WEBVTT 1 00:00:01.080 --> 00:00:03.799 How'd you like to listen to dot NetRocks with no ads? 2 00:00:04.440 --> 00:00:04.799 Easy? 3 00:00:05.360 --> 00:00:08.560 Become a patron for just five dollars a month. You 4 00:00:08.599 --> 00:00:11.320 get access to a private RSS feed where all the 5 00:00:11.359 --> 00:00:14.560 shows have no ads. Twenty dollars a month. We'll get 6 00:00:14.599 --> 00:00:18.440 you that and a special dot NetRocks patron mug. Sign 7 00:00:18.519 --> 00:00:34.840 up now at Patreon dot dot NetRocks dot com. Hey 8 00:00:34.840 --> 00:00:37.560 guess what. It's dot net rocks from build. I'm Carl 9 00:00:37.600 --> 00:00:40.240 Franklin Capel. We've been here for three days. This is 10 00:00:40.240 --> 00:00:42.719 our third day, and I think it's our third day. 11 00:00:42.759 --> 00:00:45.159 It's our third day, yeah, yeah, and we're here where 12 00:00:45.159 --> 00:00:48.719 the Imagined Cup finalists. Representatives from each of the three 13 00:00:48.759 --> 00:00:51.479 teams are going to interview them in a minute. But 14 00:00:51.679 --> 00:00:54.600 first let's talk about better know a framework. 15 00:00:54.799 --> 00:01:04.319 Roll the crazy music? All right? Man? What do you got? 16 00:01:04.359 --> 00:01:08.359 I went searching again for things repos on GitHub that 17 00:01:08.400 --> 00:01:13.000 are trending, and the number one repo right now is 18 00:01:13.680 --> 00:01:14.760 WSL because. 19 00:01:14.719 --> 00:01:17.840 Interesting just Windows Subsystem for Linux. Yeah, because they just 20 00:01:17.879 --> 00:01:19.159 open it, just open starts it. 21 00:01:19.239 --> 00:01:22.159 But before that, this one has been at the top 22 00:01:22.719 --> 00:01:23.799 AI hedge fund. 23 00:01:24.079 --> 00:01:25.760 Okay, what's that? Okay? 24 00:01:25.879 --> 00:01:28.640 Absolutely I haven't run it, but it's a proof of 25 00:01:28.799 --> 00:01:31.640 concept for an AI powered hedge fund. The goal of 26 00:01:31.680 --> 00:01:33.920 the project is to explore the use of AI to 27 00:01:33.920 --> 00:01:38.680 make trading decisions. This project is for educational purposes only 28 00:01:38.719 --> 00:01:41.480 and is not intended for real trading or investment. 29 00:01:41.680 --> 00:01:42.959 It seems like a great way to lose a lot 30 00:01:42.959 --> 00:01:43.680 of money fast. 31 00:01:43.840 --> 00:01:47.040 Yeah, But here's the cool thing if you think about it. It 32 00:01:47.040 --> 00:01:50.120 employs several agents working together. So this is a great 33 00:01:50.200 --> 00:01:56.159 example of agentic programming. The Aswaff demo, Darin agent, the 34 00:01:56.239 --> 00:02:00.599 dean of valuation, focuses on story numbers and discipline value. 35 00:02:00.760 --> 00:02:02.200 All right, I'm not going to read them all, but 36 00:02:02.359 --> 00:02:06.799 one of them is Charlie Munger, right, Barren Buffert's partner 37 00:02:07.079 --> 00:02:11.719 only buys wonderful businesses at fair prices. Michael Burry agent 38 00:02:11.960 --> 00:02:16.120 the big short contrarian who hunts for deep value. Peter Lynch, 39 00:02:16.599 --> 00:02:19.879 Phil Fisher, Stanley Druckenmeiler. Now the question is what do 40 00:02:19.960 --> 00:02:22.919 they train this on their data? You know they're trading 41 00:02:23.000 --> 00:02:27.280 data apparently, and you know they have big disclaimers. But 42 00:02:27.400 --> 00:02:32.319 it's very popular, and that makes me wonder how good 43 00:02:32.319 --> 00:02:32.639 it is. 44 00:02:32.840 --> 00:02:35.560 You know what they say, past results do not indicate 45 00:02:35.599 --> 00:02:36.240 future results. 46 00:02:36.319 --> 00:02:39.039 No, you're right about that, but I think you know 47 00:02:39.240 --> 00:02:42.960 understanding what we do about AI. It's either that it's 48 00:02:43.159 --> 00:02:45.520 really good clickbait and got a lot of people to 49 00:02:45.560 --> 00:02:49.159 download it, or it's working for some people. 