 00:45:43,979 an LLM today to make it understand the relationship 911 00:45:43,979 --> 00:45:48,925 between parent, child and child up to a level of N-3. 912 00:45:49,365 --> 00:45:52,680 Anything beyond that, we see the results are quite crazy or super 913 00:45:52,680 --> 00:45:56,299 expensive, super slow. So I think in the context of 914 00:45:56,299 --> 00:46:01,150 design-to-code, I can just say that LLMs work for, let's say, 915 00:46:01,150 --> 00:46:04,655 understanding naming, potentially changing the code 916 00:46:04,655 --> 00:46:06,695 from, let's say React to Angular. 917 00:46:07,235 --> 00:46:10,745 But then understanding designs itself is something where it 918 00:46:10,745 --> 00:46:14,591 struggles. Yet a lot of people are still, I 919 00:46:14,591 --> 00:46:18,401 think using, they're going in with the hopes of like LLMs to 920 00:46:18,401 --> 00:46:20,085 solve every goddamn problem in the world. 921 00:46:20,435 --> 00:46:23,445 So let me just try it. And it maybe works with some 922 00:46:23,445 --> 00:46:26,345 basic websites, but that's why we are hearing from our 923 00:46:26,345 --> 00:46:28,547 customers as well that it doesn't understand it. 924 00:46:28,577 --> 00:46:30,467 We have copyright issues on top of that. 925 00:46:31,157 --> 00:46:34,523 We are not necessarily sure if it produces that wallet, can 926 00:46:34,523 --> 00:46:37,757 someone claim it was their wallet design that, you know, 927 00:46:37,757 --> 00:46:41,375 our large company that cares about its reputation will be 928 00:46:41,375 --> 00:46:44,825 obviously against? So yeah, I would say it really 929 00:46:44,825 --> 00:46:46,705 depends, man. It depends on the problem you 930 00:46:46,705 --> 00:46:48,947 are solving. Do experiments. 931 00:46:49,057 --> 00:46:51,187 Don't put all your eggs in just the LLM basket. 932 00:46:51,713 --> 00:46:54,326 And in most times, what you realize is it solves some 933 00:46:54,326 --> 00:46:55,778 problems. And then that's where the 934 00:46:55,778 --> 00:46:59,717 quality of how you fine tune it. But also fill that last mile 935 00:46:59,717 --> 00:47:03,640 gap, let's say, by building your own either heuristics or, you 936 00:47:03,640 --> 00:47:07,801 know, deep learning techniques. I think that's where the future 937 00:47:07,801 --> 00:47:10,593 of solving problems with LLMs is gonna be. 938 00:47:10,593 --> 00:47:13,793 That the initial POC and all looks very easy with the LLM. 939 00:47:14,313 --> 00:47:18,321 But as you go deeper, you have to build, you know, models 940 00:47:18,321 --> 00:47:21,163 around it or wrappers around it to kind of get the results. 941 00:47:22,110 --> 00:47:25,927 I'm not sure if I answer this question very well actually, 942 00:47:25,927 --> 00:47:29,985 because, um, yeah. It's just evolving so quickly. 943 00:47:30,795 --> 00:47:33,288 But also I think there's a lot more chatter now about the 944 00:47:33,288 --> 00:47:35,060 limitations of LLMs around the world. 945 00:47:35,060 --> 00:47:38,756 It doesn't understand how humans think how the world works. 946 00:47:38,756 --> 00:47:41,672 And that's why I think there's a lot of chatter around how LLMs 947 00:47:41,672 --> 00:47:45,708 cannot alone solve AGI, for example, and we have to think of 948 00:47:45,708 --> 00:47:48,776 a completely new way rather than just think scaling laws, just 949 00:47:48,776 --> 00:47:52,269 give the LLM as much information and more compute and it'll get 950 00:47:52,269 --> 00:47:55,256 better. There's recently news about even 951 00:47:55,256 --> 00:47:58,694 to get marginal, marginal improvements on LLM accuracy, 952 00:47:58,694 --> 00:48:02,370 you'll have to get 10 to the power a hundred, you know, that 953 00:48:02,370 --> 00:48:06,233 much sort of source of energy to even get that marginal 954 00:48:06,233 --> 00:48:07,844 improvement, which Earth even doesn't have. 955 00:48:08,504 --> 00:48:12,903 So I think, yeah, LLMs are going to start hitting a wall or 956 00:48:12,903 --> 00:48:16,926 rather ceiling. For us, we hit that ceiling a 957 00:48:16,926 --> 00:48:18,529 long time ago with design to code. 958 00:48:19,150 --> 00:48:23,271 But I think a lot of other startups that are betting big on 959 00:48:23,271 --> 00:48:25,884 it will find through their own experiments, which we are not 960 00:48:25,884 --> 00:48:27,636 very well obviously aware of. Right. 961 00:48:28,211 --> 00:48:29,987 Yeah. So I think, uh, from various 962 00:48:29,987 --> 00:48:33,083 conversations I had with, you know, people who are expert on 963 00:48:33,083 --> 00:48:36,591 AI, so they also think that using the current LLM 964 00:48:36,591 --> 00:48:39,081 transformer approach may not be the path to AGI. 965 00:48:39,521 --> 00:48:41,926 But I mean, who knows? Like this field, you know, 966 00:48:41,926 --> 00:48:45,089 changes so rapidly. Every few days we will see new 967 00:48:45,089 --> 00:48:46,581 advancements. But yeah, thanks for sharing 968 00:48:46,581 --> 00:48:49,067 your perspective. So I assume building the large 969 00:48:49,067 --> 00:48:51,785 design model is partly one of the most complex engineering 970 00:48:51,785 --> 00:48:55,309 tasks that you have. Is there any other thing that is 971 00:48:55,309 --> 00:48:58,891 un-intuitively, complex as well for us to, you know, maybe learn 972 00:48:58,891 --> 00:49:00,478 about from, from your experience? 973 00:49:00,478 --> 00:49:02,381 Yeah. I guess I've already touched 974 00:49:02,381 --> 00:49:04,461 upon it. But right now, I think the 975 00:49:04,461 --> 00:49:06,567 biggest problem in the design-to-code space, not just 976 00:49:06,567 --> 00:49:08,970 for Locofy, and the biggest opportunity as well, while 977 00:49:08,970 --> 00:49:12,536 there's a lot of chatter around, hey, designer, not only 978 00:49:12,536 --> 00:49:14,511 designers, but anyone can design. 979 00:49:15,352 --> 00:49:18,756 And do you even need Figma? You can just go to Lovable, Bolt 980 00:49:18,756 --> 00:49:20,392 and just design something. It's absolutely true. 981 00:49:20,392 --> 00:49:26,014 It solves a lot of use cases. But we still think, we obviously 982 00:49:26,014 --> 00:49:30,468 have a strong bias towards advanced production products 983 00:49:30,468 --> 00:49:34,712 will require proper designs and not just an LLM producing the 984 00:49:34,712 --> 00:49:37,402 same results that your competitor can get as well. 985 00:49:37,522 --> 00:49:41,402 It is becoming commoditized. And what happens when everyone 986 00:49:41,402 --> 00:49:44,982 has the same looking website and app, is that people go back to 987 00:49:44,982 --> 00:49:47,753 saying, I will do it very custom made and designers become cool 988 00:49:47,753 --> 00:49:51,252 and design tools still have very much stay relevant and continue 989 00:49:51,252 --> 00:49:53,994 growing strong actually. So that's where we still think 990 00:49:53,994 --> 00:49:57,159 that the problem I mentioned earlier about design structure. 991 00:49:58,145 --> 00:50:04,385 It holds the key to a lot of the front-end automation around. 992 00:50:05,015 --> 00:50:06,875 And that is one of the toughest problems to solve. 993 00:50:07,280 --> 00:50:10,927 Because, uh, design structure, there are some patterns that can 994 00:50:10,927 --> 00:50:14,763 be automated. But even if you automate 80% and 995 00:50:14,763 --> 00:50:18,576 the remaining 20% doesn't work, for the user to even identify 996 00:50:18,576 --> 00:50:23,149 which part of my design is in that 20% that didn't work is one 997 00:50:23,149 --> 00:50:25,035 of the problems we are working on right now. 998 00:50:25,515 --> 00:50:29,365 And we think once we can get there, design-to-code can become 999 00:50:29,365 --> 00:50:33,186 a lot more mainstream, because of the dependency. 1000 00:50:33,186 --> 00:50:35,955 And I think especially with this notion of like, why should I do 1001 00:50:35,955 --> 00:50:38,606 the work as a designer if the engineer benefits from it? 1002 00:50:39,296 --> 00:50:44,070 If we can get the designers' existing designs as they are 1003 00:50:44,070 --> 00:50:47,476 without any requirements for them to go and manually 1004 00:50:47,476 --> 00:50:50,684 restructure it, to then go and restructure it with an 1005 00:50:50,684 --> 00:50:53,538 automation and then, you know, design to code works really 1006 00:50:53,538 --> 00:50:56,953 well, that I think will be the toughest problem in the 1007 00:50:56,953 --> 00:50:58,649 industry. And I think for, definitely, for 1008 00:50:58,649 --> 00:51:00,099 what we stand for. Yeah. 1009 00:51:00,309 --> 00:51:02,689 Since you brought it up, I think it's also a good point to 1010 00:51:02,689 --> 00:51:04,881 discuss about it, right? Because in the industry, with 1011 00:51:04,881 --> 00:51:07,956 the introduction of AI, almost every engineers are kinda like 1012 00:51:07,956 --> 00:51:09,561 concerned about their role, right? 