WEBVTT 00:00:01.820 --> 00:00:04.320 All right, welcome to the AdaptX podcast where 00:00:04.320 --> 00:00:06.339 we have conversations with individuals building 00:00:06.339 --> 00:00:08.800 accessible businesses, advocating for inclusion 00:00:08.800 --> 00:00:11.740 or excelling in adaptive sports. And joined by 00:00:11.740 --> 00:00:13.960 Shay Cohen, the co -founder and CEO of Verbali, 00:00:14.160 --> 00:00:16.660 an AI powered augmentative and alternative communication 00:00:16.660 --> 00:00:18.600 platform that helps children communicate more 00:00:18.600 --> 00:00:21.000 independently in everyday life. A veteran product 00:00:21.000 --> 00:00:23.199 leader with more than 15 years in tech, Shay 00:00:23.199 --> 00:00:25.739 has led high impact teams at Fortune 500s and 00:00:25.739 --> 00:00:27.820 startups building payment systems, real time 00:00:27.820 --> 00:00:30.519 communication platforms and consumer apps. Shay, 00:00:30.559 --> 00:00:32.679 thanks for joining me this morning. Thank you, 00:00:32.759 --> 00:00:35.719 Brendan, for letting me join your process. I'm 00:00:35.719 --> 00:00:39.340 excited. Yeah. So I came across Verbali on LinkedIn. 00:00:39.719 --> 00:00:42.479 I was really excited to see it because for a 00:00:42.479 --> 00:00:46.219 long time, I've wondered how these AAC tools, 00:00:46.500 --> 00:00:48.780 selfishly, can play more of a role in fitness 00:00:48.780 --> 00:00:51.079 environments where I typically operate out of. 00:00:51.179 --> 00:00:53.020 And so I'm sure we'll get to that at some point, 00:00:53.039 --> 00:00:55.240 but maybe let's start with the origin story. 00:00:55.299 --> 00:00:58.600 I know it's personal to you. So what inspired 00:00:58.600 --> 00:01:02.689 you to start Verbali? Yeah, absolutely. So I 00:01:02.689 --> 00:01:05.450 guess the easiest way to say it's because my 00:01:05.450 --> 00:01:09.890 kids So I have three kids and my oldest her name 00:01:09.890 --> 00:01:12.069 is Leah middle one. His name is Matan It means 00:01:12.069 --> 00:01:15.269 the reason we'll talk about in a second and my 00:01:15.269 --> 00:01:19.609 youngest. She's my latest boss Her name is Miley 00:01:19.609 --> 00:01:26.189 and so Matan He's he's seven who just turned 00:01:26.189 --> 00:01:29.430 seven And he's not verbally as a developmental 00:01:29.430 --> 00:01:33.909 delays and one goes to an amazing school regular 00:01:33.909 --> 00:01:36.590 public school is special class but one of the 00:01:36.590 --> 00:01:38.750 first thing that you'll notice with him is that 00:01:38.750 --> 00:01:42.170 he's he's not verbal and luckily for us and I 00:01:42.170 --> 00:01:45.530 always say this kid is a rock star is super charismatic 00:01:45.530 --> 00:01:49.230 and isn't. Amazing communicator, despite not 00:01:49.230 --> 00:01:54.329 being to verbalize himself is is super is super 00:01:54.329 --> 00:01:57.409 communicative. He'll take your hand. He'll he'll 00:01:57.409 --> 00:02:00.750 point. He'll try to say the beginning of a word. 00:02:01.329 --> 00:02:04.170 And so he's never frustrated. So that's why me 00:02:04.170 --> 00:02:06.790 and my wife, Anna, we always consider ourselves 00:02:06.790 --> 00:02:10.210 very lucky because the kid has a lot of to say 00:02:10.210 --> 00:02:14.909 and is able to communicate. Right. But. So me 00:02:14.909 --> 00:02:19.889 as the dad, I always was able to understand what 00:02:19.889 --> 00:02:24.370 he wants. But I think this is how it became clear 00:02:24.370 --> 00:02:27.330 to me that we need something better, because 00:02:27.330 --> 00:02:31.169 as a dad, I know what he wants, but I want him 00:02:31.169 --> 00:02:34.710 to be able to express himself, not just to the 00:02:34.710 --> 00:02:37.810 people around him that didn't know him and they 00:02:37.810 --> 00:02:40.610 know how he communicates. He goes to school, 00:02:40.610 --> 00:02:43.539 I want him to be successful at school. He, I 00:02:43.539 --> 00:02:47.060 wanted to have meaningful connections. And to 00:02:47.060 --> 00:02:50.139 all of this, you need to have meaningful conversations, 00:02:50.460 --> 00:02:54.419 right? You need to have real, genuine conversations. 