50 00:02:49.240 --> 00:02:49.439 Yeah. 51 00:02:49.479 --> 00:02:51.120 I don't know either one of those is true, but 52 00:02:51.199 --> 00:02:53.680 something's happening for sure. Okay, So that's it. 53 00:02:53.759 --> 00:02:57.680 That's weird AI things. And this one's pretty popular, So 54 00:02:57.719 --> 00:03:00.680 who's talking to us? Gravity comment to Show nineteen forty eight, 55 00:03:00.680 --> 00:03:02.400 the one we did with Rob MenShing. We were talking 56 00:03:02.439 --> 00:03:04.360 about the open source maintenance fee and just trying to 57 00:03:04.400 --> 00:03:07.080 make open source more sustainable. Got a lot of comments 58 00:03:07.120 --> 00:03:08.919 on that show. Actually, this one comes from Rob King, 59 00:03:08.960 --> 00:03:11.080 who says, in my experience of using open source in 60 00:03:11.080 --> 00:03:14.240 a commercial setting, the biggest barrier is the financial apparatus 61 00:03:14.319 --> 00:03:17.319 to get money to pay for maintenance. We talked about 62 00:03:17.360 --> 00:03:17.919 this on the show. 63 00:03:18.759 --> 00:03:20.960 There are often forms to fill out and people to 64 00:03:20.960 --> 00:03:23.680 convince to get that payment. I suspective companies just gave 65 00:03:23.800 --> 00:03:26.000 devs a small annual stipend to spend on open source 66 00:03:26.039 --> 00:03:28.919 project the money would flow more readily. After all, if 67 00:03:28.919 --> 00:03:31.280 I've taken a dependency on open source project, I've already 68 00:03:31.319 --> 00:03:35.759 done its cost benefit analysis that seems very optimistic. Actually, Rob, 69 00:03:36.520 --> 00:03:38.240 mostly I had a problem and I made that problem 70 00:03:38.319 --> 00:03:41.159 go away. The actual money change part should be as 71 00:03:41.159 --> 00:03:43.759 low frish as possible. I came at this a different 72 00:03:43.759 --> 00:03:45.719 way during the show, where it's because I, of course 73 00:03:46.000 --> 00:03:49.240 have worked with CFOs. It's like, Hey, if we had 74 00:03:49.240 --> 00:03:51.960 a mechanism so a CFO could put one chunk of 75 00:03:52.000 --> 00:03:53.879 money once a year and it would distribute into open 76 00:03:53.960 --> 00:03:57.639 source based on utilization, they'd probably do that. Just a 77 00:03:57.719 --> 00:04:00.759 question of how hard that mechanism is developed. But Rob, 78 00:04:00.800 --> 00:04:02.960 I appreciate your thoughts, and a copy of Music Cobey 79 00:04:03.039 --> 00:04:04.159 is on its way to you. And if you'd like 80 00:04:04.199 --> 00:04:05.800 a copy of Musico buy I write a comment on 81 00:04:05.800 --> 00:04:07.919 the website at dot NetRocks dot com or on the 82 00:04:07.919 --> 00:04:09.719 facebooks we publish every show there. And if you comment 83 00:04:09.759 --> 00:04:10.879 there and I read it on the show, was that 84 00:04:10.960 --> 00:04:11.919 your copy of music Kobe? 85 00:04:12.080 --> 00:04:13.759 And for those who don't know, music to code by 86 00:04:14.000 --> 00:04:15.879 is music that you put on in the background while 87 00:04:15.879 --> 00:04:19.480 you're trying to focus. It's not distracting, it's not boring, 88 00:04:20.199 --> 00:04:25.160 and it is scientifically based in what I should I say, 89 00:04:25.160 --> 00:04:28.360 it's based in science in a study about the aspects 90 00:04:28.360 --> 00:04:29.800 of music that help people focus. 91 00:04:29.879 --> 00:04:32.480 So yeah, there's after they did that show with Mark Semen. 92 00:04:32.560 --> 00:04:35.040 We were talking about flow and focus and you then 93 00:04:35.600 --> 00:04:36.160 made a thing. 94 00:04:36.319 --> 00:04:36.879 I made a thing. 95 00:04:37.160 --> 00:04:39.360 So there's twenty two tracks now they're twenty five minutes 96 00:04:39.399 --> 00:04:42.079 each to correspond with a Pomodora technique and you get 97 00:04:42.120 --> 00:04:45.480 them at musicdocodebuy dot net, or as Richard said, by 98 00:04:45.519 --> 00:04:49.439 sending us a comment, Okay, let us talk about the 99 00:04:49.639 --> 00:04:54.439 imagined cup we have, right, you know what, You're right, 100 00:04:54.480 --> 00:04:56.519 we should do nineteen fifty six, but I did not 101 00:04:56.639 --> 00:05:00.240 prepare anything. So this is episode nineteen fifty six. And 102 00:05:00.519 --> 00:05:02.839 what we've been doing lately is talking about what happened 103 00:05:02.879 --> 00:05:05.959 that year in world events. M So, I don't know 104 00:05:05.959 --> 00:05:07.120 you've got something up there. 