1013 00:51:09,561 --> 00:51:11,991 In this case, for your particular use cases like 1014 00:51:11,991 --> 00:51:13,941 designers and front-end engineers, right? 1015 00:51:14,201 --> 00:51:17,369 So now I would imagine designers have the ability to produce code 1016 00:51:17,369 --> 00:51:19,551 in some sense, and maybe can tweak it along the way. 1017 00:51:19,851 --> 00:51:23,061 And for front-end engineers, um, you know, they have design 1018 00:51:23,061 --> 00:51:25,446 capability as well, right? But obviously some people are 1019 00:51:25,446 --> 00:51:28,915 concerned, you know, like, oh, does it mean we have lesser need 1020 00:51:28,915 --> 00:51:30,631 for designers and front-end engineer? 1021 00:51:30,931 --> 00:51:33,591 What do you think about, you know, the future of development, 1022 00:51:33,591 --> 00:51:34,786 engineering, you know? Yeah. 1023 00:51:34,786 --> 00:51:36,856 And this is a topic I'm very, very passionate about as well. 1024 00:51:37,109 --> 00:51:40,229 My co-founder and I did a talk at Figma Config called the Rise 1025 00:51:40,229 --> 00:51:43,146 of the Design Engineer just, 3, 3, 4 months ago. 1026 00:51:43,806 --> 00:51:47,296 So I do believe that, uh, there are gonna be changes, and they, 1027 00:51:47,296 --> 00:51:49,578 I mean, they, changes are already kind of happening live, 1028 00:51:49,578 --> 00:51:52,160 right? Still some, uh, I'm not still 1029 00:51:52,160 --> 00:51:56,490 100% convinced on some of these, uh, claims made that no more 1030 00:51:56,490 --> 00:51:58,980 designers are needed, and then designers can do front-end 1031 00:51:58,980 --> 00:52:01,095 engineering. But I do generally agree with 1032 00:52:01,095 --> 00:52:03,950 the notion that designers can do the front-end development, and 1033 00:52:03,950 --> 00:52:06,919 they have absolutely every reason and all the tools 1034 00:52:06,919 --> 00:52:09,791 available today to do it. The reason is simply because if 1035 00:52:09,791 --> 00:52:12,908 you look at a designer, they do the designs and then they gave 1036 00:52:12,908 --> 00:52:14,455 it to engineers. And then traditionally, two 1037 00:52:14,455 --> 00:52:17,045 months, three months later, the engineer would come back and 1038 00:52:17,045 --> 00:52:19,331 say, hey, I've built it, send it back to the designer. 1039 00:52:19,361 --> 00:52:22,158 And then the designer goes, this is nowhere close to what I 1040 00:52:22,158 --> 00:52:24,490 design. And then that back and forth 1041 00:52:24,490 --> 00:52:26,956 between designers and engineers to essentially produce the same 1042 00:52:26,956 --> 00:52:29,718 thing. But for the designer, what it 1043 00:52:29,718 --> 00:52:33,153 means is frustration that I, this is was my vision of what I 1044 00:52:33,153 --> 00:52:35,942 wanted to design, and then I have to keep talking to this 1045 00:52:35,942 --> 00:52:37,764 engineer and try to get it to work that way. 1046 00:52:38,244 --> 00:52:40,944 And they want ownership. So when we've talked to 1047 00:52:40,944 --> 00:52:43,716 designers, we've said, look, if you, for the designers who are 1048 00:52:43,716 --> 00:52:47,134 using Locofy, at least this is what they tell us, that I'm not 1049 00:52:47,134 --> 00:52:49,879 a expert at front-end but I understand front-end techniques 1050 00:52:49,879 --> 00:52:52,298 enough. And secondly, I use it because I 1051 00:52:52,298 --> 00:52:56,334 get that code and I can make sure that that code when it runs 1052 00:52:56,334 --> 00:52:59,584 on the user side is actually showing exactly the design I 1053 00:52:59,584 --> 00:53:01,897 design through code. Whereas earlier, I had to kinda 1054 00:53:01,897 --> 00:53:05,444 rely on the engineer. So I think, yeah, that's why we 1055 00:53:05,444 --> 00:53:09,999 think that the designer's role will evolve and should evolve. 1056 00:53:10,419 --> 00:53:14,003 Also for the other reason that if you look at the last 15 years 1057 00:53:14,003 --> 00:53:16,764 or so, engineering salaries have tripled, quadrupled because 1058 00:53:16,764 --> 00:53:18,759 there's a shortage of developers. 1059 00:53:19,539 --> 00:53:23,087 Designers, on the other hand, I think there's enough supply of 1060 00:53:23,087 --> 00:53:25,054 designers. I have never heard of a company 1061 00:53:25,054 --> 00:53:27,489 dying because, hey, I never found a designer, right? 1062 00:53:27,849 --> 00:53:29,369 There obviously, there's a spectrum, right? 1063 00:53:29,649 --> 00:53:32,769 Really good designers are still difficult, but you can find 1064 00:53:32,769 --> 00:53:35,426 designers with a lot more ease. And that's why designers 1065 00:53:35,426 --> 00:53:38,962 salaries have not grown anywhere compared to engineers in the 1066 00:53:38,962 --> 00:53:41,682 last decade. And now with design automation 1067 00:53:41,682 --> 00:53:44,477 coming up, I think designers feel even more like squeezed 1068 00:53:44,477 --> 00:53:47,180 from both sides. That a) there's a lot of 1069 00:53:47,180 --> 00:53:49,172 designers and how come I'm not getting a seat at the table? 1070 00:53:49,648 --> 00:53:52,738 My salaries are not rising. And the second thing is now, oh, 1071 00:53:52,738 --> 00:53:55,609 whoa, hold on. The role of a design intern can 1072 00:53:55,609 --> 00:53:59,648 now be done by, you know, like Figma and Stitch, for example. 1073 00:53:59,668 --> 00:54:02,068 There's so many tools. UIzard, as it's called. 1074 00:54:02,698 --> 00:54:05,757 So I think for designers, it's almost, I think, inevitable that 1075 00:54:05,757 --> 00:54:09,581 they will end up owning more partly because of the need for 1076 00:54:09,581 --> 00:54:13,393 them to kind of own the front-end, and partly because 1077 00:54:13,393 --> 00:54:17,551 they themselves realize that, you know, I need to be doing 1078 00:54:17,551 --> 00:54:19,579 more. I need to be having more impact. 1079 00:54:20,269 --> 00:54:22,759 And today I have the tools, so why not? 1080 00:54:22,969 --> 00:54:26,079 But of course, it'll still require them to have, like learn 1081 00:54:26,079 --> 00:54:30,214 some front-end expertise because no AI is magic where it works 1082 00:54:30,214 --> 00:54:32,986 until it works. But when it breaks, you need to 1083 00:54:32,986 --> 00:54:35,895 know where it broke and even realize how to fix it after 1084 00:54:35,895 --> 00:54:37,795 that. So that's where I've been urging 1085 00:54:37,795 --> 00:54:40,287 at least the designers in my team to start learning front-end 1086 00:54:40,287 --> 00:54:42,877 concepts. Not be like front-end gurus, but 1087 00:54:42,877 --> 00:54:45,973 at least understand concepts enough that if they're using a 1088 00:54:45,973 --> 00:54:50,365 tool, they don't feel scared. They can own it and they can 1089 00:54:50,365 --> 00:54:52,624 actually, they don't have to keep going to engineers to keep 1090 00:54:52,624 --> 00:54:54,699 asking for help. So I think from design 1091 00:54:54,699 --> 00:54:57,755 engineering point of view, I feel there's already a 1092 00:54:57,755 --> 00:55:01,020 revolution. But it doesn't mean that, you 1093 00:55:01,020 --> 00:55:04,569 know, the really experienced, high functioning, you know, 1094 00:55:04,569 --> 00:55:09,179 designers who build complex systems should worry about this 1095 00:55:09,179 --> 00:55:11,581 automation, because this automation is just producing. 1096 00:55:12,061 --> 00:55:14,041 Like in Japan, there's a term called Wabi-sabi, right? 1097 00:55:14,041 --> 00:55:17,761 Which is, look, if everyone in your neighborhood, everyone, all 1098 00:55:17,761 --> 00:55:20,519 your friends have the same table from IKEA, then you kind of want 1099 00:55:20,519 --> 00:55:22,051 to make something custom made, right? 1100 00:55:22,321 --> 00:55:24,373 And I think that's essentially what's going to happen with a 1101 00:55:24,373 --> 00:55:28,267 lot of these design automations that real designers and advanced 1102 00:55:28,267 --> 00:55:32,661 design tools will still be very much hot and in demand. 