00:02:54.979 --> 00:02:58.560 And for that, you need those conversations to 00:02:58.560 --> 00:03:01.939 be natural. And the same way that you and I have, 00:03:01.939 --> 00:03:04.379 right? Like we just before we started, we talked 00:03:04.379 --> 00:03:06.340 about a little bit of our background. I told 00:03:06.340 --> 00:03:08.300 I explained to you how to pronounce my name, 00:03:08.439 --> 00:03:10.719 how to pronounce Verbali, right? We had a little 00:03:10.719 --> 00:03:15.259 bit of jokes and but it because this fluency, 00:03:15.719 --> 00:03:19.759 the ability to switch between topic quickly. 00:03:20.479 --> 00:03:25.560 This, I would say this luxury for us is something 00:03:25.560 --> 00:03:30.469 that he doesn't have. And I want him to have 00:03:30.469 --> 00:03:34.590 it. So that's why, that's how it started. I can, 00:03:34.689 --> 00:03:36.990 we'll probably dive more about it, but this is 00:03:36.990 --> 00:03:39.330 just, you know, like the, the origin story of 00:03:39.330 --> 00:03:42.129 what, what was the motivation for me to start 00:03:42.129 --> 00:03:45.389 thinking about the better way. Yeah. So the latency 00:03:45.389 --> 00:03:50.129 between you making a comment, asking a question 00:03:50.129 --> 00:03:53.889 and the response to the traditional AAC software 00:03:53.889 --> 00:03:57.969 provided is kind of what makes it. not fluent. 00:03:58.750 --> 00:04:02.129 So how does verbally, verbally, we'll probably 00:04:02.129 --> 00:04:05.250 go back and forth on the pronunciation, how does 00:04:05.250 --> 00:04:10.210 it address those maybe time lapses in the conversation? 00:04:11.159 --> 00:04:13.340 Absolutely. Yeah. So one of the thing I noticed 00:04:13.340 --> 00:04:16.339 that today when it's I don't know how many folks 00:04:16.339 --> 00:04:19.399 from out from the podcast know what AAC is. I'll 00:04:19.399 --> 00:04:21.560 give a quick sentence of what it is. The best 00:04:21.560 --> 00:04:24.220 way to describe it is a tool. So it stands for 00:04:24.220 --> 00:04:27.160 augmentative and alternative communication system. 00:04:27.579 --> 00:04:30.319 And the best way to think about is a tablet that 00:04:30.319 --> 00:04:33.259 acts as a keyboard. But instead of letters, you'll 00:04:33.259 --> 00:04:35.259 have images for words. So if you want to say 00:04:35.259 --> 00:04:37.620 I want to eat the banana. So they have an image 00:04:37.620 --> 00:04:40.939 for I image for one. banana and so on, and you 00:04:40.939 --> 00:04:43.439 build a sentence out of it. Now, everything is 00:04:43.439 --> 00:04:45.879 great and everything when you have a simple sentence 00:04:45.879 --> 00:04:49.839 like, I want a banana. But once you try to have 00:04:49.839 --> 00:04:52.759 more complex conversations, you need to have 00:04:52.759 --> 00:04:56.839 a bigger size of vocabulary bank, right? Meaning 00:04:56.839 --> 00:04:59.759 you need more images. And where the problem is, 00:04:59.800 --> 00:05:02.579 the problem is that you have limited real estate 00:05:02.579 --> 00:05:05.259 on your screen. Right? The more images, the more 00:05:05.259 --> 00:05:07.800 words you want, you need more images. So it becomes 00:05:07.800 --> 00:05:11.079 busy on the screen. And then you go into drill 00:05:11.079 --> 00:05:14.300 down system. So you need to like, you go to verbs, 00:05:14.339 --> 00:05:16.980 you have any page for verbs or page for fruits 00:05:16.980 --> 00:05:19.160 and stuff like that. So you become like chasing 00:05:19.160 --> 00:05:22.560 the word that you're looking for. And that chasing, 00:05:23.519 --> 00:05:27.740 it's time consuming. It's energy consuming. So 00:05:28.009 --> 00:05:30.389 That's where the problem is that if you want 00:05:30.389 --> 00:05:33.610 to have a fluency conversation, you want those 00:05:33.610 --> 00:05:37.230 words to come really, really fast to you, right? 