105 00:05:07.120 --> 00:05:09.839 We're beginning of the Suez crisis. Oh, that was the 106 00:05:09.839 --> 00:05:15.360 invasion of Britain and trying to control that the waterway there, 107 00:05:15.360 --> 00:05:18.839 which is very big deal. Yeah, and the first time 108 00:05:19.120 --> 00:05:22.560 Elvis Presley became famous, right, the first detection of a 109 00:05:22.600 --> 00:05:25.279 neutrinoy an undetectable particle. 110 00:05:25.000 --> 00:05:27.920 An undetectable particle that can go through walls and lead 111 00:05:27.959 --> 00:05:31.120 and everything else. How did they even detect it with really? 112 00:05:31.879 --> 00:05:34.399 Once in a while the Trino's decay. It's very rare 113 00:05:34.959 --> 00:05:38.480 and by far most importantly, General Electric releases the first 114 00:05:38.759 --> 00:05:42.600 snoozeable alarm clock. Wow, there you go, but then you 115 00:05:42.600 --> 00:05:43.399 couldn't snooze. 116 00:05:43.600 --> 00:05:43.920 Wow. 117 00:05:44.680 --> 00:05:47.120 I wonder if it sounded like modern techno music. 118 00:05:47.160 --> 00:05:51.639 I'm thinking that's the thing. The beat nineteen fifty six. 119 00:05:51.800 --> 00:05:55.079 Yeah, so all right, well there's more stuff to talk 120 00:05:55.120 --> 00:05:57.079 about in nineteen fifty six, but we want to talk 121 00:05:57.120 --> 00:06:02.199 to the Imagined Cup finalists builds. So we have representatives 122 00:06:02.279 --> 00:06:06.759 of each team, just one representative from each team. So 123 00:06:06.800 --> 00:06:10.040 I'll start with Matt. Matt Steele is on the hair 124 00:06:10.079 --> 00:06:13.680 Match team. He's a Georgia Tech computer science grad with 125 00:06:13.720 --> 00:06:17.759 a concentration in AI and embedded systems. He has published 126 00:06:17.800 --> 00:06:20.399 more than five apps to the iOS app Store with 127 00:06:20.600 --> 00:06:24.560 over twenty thousand monthly active users. Matt's also a former 128 00:06:24.800 --> 00:06:28.160 US Olympic Trials qualifier in swimming and took fifth place 129 00:06:28.480 --> 00:06:30.160 at the US Open in swimming. 130 00:06:30.279 --> 00:06:31.399 Wow. How about that? 131 00:06:31.920 --> 00:06:34.240 And on top of that, you're in the Imagined Cup 132 00:06:34.279 --> 00:06:39.480 Finalists And so that's from the hair Match team. Guida 133 00:06:39.519 --> 00:06:44.279 Abdallah Omar is on the sign Verse team, and she 134 00:06:44.399 --> 00:06:48.720 is a passionate innovator and project associative Signverse that's sign 135 00:06:48.879 --> 00:06:53.800 v r S, where she leverages technology to address social challenges. 136 00:06:54.480 --> 00:06:57.920 Is the founder of Girls I Save, an initiative empowering 137 00:06:57.920 --> 00:07:01.720 girls in STEM and advocating against gender based violence. She 138 00:07:01.839 --> 00:07:06.000 brings a strong commitment to community driven tax solutions. A 139 00:07:06.040 --> 00:07:11.120 Guide is also a Generation Connect Youth envoy with the ITU, 140 00:07:11.800 --> 00:07:16.040 a judge for Young Scientists Kenya, and a recognized change 141 00:07:16.040 --> 00:07:19.399 maker in the Coastal region of Kenya. Her work combines 142 00:07:19.439 --> 00:07:25.560 information technology, advocacy and innovation to create inclusive, impactful change. 