1103 00:55:33,441 --> 00:55:36,469 On the engineering side, I do feel that there's been always 1104 00:55:36,469 --> 00:55:39,949 been full-stack engineers. But I think there was a time 1105 00:55:39,949 --> 00:55:41,887 when having just front-end engineers and back-end engineers 1106 00:55:41,887 --> 00:55:44,819 made sense. I do think the lines will blur a 1107 00:55:44,819 --> 00:55:48,232 little bit over there. So I don't think of it as like 1108 00:55:48,232 --> 00:55:51,048 jobs will go away, but team sizes and roles and 1109 00:55:51,048 --> 00:55:54,088 responsibilities I think will merge a little bit. 1110 00:55:54,108 --> 00:55:56,455 I mean, Figma did it to a large extent, if you think about it. 1111 00:55:56,515 --> 00:55:59,323 Product managers and designers started designing together, and 1112 00:55:59,323 --> 00:56:01,375 they brought them kind of close to each other. 1113 00:56:01,795 --> 00:56:03,901 What we are doing can bring designers and front-end 1114 00:56:03,901 --> 00:56:05,035 engineers very close to each other. 1115 00:56:05,755 --> 00:56:09,230 And what Cursor kind of does, allows front-end engineers to do 1116 00:56:09,230 --> 00:56:12,433 more full stack. So it's less of replacing people 1117 00:56:12,433 --> 00:56:15,460 and more of like giving them the opportunity to make themselves 1118 00:56:15,460 --> 00:56:18,610 more full stack. That's I think something, yeah, 1119 00:56:18,610 --> 00:56:20,650 it's inevitable. And product managers as well, 1120 00:56:20,650 --> 00:56:23,547 not just writing specs. You can solve your problems, do 1121 00:56:23,547 --> 00:56:27,043 the analysis, create the first mockup, even if it's done using 1122 00:56:27,043 --> 00:56:31,201 LLM and everyone gets the same result, it's enough to get an 1123 00:56:31,201 --> 00:56:34,010 approval, for example. Then design it again from 1124 00:56:34,010 --> 00:56:35,749 scratch, right? So where versus earlier, a 1125 00:56:35,749 --> 00:56:39,092 product manager would go to a lead UX researcher to go do a 1126 00:56:39,092 --> 00:56:42,672 research, a business PM to write the business use case, then a 1127 00:56:42,672 --> 00:56:45,526 designer to do the design, and that would just slow them down. 1128 00:56:45,526 --> 00:56:49,099 Today, I think any product manager should absolutely be 1129 00:56:49,099 --> 00:56:52,449 100% into AI and rather than having 10 in the team, you can 1130 00:56:52,449 --> 00:56:55,860 have probably two or three. So yeah, I think roles will 1131 00:56:55,860 --> 00:57:00,174 transform, but I don't think it's gonna be as crazy as it 1132 00:57:00,174 --> 00:57:04,374 sounds that proper jobs will just completely be wiped out. 1133 00:57:05,235 --> 00:57:08,644 I think one key message I would say that there will be more 1134 00:57:08,644 --> 00:57:12,352 blending of skillset, right? So most likely, the adjacent 1135 00:57:12,352 --> 00:57:14,346 thing, right? So designer and front-end 1136 00:57:14,346 --> 00:57:16,666 engineer, maybe front-end engineer with design and 1137 00:57:16,666 --> 00:57:18,712 backend, right? So definitely people need to 1138 00:57:18,712 --> 00:57:21,004 upskill themselves and play around with these tools, right? 1139 00:57:21,274 --> 00:57:23,494 So I like the term design engineer you mentioned. 1140 00:57:23,704 --> 00:57:26,250 Maybe one day we'll start seeing more roles design engineer, 1141 00:57:26,250 --> 00:57:28,618 right? My team has two designers. 1142 00:57:29,308 --> 00:57:32,428 When the growth team basically wants to create some pages for 1143 00:57:32,428 --> 00:57:35,883 SEO, AIO, for whatever, right? Now, the designers take that 1144 00:57:35,883 --> 00:57:39,463 design, use Locofy, get the code, pull it into Cursor, just 1145 00:57:39,463 --> 00:57:41,314 prompt, explain, and then publish it. 1146 00:57:41,314 --> 00:57:43,564 And maybe an engineer needs 10 minutes to just review. 1147 00:57:44,194 --> 00:57:48,094 So we are already freeing up the engineer's time from slightly 1148 00:57:48,094 --> 00:57:51,284 lower value tasks so they can focus on the core problems 1149 00:57:51,284 --> 00:57:53,458 itself. And we are now starting to say 1150 00:57:53,458 --> 00:57:54,994 we don't have a product manager, for example. 1151 00:57:55,550 --> 00:57:57,680 Our solution architects are the product managers. 1152 00:57:57,699 --> 00:57:59,913 They can write. I mean, writing a PRD, I've 1153 00:57:59,913 --> 00:58:01,029 never considered as product management. 1154 00:58:01,269 --> 00:58:02,499 But they're talking to customers. 1155 00:58:02,499 --> 00:58:05,229 Why should there then be a product manager? 1156 00:58:05,859 --> 00:58:08,096 The more number of people you add, the more the complexity, 1157 00:58:08,096 --> 00:58:10,124 right? So our solution architects who 1158 00:58:10,124 --> 00:58:12,849 help customers with their training and workshops can also 1159 00:58:12,849 --> 00:58:16,642 come back and work directly with our designers and engineers. 1160 00:58:16,882 --> 00:58:19,937 So I think, yeah, these unnecessary roles that have been 1161 00:58:19,937 --> 00:58:22,882 created during the height of the pandemic, I think, yeah, will go 1162 00:58:22,882 --> 00:58:26,038 back. No more Business PM, Technical 1163 00:58:26,038 --> 00:58:30,280 PM, Lead UX researcher, UI designer, UX designer, could 1164 00:58:30,280 --> 00:58:32,242 just be a product manager and a designer. 1165 00:58:32,812 --> 00:58:35,332 So I think that's where things will be just more efficient and 1166 00:58:35,332 --> 00:58:35,692 faster. Yeah. 1167 00:58:35,863 --> 00:58:37,944 Yeah. So I think it was also brought 1168 00:58:37,944 --> 00:58:39,448 up in one of my conversations before, right? 1169 00:58:39,478 --> 00:58:42,598 So we used to have specializations simply because, 1170 00:58:42,598 --> 00:58:45,835 like almost every part of the software engineering is like 1171 00:58:45,835 --> 00:58:47,878 very complex, right? So if you think about it, right? 1172 00:58:47,878 --> 00:58:50,051 So front-end itself, you know, you have so many frameworks, you 1173 00:58:50,051 --> 00:58:52,026 have to understand maybe JavaScript, TypeScript, whatever 1174 00:58:52,026 --> 00:58:53,643 that is. You have to understand the 1175 00:58:53,643 --> 00:58:54,868 tools, the ecosystem and all that. 1176 00:58:54,868 --> 00:58:58,028 It becomes like very complex and you need specialized skills. 1177 00:58:58,283 --> 00:59:00,593 And now with maybe the help of AI, right? 1178 00:59:00,753 --> 00:59:04,722 So the specialization becomes blurry and it's turning into 1179 00:59:04,722 --> 00:59:07,122 more generalization. This is where the trend is going 1180 00:59:07,122 --> 00:59:09,411 as well, I believe. And, uh, I would say that people 1181 00:59:09,411 --> 00:59:12,558 need to be able to play around with these tools, which is my 1182 00:59:12,558 --> 00:59:14,888 next question. With so many proliferation of 1183 00:59:14,888 --> 00:59:18,837 tools, you know, you have maybe low-code, no-code, uh, you have 1184 00:59:18,837 --> 00:59:21,858 AI driven tools, right? Well, what's the future of this? 1185 00:59:21,858 --> 00:59:24,987 Actually, I'm sometimes very confused, like where should we 1186 00:59:24,987 --> 00:59:27,092 invest our effort and time? Yeah. 1187 00:59:28,052 --> 00:59:31,396 I don't like to make bold predictions as if I know what's 1188 00:59:31,396 --> 00:59:33,764 gonna happen. Two years ago, we, I wouldn't 1189 00:59:33,764 --> 00:59:35,514 have predicted tool like Cursor, for example. 1190 00:59:36,127 --> 00:59:42,264 I do think that engineering will become a lot more about some 1191 00:59:42,264 --> 00:59:47,246 generalists who can do the work to get to a decision, to get to 1192 00:59:47,246 --> 00:59:51,031 a point where, I mean, in large organizations, uh, it's not just 1193 00:59:51,031 --> 00:59:54,230 that engineering is complex, it's also about getting people 1194 00:59:54,230 --> 00:59:57,571 aligned on something. So I think over there, I think 1195 00:59:57,571 --> 01:00:00,356 the hundreds of tools that require, and like generalists 1196 01:00:00,356 --> 01:00:04,160 can go in and absolutely should. But I think it'll be a 1197 01:00:04,160 --> 01:00:06,965 combination of some sort of like a ChatGPT for research, for 1198 01:00:06,965 --> 01:00:08,655 example. There are many tools out there 1199 01:00:08,655 --> 01:00:11,980 obviously that can do that. Something like a Miro, Figma, 1200 01:00:11,980 --> 01:00:16,402 Lovable, Bolt to do the initial sort of converting ideas into 1201 01:00:16,402 --> 01:00:19,977 some sort of visuals because people tend to, you know, when 1202 01:00:19,977 --> 01:00:23,496 you talk, like six people talk about a homepage, you're all 1203 01:00:23,496 --> 01:00:25,825 imagining different things. But a design essentially gets 1204 01:00:25,825 --> 01:00:29,000 everyone to be looking at the same thing and align and make 1205 01:00:29,000 --> 01:00:32,930 decisions. Beyond that, I think with a 1206 01:00:32,930 --> 01:00:37,556 design tool that can help bring design and front-end together 1207 01:00:37,556 --> 01:00:41,512 makes a ton of sense. And I think obviously like a 1208 01:00:41,512 --> 01:00:43,144 Cursor, Windsurf, Copilot for example, right? 