00:05:37.449 --> 00:05:39.990 Again, going back to how you and I are speaking, 00:05:40.149 --> 00:05:44.069 we don't think about ours, we just speak. And 00:05:44.069 --> 00:05:47.990 today with AAC technology, before Verbali, you 00:05:47.990 --> 00:05:52.189 really needed to plan your sentence. And planning 00:05:52.189 --> 00:05:57.550 takes time. Planning takes away from the natural. 00:05:58.470 --> 00:06:02.550 experience. So what happens that kids or adults 00:06:02.550 --> 00:06:07.949 who use AAC, the conversation flow breaks because 00:06:07.949 --> 00:06:12.350 now they're building the sentence. So the downside 00:06:12.350 --> 00:06:17.250 of it is that, and it can take 45 seconds on 00:06:17.250 --> 00:06:22.889 average. So unfortunately, the downside of it 00:06:22.889 --> 00:06:24.870 is that, yeah, you might get a good sentence, 00:06:24.889 --> 00:06:30.420 maybe, eventually. but the other side, the partner 00:06:30.420 --> 00:06:33.279 for the conversation, will either, if they're 00:06:33.279 --> 00:06:35.199 a kid, they might walk away because they get 00:06:35.199 --> 00:06:40.519 bored, or they'll change topics. In both cases, 00:06:40.720 --> 00:06:44.500 the outcome is that the kid wasn't able to express 00:06:44.500 --> 00:06:47.899 themselves, right? So with Verbali, what we're 00:06:47.899 --> 00:06:50.519 doing, instead of the kid chasing for the word, 00:06:50.620 --> 00:06:55.199 instead of the kid... Looking what is the planning 00:06:55.199 --> 00:07:00.480 for for the sentence. We are using AI to predict 00:07:00.480 --> 00:07:03.500 help predict what the kid might want to speak 00:07:03.500 --> 00:07:08.519 of right and responding much faster and for it 00:07:08.519 --> 00:07:10.920 for the for the conversation itself. So we were 00:07:10.920 --> 00:07:15.720 put the app which is called Ma-Talk AI and there's 00:07:15.720 --> 00:07:19.040 a nice story about the rhythm name behind it 00:07:19.040 --> 00:07:23.149 and. basically understand the conversations. 00:07:23.569 --> 00:07:25.810 We're using AI to understand the conversation, 00:07:25.949 --> 00:07:30.170 using AI to take context signals from the environment, 00:07:31.089 --> 00:07:33.910 like the weather, the time of day, are we in 00:07:33.910 --> 00:07:36.069 a holiday, right? If it's Christmas, do you have 00:07:36.069 --> 00:07:38.009 different topics that you might talk about in 00:07:38.009 --> 00:07:40.230 Christmas that you won't talk about Thanksgiving 00:07:40.230 --> 00:07:42.170 or something like that, right? Or regularly. 00:07:43.209 --> 00:07:47.370 And we're using all of that in order to hand 00:07:47.370 --> 00:07:52.850 over the statistically possible answers for the 00:07:52.850 --> 00:07:55.069 kid to respond to. So instead of they are chasing 00:07:55.069 --> 00:07:58.089 the words or chasing the answers, we're handing 00:07:58.089 --> 00:08:02.050 them options that are relevant, right, in order 00:08:02.050 --> 00:08:05.410 for them to speed up the conversation in a more 00:08:05.410 --> 00:08:08.329 natural way. I would pause here and just say 00:08:08.329 --> 00:08:13.430 that the kid or the adult who's using, the individual 00:08:13.430 --> 00:08:16.769 who's using the Ma-Talk AI is always the driver. 