143 00:07:25.800 --> 00:07:30.879 Welcome and Daniel Kim is studying math in neuroscience at 144 00:07:30.920 --> 00:07:35.079 Stamford and seeing his grandmother struggle with her visual impairment, 145 00:07:35.160 --> 00:07:38.720 Daniel was inspired to pursue Argus, which is one of 146 00:07:38.759 --> 00:07:41.759 the teams in the finalists. In his free time, he 147 00:07:41.800 --> 00:07:45.800 loves spending time with animals in any capacity he can. Okay, well, 148 00:07:45.839 --> 00:07:50.120 welcome guys to dot ne Rocks and congratulations, thank you 149 00:07:50.160 --> 00:07:53.519 for having us. You bet, Matt, Let's start with you. 150 00:07:53.600 --> 00:07:55.120 So here. 151 00:07:56.480 --> 00:08:00.240 Match is the team that you are on and tell 152 00:08:00.279 --> 00:08:01.199 us a little bit about that. 153 00:08:01.439 --> 00:08:01.959 Yeah sure. 154 00:08:02.040 --> 00:08:02.120 So. 155 00:08:02.720 --> 00:08:05.519 Hair Match is a mobile app which allows women with 156 00:08:05.600 --> 00:08:07.959 curly hair to take pictures of their hair, and then 157 00:08:07.959 --> 00:08:10.879 we use AI to be able to determine their hair type, 158 00:08:10.920 --> 00:08:13.079 and from there we've done the R and D to 159 00:08:13.079 --> 00:08:15.560 be able to curate products based on their specific hair 160 00:08:15.600 --> 00:08:17.920 type that work for them. And so there's a very 161 00:08:17.920 --> 00:08:21.040 big problem with curly hair care where people will spend 162 00:08:21.079 --> 00:08:23.480 a lot of money on products and then the products 163 00:08:23.480 --> 00:08:25.639 don't actually work for their hair because there's something really 164 00:08:25.680 --> 00:08:28.279 specific about that one product, you know, might be one 165 00:08:28.319 --> 00:08:31.720 chemical or one molecule that's often the formulation that doesn't 166 00:08:31.759 --> 00:08:33.919 work for that person's hair type. And then you end 167 00:08:34.000 --> 00:08:35.759 up with a bunch of products under yourself. And so 168 00:08:36.159 --> 00:08:39.279 my co founder, Relisha, has very very curly, very long hair, 169 00:08:39.639 --> 00:08:41.360 and this is a problem that she faced first. 170 00:08:41.399 --> 00:08:44.159 I imagine just I've seen shelves like that just yeah, 171 00:08:44.200 --> 00:08:45.480 bottles that were used once. 172 00:08:46.159 --> 00:08:48.240 Yeah, just you know, you use it once, but once 173 00:08:48.240 --> 00:08:49.759 you use it once, you can't return it. And so 174 00:08:50.159 --> 00:08:52.919 she just had this. She called it the shelf of shame. 175 00:08:53.240 --> 00:08:56.600 So it's like our phones. 176 00:08:57.480 --> 00:09:00.519 And so I've been friends with her since we interned 177 00:09:00.559 --> 00:09:03.279 together about two years ago. We were both college athletes. 178 00:09:03.320 --> 00:09:05.600 She cheered at University of North Carolina and I swam 179 00:09:05.639 --> 00:09:08.399 at Georgia Tech, and so we just became friends over that, 180 00:09:08.440 --> 00:09:11.639 and then eventually we both are really interested in building things, 181 00:09:11.679 --> 00:09:13.919 and so we decided to make hair match and that's 182 00:09:14.000 --> 00:09:14.840 kind of how it was born. 183 00:09:14.960 --> 00:09:18.919 Now she's technological also, or was she just interested in 184 00:09:19.039 --> 00:09:19.759 the problem. 185 00:09:20.159 --> 00:09:23.879 So she she is technical, she can code, but she 186 00:09:24.080 --> 00:09:27.320 primarily her strengths are in the media and the product 187 00:09:27.320 --> 00:09:30.320 side of things, and so primarily those two things were 188 00:09:30.360 --> 00:09:31.879 her focus on the team. And then she has a 189 00:09:32.000 --> 00:09:34.519 very very deep understanding of the problem because she is 190 00:09:34.559 --> 00:09:35.519 the customer. 191 00:09:35.120 --> 00:09:37.360 Of her Perfect Perfect Domain Expert. 192 00:09:37.639 --> 00:09:40.960 So people listening might be going, what you know, that 193 00:09:41.639 --> 00:09:44.720 seems kind of silly, But you know, I have two 194 00:09:44.720 --> 00:09:49.039 stepdaughters with curly hair, and they struggled all their growing 195 00:09:49.120 --> 00:09:52.879 up lives and into adulthood probably still do now with 196 00:09:53.120 --> 00:09:57.320 this problem. I'm you know, straightening hair was and just 197 00:09:57.440 --> 00:10:01.639 combing it could be a huge shoe, right, And some 198 00:10:01.679 --> 00:10:05.039 people want to straighten curly hair. Some people want to 199 00:10:05.039 --> 00:10:07.799 have smaller curls when they have big curls. Like there's 200 00:10:07.799 --> 00:10:11.440 a whole bunch of styling things and just maintenance. 