1209 01:00:43,880 --> 01:00:48,155 May not necessarily be your solution to everything. 1210 01:00:48,575 --> 01:00:52,770 But things like code reviews, I think writing the logic with 1211 01:00:52,770 --> 01:00:55,306 auto-suggest for certain things that you've designed already, 1212 01:00:55,306 --> 01:00:58,560 that's where I think the Cursors and the Windsurfs of the world 1213 01:00:58,560 --> 01:01:02,261 obviously makes a lot of sense. I'm not very familiar with the 1214 01:01:02,261 --> 01:01:04,391 back-end side of things. I obviously am a front-end guy. 1215 01:01:04,986 --> 01:01:07,461 Essentially what I'm trying to say is that there will be 1216 01:01:07,461 --> 01:01:11,310 certain tools that will transform how you've done things 1217 01:01:11,310 --> 01:01:14,684 in the past. I don't think there will be one 1218 01:01:14,684 --> 01:01:16,754 tool to rule them all. It'll be a combination depending 1219 01:01:16,754 --> 01:01:19,259 on your use case. So in the enterprise side, if 1220 01:01:19,259 --> 01:01:21,965 you are a part of an innovation team, and you're just producing 1221 01:01:21,965 --> 01:01:24,539 things very quickly, use a Lovable or Bolt, for example. 1222 01:01:24,719 --> 01:01:26,713 But if you're building a production grade app, that may 1223 01:01:26,713 --> 01:01:29,329 not be your answer. It could be a combination of a, 1224 01:01:29,329 --> 01:01:32,782 like a Locofy plus Cursor. So the same organization might 1225 01:01:32,782 --> 01:01:36,898 end up having a stack of tools rather than just invest into one 1226 01:01:36,898 --> 01:01:39,096 and say, it'll solve all your problems. 1227 01:01:39,516 --> 01:01:42,950 And I think eventually, a couple of winners will emerge and they 1228 01:01:42,950 --> 01:01:47,250 will start either merging, acquiring, or getting into these 1229 01:01:47,250 --> 01:01:50,421 spaces. But I think to think that one 1230 01:01:50,421 --> 01:01:54,211 platform to rule it all or just one tool today that works today 1231 01:01:54,211 --> 01:01:57,627 will be the tool you use in 10 years from now, no one knows, 1232 01:01:57,627 --> 01:02:00,159 right? You try Cursor today and then 1233 01:02:00,159 --> 01:02:02,405 Windsurf gets better tomorrow. And then one LLM doesn't work 1234 01:02:02,405 --> 01:02:05,991 today, and then all of a sudden Claude comes out and works well, 1235 01:02:05,991 --> 01:02:10,032 and people say ChatGPT, Open, you know, the OpenAI models are 1236 01:02:10,032 --> 01:02:12,501 not working well. But then we try the actual 1237 01:02:12,501 --> 01:02:15,089 models and it works. No one knows what the real 1238 01:02:15,089 --> 01:02:17,999 answers are. But I think the key is 1239 01:02:17,999 --> 01:02:20,848 experimentation and the key is also giving the tools you've 1240 01:02:20,848 --> 01:02:23,873 chosen a little bit more time, because it always starts with 1241 01:02:23,873 --> 01:02:27,225 initial excitement, initially looks really good, then a 1242 01:02:27,225 --> 01:02:29,797 learning curve. And that's when you realize, oh, 1243 01:02:29,797 --> 01:02:33,148 it's not what I thought it would be because I have to do a 1244 01:02:33,148 --> 01:02:35,098 certain amount of things and there's a learning curve over 1245 01:02:35,098 --> 01:02:37,744 there. So I would just say like I 1246 01:02:37,744 --> 01:02:40,557 advise people against every month trying five, six new 1247 01:02:40,557 --> 01:02:42,470 tools. But yeah, do your research, find 1248 01:02:42,470 --> 01:02:43,942 a couple of tools that work for you. 1249 01:02:44,332 --> 01:02:48,045 Define the grunt work that either your team doesn't like or 1250 01:02:48,045 --> 01:02:49,302 you don't have the right people for. 1251 01:02:49,862 --> 01:02:52,226 Start with that and then eventually start like kind of 1252 01:02:52,226 --> 01:02:53,897 growing. Yeah. 1253 01:02:53,897 --> 01:02:55,977 From my experience as well, like what you mentioned, right? 1254 01:02:56,018 --> 01:02:59,242 So with so many different tools, different models, different 1255 01:02:59,242 --> 01:03:02,421 whatever, right? So I think the key thing is 1256 01:03:02,421 --> 01:03:05,775 obviously experimenting and know when to switch, right? 1257 01:03:05,835 --> 01:03:08,499 Like sometimes like you can bang over your head with the same 1258 01:03:08,499 --> 01:03:10,725 tools, but actually it doesn't actually solve the problems. 1259 01:03:10,905 --> 01:03:12,975 Now you have a plethora of tools that you can choose. 1260 01:03:13,305 --> 01:03:16,648 And I find that as long as you still rely on these so-called 1261 01:03:16,648 --> 01:03:18,930 popular foundational models, actually any tools you can use, 1262 01:03:18,930 --> 01:03:22,065 you can just switch, right? There's no moat per se, right? 1263 01:03:22,185 --> 01:03:23,622 Maybe... Especially with large language 1264 01:03:23,622 --> 01:03:25,044 models, that's exactly what it is. 1265 01:03:25,494 --> 01:03:30,093 I think, uh, we've seen it every two months when I meet the team. 1266 01:03:30,963 --> 01:03:34,173 Oh yeah, we ditched that one. Sonnet is doing better now. 1267 01:03:34,653 --> 01:03:36,603 Like what happened? No, we are not using that. 1268 01:03:36,603 --> 01:03:38,433 Llama is doing better now. So it keeps on changing. 1269 01:03:38,463 --> 01:03:41,696 It's hard to predict, but what's constant is that they will play 1270 01:03:41,696 --> 01:03:44,279 a big role. Which LLM, it would be a 1271 01:03:44,279 --> 01:03:45,573 combination of two or three maybe. 1272 01:03:45,573 --> 01:03:47,662 I don't know. People will start realizing that 1273 01:03:47,662 --> 01:03:49,627 when it comes to code, maybe Claude is better. 1274 01:03:49,627 --> 01:03:53,081 When it comes to, I don't know, research and product manager 1275 01:03:53,081 --> 01:03:56,432 work, maybe ChatGPT and OpenAI models are better. 1276 01:03:56,912 --> 01:03:59,072 And I think eventually people will just like stop 1277 01:03:59,072 --> 01:04:02,384 experimenting too much and just trust that even if at a certain 1278 01:04:02,384 --> 01:04:05,917 one of your chosen models is lagging behind, three or four 1279 01:04:05,917 --> 01:04:07,862 months later, it'll catch up eventually. 1280 01:04:08,492 --> 01:04:12,317 So yeah, but hard to say, man. Even for founders like me 1281 01:04:12,317 --> 01:04:16,096 building an AI developer tool, it's very hard to say where this 1282 01:04:16,096 --> 01:04:18,681 is going to be, because there's so much hype right now. 1283 01:04:19,281 --> 01:04:24,139 So much hype, so much of initial trying and POCs and below in the 1284 01:04:24,139 --> 01:04:26,271 next, I would say 12 to 18 months hopefully. 1285 01:04:27,051 --> 01:04:30,651 But I do think that we are at the stage where in the next 12 1286 01:04:30,651 --> 01:04:32,391 to 18 months, a lot of startups are gonna die. 1287 01:04:33,274 --> 01:04:35,826 And then the winners will emerge and then people can focus on 1288 01:04:35,826 --> 01:04:38,757 those two or three. So thanks for sharing all that, 1289 01:04:38,757 --> 01:04:40,142 right? So I think one particular thing 1290 01:04:40,142 --> 01:04:42,164 that I wanna pick from this conversation as well, because 1291 01:04:42,164 --> 01:04:43,652 you are kind of like unique, right? 1292 01:04:43,652 --> 01:04:48,092 You are building an application developer tools, right, from 1293 01:04:48,092 --> 01:04:50,042 Southeast Asia for global markets. 1294 01:04:50,042 --> 01:04:51,962 And I know this is one of your passion, right? 1295 01:04:51,962 --> 01:04:55,121 Building, I dunno, global solutions from Southeast Asia 1296 01:04:55,121 --> 01:04:57,339 region. So tell us about, you know, 1297 01:04:57,339 --> 01:04:59,951 first of all, right, the promise of Southeast Asia region. 