201 00:10:11.600 --> 00:10:13.960 Yeah, and there's a lot of societal pressure as well 202 00:10:14.000 --> 00:10:16.159 on people with curly hair and either straighten it or 203 00:10:16.200 --> 00:10:19.320 to keep it in very traditional hairstyles. They've been commonly 204 00:10:19.320 --> 00:10:22.879 accepted in the workplace for decades here especially, I guess 205 00:10:22.879 --> 00:10:25.120 primarily in the US because that's where we're based out of. 206 00:10:25.240 --> 00:10:27.639 But you know, one of the things that Alicia and 207 00:10:27.639 --> 00:10:29.759 I believe in is we want people to be able 208 00:10:29.799 --> 00:10:32.720 to have equal opportunity professionally and from a self confidence 209 00:10:32.720 --> 00:10:35.320 perspective as well, and so being able to build something 210 00:10:35.679 --> 00:10:38.919 that directly improves people's ability to style their hair and 211 00:10:39.360 --> 00:10:40.519 leave their confidence is great. 212 00:10:40.519 --> 00:10:42.360 We can go to the app store and download hair 213 00:10:42.399 --> 00:10:43.080 Match right now. 214 00:10:43.200 --> 00:10:45.279 Yeah, it's been live for six months now. 215 00:10:45.320 --> 00:10:47.519 Actually, wow, what's the response been. 216 00:10:48.039 --> 00:10:51.360 The response has been great. So in the app, we 217 00:10:51.399 --> 00:10:53.360 did for a while have my phone number in there. 218 00:10:53.399 --> 00:10:55.080 People could just text me if there was a problem, 219 00:10:55.600 --> 00:10:58.480 and I would just get these texts at random hours 220 00:10:58.519 --> 00:11:01.080 of the day like I love this, this is amazing, guys, 221 00:11:01.080 --> 00:11:03.240 you're doing great. We get emails like that as well, 222 00:11:03.240 --> 00:11:05.879 and so the response has been overwhelmingly positive, and that's 223 00:11:05.919 --> 00:11:07.039 something that we've really enjoyed. 224 00:11:07.159 --> 00:11:08.919 Talk to me about the workflow, like, how does people 225 00:11:09.000 --> 00:11:09.519 use the app. 226 00:11:09.639 --> 00:11:12.320 Yeah, So a user will download the app from the 227 00:11:12.360 --> 00:11:14.559 app Store and they go through onboarding. We ask them 228 00:11:14.600 --> 00:11:17.480 a few questions about their experience with hair and the 229 00:11:17.679 --> 00:11:19.759 sort of what they already know versus not, so we 230 00:11:19.799 --> 00:11:21.840 can tailor the experience a little bit to that. Then 231 00:11:22.279 --> 00:11:24.600 the primary magic is we have them take two pictures 232 00:11:24.600 --> 00:11:27.279 of their hair very up close in good lighting, so 233 00:11:27.320 --> 00:11:28.519 we're able to actually. 234 00:11:28.240 --> 00:11:29.639 See that in the more in detail. 235 00:11:29.679 --> 00:11:32.480 We process that using a visual transformer that we run 236 00:11:32.559 --> 00:11:35.240 in Azure ai foundry, and we're able to get the 237 00:11:35.240 --> 00:11:37.120 hair type. Once we get that hair type, we give 238 00:11:37.159 --> 00:11:39.639 them a nice little savable card that they can save 239 00:11:39.679 --> 00:11:42.639 to the chemeeral. It's like, oh, this is your hair analysis. 240 00:11:42.480 --> 00:11:43.519 It's really cool. 241 00:11:43.519 --> 00:11:46.240 We it's like five or six different categories that you 242 00:11:46.440 --> 00:11:47.320 check exactly. 