1298 01:04:59,951 --> 01:05:02,809 Like do you see the skillset here is capable enough to 1299 01:05:02,809 --> 01:05:05,759 actually solve global problems? And what do you see the trend so 1300 01:05:05,759 --> 01:05:06,478 far? Yeah. 1301 01:05:06,818 --> 01:05:10,068 I mean, I'm super passionate, because I've been here 20 years. 1302 01:05:10,068 --> 01:05:13,422 And yeah, I kind of decided I like to stay in Singapore and 1303 01:05:13,422 --> 01:05:16,501 kind of make that chance that the Singapore maybe government 1304 01:05:16,501 --> 01:05:17,851 took on me through a scholarship. 1305 01:05:17,911 --> 01:05:20,702 Make it worth it. There've been enough times in my 1306 01:05:20,702 --> 01:05:23,233 career where I've heard from people that Southeast Asia does 1307 01:05:23,233 --> 01:05:25,791 not have the talent. And I understand where it comes 1308 01:05:25,791 --> 01:05:27,771 from. But how does talent actually 1309 01:05:27,771 --> 01:05:30,579 started? I mean, India didn't have the 1310 01:05:30,579 --> 01:05:34,209 engineering talent to build SaaS until Zoho built it against all 1311 01:05:34,209 --> 01:05:36,561 the odds, and then now it's becoming a SaaS factory. 1312 01:05:37,101 --> 01:05:40,341 And I always believed that Southeast Asia needs the same. 1313 01:05:40,551 --> 01:05:45,453 We don't have a lack of talent. We do have a lack of taking big 1314 01:05:45,453 --> 01:05:49,189 bets, I would say. And then the more I start 1315 01:05:49,189 --> 01:05:51,997 understanding how the whole value chain works, it really 1316 01:05:51,997 --> 01:05:54,487 comes down to risk capital, basically. 1317 01:05:55,417 --> 01:06:00,007 The appetite for high risk moonshot projects, investing 1318 01:06:00,007 --> 01:06:03,232 heavy amounts of money into it. In the West, we just don't have 1319 01:06:03,232 --> 01:06:06,745 it over here. So I think over here, as much as 1320 01:06:06,745 --> 01:06:09,822 I believe in Southeast Asia being, uh, you know, competitive 1321 01:06:09,822 --> 01:06:13,507 and we have talent from around the world and stable government 1322 01:06:13,507 --> 01:06:16,049 and like, founder-friendly, I would say, terms through capital 1323 01:06:16,049 --> 01:06:19,055 gain tax and everything, just easy to do business. 1324 01:06:20,023 --> 01:06:24,223 It's still, I would say, not where the hype machine is. 1325 01:06:25,973 --> 01:06:29,784 Um, but we do plan to or at least hope to change that. 1326 01:06:29,784 --> 01:06:32,448 And there are companies like Supabase, LottieFiles that have 1327 01:06:32,448 --> 01:06:35,734 origins over here. And there are ways of kind of 1328 01:06:35,734 --> 01:06:38,894 being across two markets. So have your team in the US and 1329 01:06:38,894 --> 01:06:41,194 Singapore, and get access to the best of both worlds. 1330 01:06:41,967 --> 01:06:44,531 But yeah, I mean, I think Southeast Asia has the 1331 01:06:44,531 --> 01:06:47,327 ingredients. I think in the last two to three 1332 01:06:47,327 --> 01:06:50,270 years, obviously the risk capital has dropped 1333 01:06:50,270 --> 01:06:52,820 tremendously. That doesn't help any founders 1334 01:06:52,820 --> 01:06:55,555 like us who want to change that from this part of the world. 1335 01:06:56,285 --> 01:06:59,082 But yeah. I would say I'm quite positive 1336 01:06:59,082 --> 01:07:02,144 about the future, just how quickly it'll happen and the 1337 01:07:02,144 --> 01:07:04,884 role we can play into it remains to be seen. 1338 01:07:05,304 --> 01:07:09,149 I do think we need, uh, when it comes to investments, which any 1339 01:07:09,149 --> 01:07:12,314 startup needs, I still think for building a developer tool or an 1340 01:07:12,314 --> 01:07:14,004 AI company, you just have to go to the West. 1341 01:07:14,364 --> 01:07:19,803 And that's been a realization lately that there's no point 1342 01:07:19,803 --> 01:07:23,133 debating and no point, like talking about what is the 1343 01:07:23,133 --> 01:07:25,611 idealistic case? The reality is investors in this 1344 01:07:25,611 --> 01:07:29,542 part of the world don't have the same maybe risk appetite or the 1345 01:07:29,542 --> 01:07:32,403 background of the history. Otherwise, Figma would be built 1346 01:07:32,403 --> 01:07:34,168 from here. OpenAI would be built from here. 1347 01:07:34,468 --> 01:07:36,598 I don't think it's necessarily because of the talent. 1348 01:07:36,958 --> 01:07:39,734 Because we've had Singapore talent go to Silicon Valley and 1349 01:07:39,734 --> 01:07:42,998 do well. And Silicon Valley also attracts 1350 01:07:42,998 --> 01:07:44,518 talent. I mean, some of the best 1351 01:07:44,518 --> 01:07:46,900 companies are being built by foreigners, but they go over 1352 01:07:46,900 --> 01:07:48,178 there because they know there's an opportunity. 1353 01:07:48,178 --> 01:07:50,278 Not just that you get smarter by just being over there. 1354 01:07:51,018 --> 01:07:54,451 We have enough talent over here. We just need the right sort of 1355 01:07:54,451 --> 01:07:57,720 balance in the ecosystem. And I think it's a work in 1356 01:07:57,720 --> 01:07:58,341 progress. Yeah. 1357 01:07:58,881 --> 01:08:00,831 I think, not to mention as well, when you mentioned about risk 1358 01:08:00,911 --> 01:08:02,831 kind of like averse thing, right? 1359 01:08:02,831 --> 01:08:06,791 So individuals themselves also probably take less bet. 1360 01:08:07,091 --> 01:08:09,611 Uh, you know, that we want like a more stable jobs, you know, 1361 01:08:09,641 --> 01:08:12,288 proper career and all that. So I think that mindset from 1362 01:08:12,288 --> 01:08:14,277 individuals as well I think might need to change. 1363 01:08:14,597 --> 01:08:17,532 So you mentioned about building dev tools kind of like difficult 1364 01:08:17,532 --> 01:08:19,508 here. I know that we have many 1365 01:08:19,508 --> 01:08:22,514 startups here that are turning into big as well, but yeah, 1366 01:08:22,514 --> 01:08:24,689 they're solving more like, I dunno, like society problem, 1367 01:08:24,689 --> 01:08:26,069 maybe business problems and all that. 1368 01:08:26,249 --> 01:08:29,399 Why particularly building dev tools is not popular here? 1369 01:08:29,399 --> 01:08:32,792 Is it like, I dunno, some trends or some mindset because we have 1370 01:08:32,792 --> 01:08:35,694 talents, we have developers here building solutions. 1371 01:08:35,694 --> 01:08:39,062 I think Southeast Asia startup scene is now kind of going into 1372 01:08:39,062 --> 01:08:42,747 the second wave, right? The first wave is generally 1373 01:08:42,747 --> 01:08:44,948 communication, e-commerce, logistics, right? 1374 01:08:45,127 --> 01:08:47,899 FinTech. It's just been a decade 1375 01:08:47,899 --> 01:08:51,964 actually, not more than that. And I think now we have enough 1376 01:08:51,964 --> 01:08:54,050 experienced engineering leaders and startup builders and 1377 01:08:54,050 --> 01:08:57,144 operators who go and say, yeah, you know what, it's, it was fun 1378 01:08:57,144 --> 01:08:59,094 building Lazada, fun building Grab. 1379 01:08:59,573 --> 01:09:02,947 And we learned a lot. But the real geeks will say, I 1380 01:09:02,947 --> 01:09:05,831 wanna go deeper. What seemed impossible almost 1381 01:09:05,831 --> 01:09:09,040 five years ago, building an app that, you know, you can show 1382 01:09:09,040 --> 01:09:11,917 location on and like connect Stripe to it, today doesn't 1383 01:09:11,917 --> 01:09:15,193 excite me anymore. I wanna go deeper into either 1384 01:09:15,193 --> 01:09:16,645 front-end, back-end, or anything. 1385 01:09:16,645 --> 01:09:18,055 Anything that people are passionate about. 1386 01:09:18,595 --> 01:09:21,337 I think 10 years ago, we just didn't have enough examples 1387 01:09:21,337 --> 01:09:26,100 around us, enough investors, and enough, I think, experience in 1388 01:09:26,100 --> 01:09:29,319 the operator side to go that deep into deep tech. 1389 01:09:30,069 --> 01:09:32,555 I think that's changing now, which is why we have, I think, 1390 01:09:32,555 --> 01:09:35,606 in the last five years I've seen maybe seven or eight developer 1391 01:09:35,606 --> 01:09:37,974 tool companies come from this part of the world and eventually 1392 01:09:37,974 --> 01:09:39,054 decided to move to Silicon Valley. 