243 00:11:47.519 --> 00:11:50.840 We have five different properties on that card that they 244 00:11:50.840 --> 00:11:53.639 can see. And then from there they're able to enter 245 00:11:53.639 --> 00:11:56.879 the app and they can go to the product recommendations list, 246 00:11:56.919 --> 00:12:01.200 which is curated directly for their specific match of those properties, 247 00:12:01.279 --> 00:12:03.240 and it goes, Okay, this is what this person needs, 248 00:12:03.279 --> 00:12:05.399 and it shows them only those products, and they're all 249 00:12:05.440 --> 00:12:08.200 sorted by the category as well, so whether it's gel, 250 00:12:08.240 --> 00:12:11.039 whether it's shampoo, whether it's conditioner, they can get easy 251 00:12:11.080 --> 00:12:13.039 access to those. And so instead of having to go 252 00:12:13.120 --> 00:12:15.399 to a store and you have, you know, a million 253 00:12:15.399 --> 00:12:17.320 products on the shelves, you don't know which ones actually 254 00:12:17.320 --> 00:12:19.799 work for you unless you, you know, look each one 255 00:12:19.840 --> 00:12:21.320 up one by one, and you have to go to 256 00:12:21.360 --> 00:12:23.120 some obscure web page to find that. We have that 257 00:12:23.320 --> 00:12:25.840 information all in front of the user, all at once, 258 00:12:26.039 --> 00:12:27.600 and it's very streamlined, very easy. 259 00:12:27.720 --> 00:12:29.240 And of course my instinct is now you'll go to 260 00:12:29.279 --> 00:12:30.720 the store with the list and look at that. Was like, no, 261 00:12:30.799 --> 00:12:32.519 you won't, You'll go on Amazon in order. 262 00:12:33.159 --> 00:12:35.200 And so the other thing we do is we have 263 00:12:35.279 --> 00:12:37.720 links to every single product inside the app, so we 264 00:12:37.759 --> 00:12:40.360 don't actually sell the products, but we connect the people 265 00:12:40.399 --> 00:12:42.200 to the source. And so what we've actually been able 266 00:12:42.240 --> 00:12:44.639 to do with that is we are able to offer 267 00:12:44.679 --> 00:12:48.000 discounts on those products because we reach out to the 268 00:12:48.159 --> 00:12:51.320 brand and manufacturers that we've found through our R and. 269 00:12:51.360 --> 00:12:53.080 D to be a monetizing strategy. 270 00:12:53.080 --> 00:12:56.480 Then, not necessarily for the modestation, we did it more 271 00:12:56.519 --> 00:13:00.440 for the customer. We do get the small affiliates, but 272 00:13:00.879 --> 00:13:03.399 it's been primarily good for us because we're able to 273 00:13:03.440 --> 00:13:06.279 not only you know, give the customer exactly what they want, 274 00:13:06.279 --> 00:13:08.519 but we can also save the money, which is really 275 00:13:08.559 --> 00:13:09.120 really cool. 276 00:13:09.360 --> 00:13:12.799 Yeah, that is cool and congratulations, that's a great thing. 277 00:13:12.960 --> 00:13:13.320 Thank you. 278 00:13:14.080 --> 00:13:17.039 Just curious and any of you can answer this. How 279 00:13:17.039 --> 00:13:19.720 many other teams did you guys beat out to get 280 00:13:19.759 --> 00:13:20.559 in the top three? 281 00:13:20.720 --> 00:13:22.960 A lot? But yeah, the match kept huge. 282 00:13:23.120 --> 00:13:26.200 Yeah, I think it was. There were over fifteen thousand 283 00:13:26.240 --> 00:13:28.000 students that participated. 284 00:13:27.399 --> 00:13:30.200 This year, one and fifty countries. 285 00:13:30.679 --> 00:13:31.120 Wow. 286 00:13:31.639 --> 00:13:34.720 So just being here and you guys were on stage 287 00:13:34.799 --> 00:13:38.320 right before the keynote, yes, on the first day of build. 288 00:13:38.639 --> 00:13:39.919 That must have been really exciting. 289 00:13:40.320 --> 00:13:42.919 Oh yeah, yeah, it was. It was amazing, Just a 290 00:13:42.960 --> 00:13:45.519 great experience to be up there. You know, you look 291 00:13:45.600 --> 00:13:48.399 into the crowd and you know, the lights are a 292 00:13:48.399 --> 00:13:50.440 little bit blinding on stage, but you can see so 293 00:13:50.679 --> 00:13:53.120 many people and it really hits you in that moment 294 00:13:53.200 --> 00:13:53.919 that you know. 295 00:13:53.840 --> 00:13:55.440 You're there thousands of people in them. 296 00:13:55.600 --> 00:13:57.720 Yeah, it's like you've made something impact. 297 00:13:57.519 --> 00:13:59.960 We have been hundreds of yers watching online. 298 00:14:00.120 --> 00:14:02.559 Yeah, not the freaky out that it was like thinking, 299 00:14:02.840 --> 00:14:05.000 you know, there are almost as many people in there 300 00:14:05.399 --> 00:14:07.120 as we had users on hair Match in the first 301 00:14:07.120 --> 00:14:08.960 couple of months. Sure, that's insane. 302 00:14:09.039 --> 00:14:11.240 Yeah. Well, also realization like that, how many people you 303 00:14:11.279 --> 00:14:12.159 attached to? Yeah? 304 00:14:12.240 --> 00:14:14.919 Yeah, it seems like a niche product, but clearly for 305 00:14:14.960 --> 00:14:17.399 a certain group of folks it's really important, really important. 306 00:14:17.919 --> 00:14:21.039 We estimate that there are probably about a billion people 307 00:14:21.320 --> 00:14:23.679 or more that could benefit from our product in this 308 00:14:23.759 --> 00:14:24.279 current state. 309 00:14:24.480 --> 00:14:27.960 Wow, should we switch to you gata? Should we talk 310 00:14:28.000 --> 00:14:29.000 about sign Verse? 311 00:14:29.080 --> 00:14:29.440 Of course? 312 00:14:29.840 --> 00:14:33.639 Yeah? So what is the product and tell us about? 313 00:14:33.759 --> 00:14:36.799 The sign Verse is an AI pod platform that chooses 314 00:14:36.879 --> 00:14:40.519 three D avatar to convert speech and text to real 315 00:14:40.559 --> 00:14:42.120 time sign language interpretation. 316 00:14:42.360 --> 00:14:43.960 Wow. Wow, that's really cool. 317 00:14:44.039 --> 00:14:48.600 So in basic what it's communication breaking the barriers between 318 00:14:48.600 --> 00:14:50.919 a person who's had of hearing and a person who 319 00:14:50.960 --> 00:14:54.600 can hear, because sometimes it becomes very challenging when people 320 00:14:54.639 --> 00:14:56.600 you know, communicate and someone cannot hear. 321 00:14:57.000 --> 00:14:59.120 So with the app, you can be able to type 322 00:14:59.120 --> 00:15:00.919 in list hello, good morning. 323 00:15:00.840 --> 00:15:03.440 So the avatar will be able to do the sign language, 324 00:15:03.519 --> 00:15:05.399 and then the person who is next will be able 325 00:15:05.440 --> 00:15:08.320 to see that through the application. So you can either 326 00:15:08.440 --> 00:15:11.919 use text or speech and making communications simless for both 327 00:15:11.919 --> 00:15:13.559 the hearing and the head of here. 328 00:15:13.639 --> 00:15:15.879 So I could hold up my phone just talk and 329 00:15:15.919 --> 00:15:16.720 it's going to do the. 330 00:15:16.720 --> 00:15:22.200 Sign language, Yes, absolutely, And we drove the inspiration because 331 00:15:22.960 --> 00:15:25.759 Brandis and I used to participate in Young Scientist Kenya 332 00:15:26.080 --> 00:15:29.000 That's Pack Home, and we used to do robotics training 333 00:15:29.080 --> 00:15:31.919 for young children and one time we found ourselves in 334 00:15:31.960 --> 00:15:35.000 a situation where the students that we're supposed to train 335 00:15:35.399 --> 00:15:37.519 were not able to hear, like the head of hearing 336 00:15:37.559 --> 00:15:39.840 and some of them were completely deaf. And you know, 337 00:15:39.919 --> 00:15:42.679 the amount of signed interpreters in the room like what 338 00:15:42.919 --> 00:15:45.879 to go to different classes, so there was already a gap. 339 00:15:46.159 --> 00:15:48.480 We had to you know, find ways to be able 340 00:15:48.559 --> 00:15:51.679 to bring that message right in terms of coding and 341 00:15:51.720 --> 00:15:55.000 talking to the student. But ever since then we realize 342 00:15:55.200 --> 00:15:57.960 there is a gap actually, and then we met Daniel 343 00:15:57.960 --> 00:16:00.159 who is one of our team who literally had to 344 00:16:00.159 --> 00:16:03.039 switch from being a computer student to being in the 345 00:16:03.080 --> 00:16:06.879 business because the school did not have sign language interpreters 346 00:16:06.879 --> 00:16:10.080 for that cause. So you can imagine the change he 347 00:16:10.200 --> 00:16:13.440 literally had to switch. And you know, it also comes 348 00:16:13.480 --> 00:16:16.159 with a lot of emotions to these people, they feel neglated. 