1393 01:09:39,654 --> 01:09:41,928 Because people, you will have a reminder on a daily basis, why 1394 01:09:41,928 --> 01:09:43,919 are you over here? Why are you building from 1395 01:09:43,919 --> 01:09:45,912 Singapore? Just go where all the developers 1396 01:09:45,912 --> 01:09:48,033 are. Because one thing true about 1397 01:09:48,033 --> 01:09:51,886 Silicon Valley is that the amount of early adopters and 1398 01:09:51,886 --> 01:09:55,976 people who are willing to try something that's not yet 1399 01:09:55,976 --> 01:10:00,266 perfect, the density in Silicon Valley is just insanely high on 1400 01:10:00,266 --> 01:10:03,530 the addition of obviously the risk capital that I talked about 1401 01:10:03,530 --> 01:10:06,247 earlier. And I think that's why in 1402 01:10:06,247 --> 01:10:10,088 Southeast Asia, it's difficult because I think the early 1403 01:10:10,088 --> 01:10:12,962 adopters may exist, but it's a very, very small number. 1404 01:10:13,022 --> 01:10:15,792 Like Indonesia, Southeast Asia, they generally tend to follow 1405 01:10:15,792 --> 01:10:18,905 what's coming from the West. I don't think it's something 1406 01:10:18,905 --> 01:10:20,165 that cannot change and will not change. 1407 01:10:20,165 --> 01:10:22,799 It will. But we just need maybe three or 1408 01:10:22,799 --> 01:10:26,315 four strong examples built from this part of the world to both 1409 01:10:26,315 --> 01:10:29,372 attract more founders to solve these kind of problems, global 1410 01:10:29,372 --> 01:10:31,115 problems, and not just Southeast Asia problems. 1411 01:10:31,415 --> 01:10:34,252 More deep, deep tech problems than just on the application 1412 01:10:34,252 --> 01:10:36,631 there. And also will attract the right 1413 01:10:36,631 --> 01:10:38,737 investors after we have some exits, I would say. 1414 01:10:39,368 --> 01:10:41,690 Yeah. So I think one example is 1415 01:10:41,690 --> 01:10:43,778 definitely Locofy.ai, right? Yeah. 1416 01:10:44,228 --> 01:10:47,977 And, and I know that you are interestingly a remote company, 1417 01:10:47,977 --> 01:10:50,102 right? Fully remote company and you 1418 01:10:50,102 --> 01:10:51,887 have developers across different countries. 1419 01:10:52,067 --> 01:10:55,187 So tell us, I think one problem for Southeast Asia is like how 1420 01:10:55,187 --> 01:10:57,113 fragmented, you know, the region are, right? 1421 01:10:57,203 --> 01:10:59,387 So many different countries, different culture, different 1422 01:10:59,387 --> 01:11:01,803 language, you know, different norms, whatever that is, right? 1423 01:11:01,803 --> 01:11:05,667 Different even like lifestyle and, you know, rate of living, 1424 01:11:05,667 --> 01:11:08,490 right? So tell us like what's, uh, the 1425 01:11:08,490 --> 01:11:11,046 challenge here when you built fully remote company from 1426 01:11:11,046 --> 01:11:12,546 Southeast Asia? Yeah. 1427 01:11:13,190 --> 01:11:16,518 So I think nature of the problem was such that we thought we 1428 01:11:16,518 --> 01:11:18,600 don't necessarily have to be in the office, right? 1429 01:11:18,600 --> 01:11:22,631 Like because of COVID, I guess designers and engineers and 1430 01:11:22,631 --> 01:11:25,065 product managers became very comfortable working with each 1431 01:11:25,065 --> 01:11:26,704 other. As long as you have the right 1432 01:11:26,704 --> 01:11:29,300 people you can trust. The bigger challenge was still 1433 01:11:29,300 --> 01:11:31,820 that the early adopters of our product are still very much like 1434 01:11:31,820 --> 01:11:33,555 around the world and not just in Southeast Asia. 1435 01:11:34,365 --> 01:11:38,104 So today, we are building from Singapore, but less than 5% of 1436 01:11:38,104 --> 01:11:39,695 our revenue is from this part of the world. 1437 01:11:40,025 --> 01:11:43,223 We were solving a global problem from day one, and we wanted to 1438 01:11:43,223 --> 01:11:45,079 build from this part of the world because there's a lot of 1439 01:11:45,079 --> 01:11:46,719 benefits. Like I've said, easy to start a 1440 01:11:46,719 --> 01:11:48,203 company and we had background over here. 1441 01:11:48,593 --> 01:11:52,121 Our first 10, 15 people we knew we wanted to work with were over 1442 01:11:52,121 --> 01:11:55,017 here. But yeah, as the company matures 1443 01:11:55,017 --> 01:11:59,057 a little bit and we start going into GTM, start going into 1444 01:11:59,057 --> 01:12:03,202 fundraising, the challenge can be not just being remote, but 1445 01:12:03,202 --> 01:12:05,697 being in this part of the world. And secondly, now we have people 1446 01:12:05,697 --> 01:12:08,697 from all around the world. And if majority of the customers 1447 01:12:08,697 --> 01:12:11,668 are going to be in the west, time zones can become a bit of a 1448 01:12:11,668 --> 01:12:13,569 problem. Culturally, I think it, it's a 1449 01:12:13,569 --> 01:12:16,830 little bit difficult to manage maybe more than 30 people remote 1450 01:12:16,830 --> 01:12:20,323 than with 10 people. 10 people remote, we hung out every day. 1451 01:12:20,673 --> 01:12:23,502 30 people, sometimes we see people and I'm like other than 1452 01:12:23,502 --> 01:12:25,893 making them an offer, I don't remember talking to them. 1453 01:12:26,643 --> 01:12:29,927 So the challenges will always be that, look, problems get solved 1454 01:12:29,927 --> 01:12:33,915 much faster when you sit together, have a whiteboard, or 1455 01:12:33,915 --> 01:12:37,679 sometimes you, online will take you five days, six days to get 1456 01:12:37,679 --> 01:12:40,969 people to really get going versus in person is just better 1457 01:12:40,969 --> 01:12:43,074 that way. Selling to customers is 1458 01:12:43,074 --> 01:12:46,595 generally better in person. But what works remotely very 1459 01:12:46,595 --> 01:12:50,650 well is that our engineers are not spending any time in 1460 01:12:50,650 --> 01:12:53,048 traffic. They're not living in expensive 1461 01:12:53,048 --> 01:12:55,679 CBD areas to be closer to office. 1462 01:12:55,969 --> 01:12:59,507 Not just in Singapore, but in the six, seven markets we have 1463 01:12:59,507 --> 01:13:03,197 our team members in and they get the time to do more important 1464 01:13:03,197 --> 01:13:06,453 things in life, right? Like there's no reason why you 1465 01:13:06,453 --> 01:13:10,851 have to work 12 hours a day and, uh, miss out, like your kid goes 1466 01:13:10,851 --> 01:13:13,735 to school and you don't get to see them at night, for example, 1467 01:13:13,735 --> 01:13:15,906 right? And sometimes you just wanna 1468 01:13:15,906 --> 01:13:18,359 take a nap, a 15-minute nap and you get a productivity, you 1469 01:13:18,359 --> 01:13:19,811 know, like boost all of a sudden. 1470 01:13:20,251 --> 01:13:23,507 Those kind of things, I think is very, very hard to get in an 1471 01:13:23,507 --> 01:13:26,791 office environment. And we've seen that people in 1472 01:13:26,791 --> 01:13:29,671 our team appreciate that they can stay home. 1473 01:13:30,271 --> 01:13:32,581 They're working hard as hell, but they don't miss out the 1474 01:13:32,581 --> 01:13:34,781 important things in life, whether they're parents, whether 1475 01:13:34,781 --> 01:13:37,951 it's taking care of their family, even walking their dog, 1476 01:13:37,951 --> 01:13:40,231 for example, taking care of their health sometimes. 1477 01:13:40,681 --> 01:13:44,521 I think, yeah, we are still very much going to try and keep it 1478 01:13:44,521 --> 01:13:48,497 remote as long as possible. But I was say, as I was saying 1479 01:13:48,497 --> 01:13:51,095 to you earlier, I've started now feeling and some other people in 1480 01:13:51,095 --> 01:13:53,275 the team have started saying as well that sometimes we just wish 1481 01:13:53,275 --> 01:13:56,410 we were sitting in an office in a, to solve this tougher 1482 01:13:56,410 --> 01:13:59,457 problem. Sit in one room for a full day, 1483 01:13:59,457 --> 01:14:02,029 we would've found a solution whereas online can be a bit 1484 01:14:02,029 --> 01:14:04,651 challenging that way. And also you have good days and 1485 01:14:04,651 --> 01:14:06,184 bad days. And when you have bad days, it's 1486 01:14:06,184 --> 01:14:09,146 easy to have a colleague who you can go down for a coffee with, 1487 01:14:09,146 --> 01:14:11,259 went out and move on versus being remote. 1488 01:14:11,259 --> 01:14:14,408 I think sometimes you, no one even knows you're going through 1489 01:14:14,408 --> 01:14:16,642 something. So I think, yeah, it has its 1490 01:14:16,642 --> 01:14:19,344 pros and cons, but it allows us to hire in every part of the 1491 01:14:19,344 --> 01:14:22,038 world, which is a huge pro. And engineers and designers like 1492 01:14:22,038 --> 01:14:25,349 the fact that they can be just in front of a laptop and not 1493 01:14:25,349 --> 01:14:28,089 like spend two hours in the morning getting ready, getting 1494 01:14:28,089 --> 01:14:31,289 to work, and then back. They would rather spend those 1495 01:14:31,289 --> 01:14:34,419 three, four hours a day doing other things outside of work and 1496 01:14:34,419 --> 01:14:36,204 still get the productivity boost they need. 1497 01:14:36,504 --> 01:14:37,209 Yeah. Right. 1498 01:14:37,868 --> 01:14:40,595 You, you mentioned that, um, interestingly, your customers 1499 01:14:40,595 --> 01:14:43,533 starting to get, you know, like you get it from, not from this 1500 01:14:43,533 --> 01:14:45,451 region, right? So tell us a little bit about 1501 01:14:45,451 --> 01:14:47,965 the traction of Locofy so far. You, you know, you have been 1502 01:14:47,965 --> 01:14:50,498 around four and a half years. What's the stage like? 1503 01:14:50,588 --> 01:14:53,598 You know, how much customers base, how much revenue you...? 1504 01:14:54,092 --> 01:14:56,072 Yeah. So we were in free beta for 1505 01:14:56,072 --> 01:14:59,453 three and a half years. And that helped us, uh, and I 1506 01:14:59,453 --> 01:15:02,930 mean, we didn't really, we did really well on Product Hunt and 1507 01:15:02,930 --> 01:15:05,778 Indie Hackers, Twitter. Like YouTube, people just kept 1508 01:15:05,778 --> 01:15:07,808 making videos about us when they found out about us. 1509 01:15:07,808 --> 01:15:12,102 TikTok went viral at one point. So we were lucky that we built a 1510 01:15:12,102 --> 01:15:15,380 strong beta community rather. And when we monetized eventually 1511 01:15:15,380 --> 01:15:18,464 last year, we had signs that enterprises and full stack 1512 01:15:18,464 --> 01:15:21,269 engineers were actually where the strong product market fit 1513 01:15:21,269 --> 01:15:23,750 like. So we started charging our 1514 01:15:23,750 --> 01:15:27,125 customers exactly a year ago. The first quarter, we started 1515 01:15:27,125 --> 01:15:29,704 realizing the pricing was a little bit off here and there, 1516 01:15:29,704 --> 01:15:32,744 and workshops were important, and we need to have a sales 1517 01:15:32,744 --> 01:15:34,220 person in the sales deck and everything. 1518 01:15:34,220 --> 01:15:36,932 It takes time. It's only earlier this year we 1519 01:15:36,932 --> 01:15:38,870 realized that we've gotten the pricing right. 1520 01:15:39,620 --> 01:15:42,380 Like not a seat-based model, but a usage-based model. 1521 01:15:43,300 --> 01:15:47,481 And also pricing that doesn't push enterprises towards self 1522 01:15:47,481 --> 01:15:51,766 service where they, we know they will struggle because there's no 1523 01:15:51,766 --> 01:15:54,475 training available. But yeah, after we figured it 1524 01:15:54,475 --> 01:15:57,661 out, it took us, uh, nine months to get to our first million 1525 01:15:57,661 --> 01:16:00,445 revenue, and now we are at a one year stage. 1526 01:16:00,445 --> 01:16:04,804 We are, just crossed 1.5. Our customers are mainly 1527 01:16:04,804 --> 01:16:08,774 actually, we've seen a strong product-market fit with system 1528 01:16:08,774 --> 01:16:12,239 integrators and agencies. Maybe because of the skill gap 1529 01:16:12,239 --> 01:16:14,847 over there, maybe because of their own business model itself, 1530 01:16:14,847 --> 01:16:17,751 it makes sense for them to be able to build faster and 1531 01:16:17,751 --> 01:16:22,085 cheaper. And when it comes to the kind of 1532 01:16:22,085 --> 01:16:25,318 enterprises other than system integrators and agencies, think 1533 01:16:25,318 --> 01:16:28,687 of like companies like Toyota, Bupa, for example. 1534 01:16:28,687 --> 01:16:31,869 Insurance companies, banks, security bank, DBS, all these 1535 01:16:31,869 --> 01:16:35,837 kind of large enterprises that have ambitions to be a tech 1536 01:16:35,837 --> 01:16:38,107 company. But they don't necessarily 1537 01:16:38,107 --> 01:16:39,592 attract the best software engineers. 1538 01:16:40,162 --> 01:16:42,754 So Toyota attracts really good mechanical engineers and 1539 01:16:42,754 --> 01:16:45,857 automobile engineers. But would the best Stanford, you 1540 01:16:45,857 --> 01:16:49,816 know, software engineer, prefer to go to a Facebook or to 1541 01:16:49,816 --> 01:16:52,692 Toyota, right? Doesn't mean that Toyota and the 1542 01:16:52,692 --> 01:16:55,223 Bupa of the world don't have ambitions too and they don't 1543 01:16:55,223 --> 01:16:58,699 have the budgets too. But that's where we've seen 1544 01:16:58,699 --> 01:17:01,678 strong product-market fit, where they haven't been able to 1545 01:17:01,678 --> 01:17:04,188 attract, you know, the 10 out of 10 engineers maybe. 1546 01:17:05,058 --> 01:17:07,268 But they're really good at what they do and they have ambitions 1547 01:17:07,268 --> 01:17:10,638 to be a tech company. So that's where we've seen a 1548 01:17:10,638 --> 01:17:13,340 strong product-market fit. And when it comes to startups, 1549 01:17:13,340 --> 01:17:18,083 we are working on a slightly simpler flow today to kind of 1550 01:17:18,083 --> 01:17:21,556 circumvent the design optimization issues I mentioned 1551 01:17:21,556 --> 01:17:23,788 to you earlier. And also to fit into the Cursors 1552 01:17:23,788 --> 01:17:25,468 of the world a little bit better. 1553 01:17:25,528 --> 01:17:28,617 More UI-less, for example, go from 15 steps to just one step, 1554 01:17:28,617 --> 01:17:31,932 where you can take your Figma design, run Locofy design models 1555 01:17:31,932 --> 01:17:34,826 with just a click, and then immediately continue into 1556 01:17:34,826 --> 01:17:37,458 Cursor, for example, rather than adding more steps. 1557 01:17:38,554 --> 01:17:41,491 Congratulations for such a rapid, you know, traction, 1558 01:17:41,491 --> 01:17:45,048 revenue and all that. So for sure that I hope these 1559 01:17:45,048 --> 01:17:48,840 tools will get more traction because I believe it solves a 1560 01:17:48,840 --> 01:17:51,464 real problem, right? Especially, for builders, right? 1561 01:17:51,464 --> 01:17:54,440 For people who would like to build products, especially the, 1562 01:17:54,440 --> 01:17:56,700 you know, front-end web products, right? 1563 01:17:57,130 --> 01:17:59,928 So we have covered a lot of things, is there anything in 1564 01:17:59,928 --> 01:18:02,615 particular that you think we missed or you want to also 1565 01:18:02,615 --> 01:18:05,218 convey in this conversation? Uh, not particularly. 1566 01:18:05,278 --> 01:18:07,738 I would say. No, man. 1567 01:18:07,738 --> 01:18:11,258 I think we've covered it, yeah. So, as we wrap up, uh, our 1568 01:18:11,258 --> 01:18:13,798 conversation today, uh, Honey, I have one last question, uh, 1569 01:18:13,798 --> 01:18:15,533 which is like a tradition in my podcast. 1570 01:18:15,533 --> 01:18:17,873 I call this the three technical leadership wisdom. 1571 01:18:17,873 --> 01:18:20,186 So think of it just like advice you want to give to the 1572 01:18:20,186 --> 01:18:22,658 listeners. Maybe if you can share a version 1573 01:18:22,658 --> 01:18:27,395 from you today, that be great. I would say that I will share 1574 01:18:27,395 --> 01:18:29,947 what I follow myself. Otherwise, what's the point? 1575 01:18:30,268 --> 01:18:32,991 I think number one is don't get too comfortable ever, especially 1576 01:18:32,991 --> 01:18:34,528 if you're working in the tech industry. 1577 01:18:35,049 --> 01:18:37,613 I think engineers, for a long time, they thought that they 1578 01:18:37,613 --> 01:18:39,324 were invincible. And I think the last three, four 1579 01:18:39,324 --> 01:18:43,240 years have been, I would say, humbling for a lot of us. 1580 01:18:43,240 --> 01:18:45,220 Product managers, engineers, designers. 1581 01:18:45,490 --> 01:18:48,550 So always stay nimble. Always keep learning. 1582 01:18:49,120 --> 01:18:50,860 Like learning shouldn't stop after university. 1583 01:18:51,580 --> 01:18:53,718 And I think you should, no matter how long you've been in 1584 01:18:53,718 --> 01:18:56,721 the company, 10 years, 15 years today, your role is never safe. 1585 01:18:56,781 --> 01:18:59,049 That's the reality. And if you keep working with 1586 01:18:59,049 --> 01:19:01,557 that mentality that I will never get comfortable just because 1587 01:19:01,557 --> 01:19:04,901 I've earned the right to be comfortable, I think you'll be 1588 01:19:04,901 --> 01:19:06,525 fine. Waves will come, innovations 1589 01:19:06,525 --> 01:19:09,656 will come and go, hype cycles will come and go, but if you 1590 01:19:09,656 --> 01:19:12,470 are... yeah, don't box yourself to be just like I'm a PWA guy, 1591 01:19:12,470 --> 01:19:16,122 or I'm a mobile app builder. Keep learning what's coming, 1592 01:19:16,122 --> 01:19:20,504 what's new with technology. Don't fall in obviously for all 1593 01:19:20,504 --> 01:19:22,556 the hype, but keep reinventing yourself. 