349 00:16:16.559 --> 00:16:19.559 The community, you know, pushes them away because of the 350 00:16:19.639 --> 00:16:21.879 stigma of them not being able to hear. And that's 351 00:16:21.879 --> 00:16:25.000 why we were driven to, you know, create a technology 352 00:16:25.000 --> 00:16:28.080 that's going to focus more on providing real time sign 353 00:16:28.159 --> 00:16:30.559 language interpretation at different levels. 354 00:16:30.600 --> 00:16:33.159 So you just mentioned sign language interpretations, so this app 355 00:16:33.200 --> 00:16:35.080 will also you can hold it up to somebody who's 356 00:16:35.120 --> 00:16:37.559 signing and it will say the words to you. 357 00:16:38.000 --> 00:16:40.639 So that is like a future vision that we want 358 00:16:40.679 --> 00:16:40.919 to do. 359 00:16:41.000 --> 00:16:43.879 For our first product, it's just converting speech to text 360 00:16:44.120 --> 00:16:46.960 and then to sign language. But in future we believe 361 00:16:47.000 --> 00:16:49.240 that we will be able to do the vice versa. 362 00:16:49.320 --> 00:16:50.960 And that's why we have signed verse so that we 363 00:16:50.960 --> 00:16:52.799 can go from sign to verse and back. 364 00:16:52.879 --> 00:16:53.679 Yesh wow. 365 00:16:54.159 --> 00:16:58.159 Because we used to have remember the Connect Microsoft Connect 366 00:16:58.519 --> 00:17:01.480 and it could track all of your joints in three 367 00:17:01.519 --> 00:17:06.680 D space. Yes, I think it was twenty twenty frames 368 00:17:06.680 --> 00:17:10.920 pertective like that, So there were applications that use that. 369 00:17:11.359 --> 00:17:13.720 But I think it wouldn't be more challenging on a 370 00:17:13.799 --> 00:17:18.359 phone to go from straight video to generative AI now right. 371 00:17:18.440 --> 00:17:21.160 Yes, we're actually trying to incorporate the use of AI 372 00:17:21.279 --> 00:17:23.319 to be able to make it more seamless and more 373 00:17:23.400 --> 00:17:24.599 interactive because. 374 00:17:24.359 --> 00:17:27.359 Even that Spelton tracking had problem with hands. Yeah, and 375 00:17:27.640 --> 00:17:30.000 sign of course it depends on the flavor signed too, 376 00:17:30.000 --> 00:17:30.359 doesn't it. 377 00:17:30.440 --> 00:17:33.759 Yes, we have different sign language and because of the 378 00:17:33.759 --> 00:17:37.559 motion capture suits that we acquired and we incorporate the 379 00:17:37.599 --> 00:17:40.759 different data sets from Kenyan sign language to Ruandan Sign 380 00:17:40.880 --> 00:17:43.799 language to even SL So in future we're hoping to 381 00:17:43.839 --> 00:17:47.680 incorporate or different sign language based on the context of 382 00:17:47.720 --> 00:17:49.599 the country and the community that it. 383 00:17:49.559 --> 00:17:51.960 Comes literally country to country, the sign language varias. 384 00:17:52.160 --> 00:17:55.599 Yeah, even in Kenyan Sign language itself, it's narrowed down 385 00:17:55.680 --> 00:17:58.359 to different community sign language stylists. 386 00:17:58.400 --> 00:17:59.880 Yes, yeah, regular language. 387 00:18:00.160 --> 00:18:01.079 It's just a language. 388 00:18:01.119 --> 00:18:03.039 It's just that it's now in form of you know, 389 00:18:03.480 --> 00:18:05.680 hand motion gestures and stuff. 390 00:18:05.839 --> 00:18:09.359 When you have your every area, every dialect has its 391 00:18:09.400 --> 00:18:10.920 own idioms and things like that. 392 00:18:11.079 --> 00:18:13.000 Yes, other people may not. 393 00:18:12.960 --> 00:18:14.759 Understand just collecting all that data it's going to be 394 00:18:14.759 --> 00:18:16.720 really powerful, exactly, But I don't know that there's a 395 00:18:16.720 --> 00:18:17.839 good repository for. 396 00:18:17.799 --> 00:18:18.640 All of the variations. 397 00:18:18.640 --> 00:18:21.440 We're actually finding very hard in Africa to get like 398 00:18:21.480 --> 00:18:23.880 a very concrete data set when it comes to sign 399 00:18:23.960 --> 00:18:28.240 language interpretation. So far, we've gotten over twenty three hundred 400 00:18:28.240 --> 00:18:32.000 glosses in terms of the sign language interpretation, but we're 401 00:18:32.039 --> 00:18:35.279