1594 01:19:22,586 --> 01:19:25,052 Like my co-founder, for example, went from being a gaming 1595 01:19:25,052 --> 01:19:28,058 engineer to a mobile app engineer, to then a PWA 1596 01:19:28,058 --> 01:19:30,746 engineer, and now an AI expert, pretty much. 1597 01:19:31,631 --> 01:19:34,710 And he's 40. Usually at this stage you'll be 1598 01:19:34,710 --> 01:19:38,196 like, I would be a line manager, I just manage some people and 1599 01:19:38,196 --> 01:19:41,608 work on some tough problems. But I'm the Ruby on Rails guy, 1600 01:19:41,608 --> 01:19:44,111 or I'm the back-end guy, or I'm the AWS expert. 1601 01:19:44,501 --> 01:19:48,158 So I think it's good to push yourself to become as close to 1602 01:19:48,158 --> 01:19:51,936 being an expert as possible, but also don't box yourself in just 1603 01:19:51,936 --> 01:19:56,160 this feeling of I've earned it. I've worked so many years on 1604 01:19:56,160 --> 01:19:58,895 something, I'm an expert and I'm safe now, and I can take it 1605 01:19:58,895 --> 01:20:00,462 easy. That's just one of the things. 1606 01:20:00,462 --> 01:20:03,492 So leadership and management are two different things. 1607 01:20:03,492 --> 01:20:06,302 I believe in leadership, but not becoming like a line manager. 1608 01:20:07,245 --> 01:20:10,261 I think 50% of the millions of people who've been fired in the 1609 01:20:10,261 --> 01:20:12,285 last three, four years are actually those managers. 1610 01:20:12,977 --> 01:20:16,238 So yeah, just always be an individual contributor and of 1611 01:20:16,238 --> 01:20:19,860 course, learn how to manage and learn how to lead teams and lead 1612 01:20:19,860 --> 01:20:22,638 tough problems. Yeah, put yourself in a position 1613 01:20:22,638 --> 01:20:26,244 where you cannot be boxed into just one thing and like that 1614 01:20:26,244 --> 01:20:28,230 whole function or that role goes away. 1615 01:20:28,260 --> 01:20:32,629 That would be number one. I would say, don't be like, oh, 1616 01:20:32,629 --> 01:20:36,209 you are a Chief Product Officer and now I can just go to next 1617 01:20:36,209 --> 01:20:37,437 company and be a Chief Product Officer. 1618 01:20:37,767 --> 01:20:41,811 I generally believe that we are living in a world where it's 1619 01:20:41,811 --> 01:20:43,882 very different from how our parents grew up. 1620 01:20:44,752 --> 01:20:47,777 Like in their generation, it was believed they will live until 1621 01:20:47,777 --> 01:20:50,541 70, maybe. And it was divided into three 1622 01:20:50,541 --> 01:20:53,157 phases, right? The first 25 years to learn. 1623 01:20:53,427 --> 01:20:57,177 Next 25 years to earn by becoming an expert at something. 1624 01:20:57,177 --> 01:21:00,478 And the next 25 years or 20 years to just retire based on 1625 01:21:00,478 --> 01:21:03,559 what you earn. Now as life expectancy has 1626 01:21:03,559 --> 01:21:06,563 obviously changed tremendously, and it will continue doing so, 1627 01:21:06,563 --> 01:21:10,891 like to think you live until 90 or 100 is not considered crazy. 1628 01:21:11,311 --> 01:21:15,301 And when that happens, whatever makes you an expert in your 1629 01:21:15,301 --> 01:21:19,317 thirties will not be enough for you to continue until you're 60, 1630 01:21:19,317 --> 01:21:22,573 70. So I think it's important for 1631 01:21:22,573 --> 01:21:25,827 people to start taking designated breaks, career breaks 1632 01:21:25,827 --> 01:21:29,499 to go and reinvent themselves, to always be hands-on to what I 1633 01:21:29,499 --> 01:21:32,939 just said. And to never feel like, but I'm 1634 01:21:32,939 --> 01:21:36,584 not an AI expert. When AI wave came, I didn't 1635 01:21:36,584 --> 01:21:40,185 study AI in university, but we got our hands dirty. 1636 01:21:40,185 --> 01:21:43,322 And it, you'll go through a period where you are 1637 01:21:43,322 --> 01:21:47,535 uncomfortable, you feel lost, you feel like I'm just not young 1638 01:21:47,535 --> 01:21:50,493 enough to learn. But you have to keep powering 1639 01:21:50,493 --> 01:21:52,697 through because very quickly you'll become that expert that 1640 01:21:52,697 --> 01:21:57,255 everyone wants you to be. And I think it applies to 1641 01:21:57,255 --> 01:22:00,047 leaders more importantly than others, because, yeah, you're 1642 01:22:00,047 --> 01:22:03,687 not just in a team or in an organization for the quality of 1643 01:22:03,687 --> 01:22:06,678 decision and the experience you bring in, but also for seeing it 1644 01:22:06,678 --> 01:22:08,790 through. And the execution piece should 1645 01:22:08,790 --> 01:22:12,468 never become like, hey, the last time I did something like this 1646 01:22:12,468 --> 01:22:14,952 was 10 years ago to stay relevant and to stay ahead of 1647 01:22:14,952 --> 01:22:16,032 the curve as well. Yeah. 1648 01:22:16,062 --> 01:22:18,686 Be a problem solver. Technologies will come and go 1649 01:22:18,686 --> 01:22:23,702 but yeah, if you're too far away from the tech, then at some 1650 01:22:23,702 --> 01:22:28,119 point of time organizations will prefer younger, hungrier, and 1651 01:22:28,119 --> 01:22:31,992 more adaptable graduates. So I think it's more that's what 1652 01:22:31,992 --> 01:22:34,341 mantra I follow. If that's the only advice I 1653 01:22:34,341 --> 01:22:35,895 would give, I'll be happy with that as well, yeah. 1654 01:22:36,641 --> 01:22:39,161 Yeah, I think that's a very great reminder, especially for 1655 01:22:39,161 --> 01:22:41,711 people who have been around for quite a number of years, right? 1656 01:22:41,711 --> 01:22:44,621 So especially line managers or maybe leaders, right? 1657 01:22:44,621 --> 01:22:48,041 So always gets your hand dirty, you know, being hands on. 1658 01:22:48,311 --> 01:22:50,675 Especially now. Actually with AI tools, I feel 1659 01:22:50,675 --> 01:22:53,761 as someone who have been around in the industry for quite some 1660 01:22:53,761 --> 01:22:55,381 time, it's actually a powerful leverage. 1661 01:22:55,531 --> 01:22:57,647 You can learn so much things quickly. 1662 01:22:58,127 --> 01:23:00,071 Yeah. And anything, anything you can 1663 01:23:00,071 --> 01:23:02,464 learn, right? So you can actually pick it up 1664 01:23:02,464 --> 01:23:05,749 and, you know, try it out and even write prototypes, code and 1665 01:23:05,749 --> 01:23:08,011 all that. Because in the past, maybe the 1666 01:23:08,011 --> 01:23:11,785 first hurdle is like the how. We don't have the muscle to 1667 01:23:11,785 --> 01:23:13,771 actually build something, you know, the know-how. 1668 01:23:14,071 --> 01:23:17,994 But now I think AI can help us in the first step to actually 1669 01:23:17,994 --> 01:23:19,684 build something and maybe iterate from there. 1670 01:23:19,684 --> 01:23:21,844 So I think that's a very great reminder. 1671 01:23:22,264 --> 01:23:25,366 So Honey, if people would love to connect with you, learn more 1672 01:23:25,366 --> 01:23:27,514 from you, learn more about Locofy, is there place where 1673 01:23:27,514 --> 01:23:29,858 they can find you online? I'm on LinkedIn all the time. 1674 01:23:31,682 --> 01:23:36,002 Other than that, I think, um, my email address should actually be 1675 01:23:36,002 --> 01:23:37,772 available on the, on our website as well. 1676 01:23:37,772 --> 01:23:41,453 But yeah, LinkedIn I would say would be the best place to find 1677 01:23:41,453 --> 01:23:44,046 us or me. And then, uh, you can always 1678 01:23:44,046 --> 01:23:47,846 book a call on our website where you book a demo or an 1679 01:23:47,846 --> 01:23:50,650 exploration session with us. And more often than not, I'll be 1680 01:23:50,650 --> 01:23:52,047 on that call, yeah. Right. 1681 01:23:52,497 --> 01:23:54,787 So again, thank you for sharing about Locofy today. 1682 01:23:54,787 --> 01:23:58,815 I'm sure many engineers or designers who are hearing this 1683 01:23:58,815 --> 01:24:01,801 conversation will be very excited, and I hope they try out 1684 01:24:01,801 --> 01:24:04,019 your tools. And yeah, all the best for the 1685 01:24:04,019 --> 01:24:05,291 journey. Thanks for having me, man. 1686 01:24:05,351 --> 01:24:08,535 This is a great conversation. I appreciate, appreciate you 1687 01:24:08,535 --> 01:24:10,221 having me on the podcast as well. 1688 01:24:10,281 --> 01:24:10,836 Yeah. Yeah.