00:08:16.939 --> 00:08:20.279 So we're not replacing them. They are saying 00:08:20.279 --> 00:08:22.639 what they want to say. The only thing we're doing 00:08:22.639 --> 00:08:26.379 is just hand them over the right options. Yeah. 00:08:26.399 --> 00:08:28.399 Yeah. That was going to be my question. Kinda 00:08:28.399 --> 00:08:32.519 like you said earlier, like I want a banana and 00:08:32.519 --> 00:08:36.799 what if the AI kinda determines that most mornings 00:08:36.799 --> 00:08:39.440 your son wakes up and says, I want cereal for 00:08:39.440 --> 00:08:41.639 breakfast. So that's the thing that it gives 00:08:41.639 --> 00:08:44.120 you. But then what if that preference changes 00:08:44.120 --> 00:08:46.620 and it's using these reinforcement? Learning 00:08:46.620 --> 00:08:49.860 strategies and does it ever does it ever kind 00:08:49.860 --> 00:08:51.899 of hallucinate and give you things that it doesn't 00:08:51.899 --> 00:08:56.080 want? And I guess you kind of answered my question 00:08:56.080 --> 00:08:58.340 with that last disclaimer at the end where the 00:08:58.340 --> 00:09:01.240 the individual is still making that final say 00:09:01.240 --> 00:09:04.799 and what they want. So is is Ma-Talk just giving 00:09:04.799 --> 00:09:08.139 you maybe 5 or 6 different options on the screen 00:09:08.139 --> 00:09:12.080 as opposed to you having to type in every word? 00:09:13.000 --> 00:09:15.669 Absolutely, yes. So we're going more than one 00:09:15.669 --> 00:09:19.149 option. Again, going back to the fact that the 00:09:19.149 --> 00:09:22.830 app, Ma-Talk, is not the kid. The kid is the driver. 00:09:23.190 --> 00:09:27.970 The kid is the pilot. Ma-Talk's job is to learn 00:09:27.970 --> 00:09:31.330 the kid, know his or hers preferences. Going 00:09:31.330 --> 00:09:34.350 back to food is a great example. The kid might 00:09:34.350 --> 00:09:39.009 like banana. My son likes oatmeal. But they might 00:09:39.009 --> 00:09:44.659 don't like eggs. Eggs is not relevant in the context 00:09:44.659 --> 00:09:48.419 of breakfast for that specific individual. So 00:09:48.419 --> 00:09:52.899 instead of that being an option, we took it away 00:09:52.899 --> 00:09:56.940 or lowered it in the statistical options, right? 00:09:57.220 --> 00:10:00.539 And then hand over only five. We are starting 00:10:00.539 --> 00:10:04.200 with five, but it is customizable, right? So 00:10:04.200 --> 00:10:06.639 eventually you'll be able to choose five or 10, 00:10:06.639 --> 00:10:09.980 depends on the... what you want, but the point 00:10:09.980 --> 00:10:12.379 is that you can always refresh and get more, 00:10:12.600 --> 00:10:16.159 right? So it's not just five. If we missed the 00:10:16.159 --> 00:10:19.559 first try, we missed and the first five options 00:10:19.559 --> 00:10:22.179 are not what you want, totally fine. You still 00:10:22.179 --> 00:10:25.519 can refresh it three more times. And so you get 00:10:25.519 --> 00:10:29.980 a total of 20 options, right? So hopefully eventually 00:10:29.980 --> 00:10:34.019 you'll get it. And the point is that we also, 00:10:35.120 --> 00:10:38.370 the app learns the kit. So it learns from different 00:10:38.370 --> 00:10:41.110 situations. So the more you use it, the better 00:10:41.110 --> 00:10:45.049 it becomes. At the beginning, we're not gonna 00:10:45.049 --> 00:10:48.889 be right about what the kid wants, because we 00:10:48.889 --> 00:10:51.169 don't know the kid. We do give the option for 00:10:51.169 --> 00:10:56.450 the parents to insert a few, I would say, who 00:10:56.450 --> 00:10:58.809 the kid is, right? Who Matan is. Matan likes 00:10:58.809 --> 00:11:02.149 oatmeal. This is information that we can, that 00:11:02.149 --> 00:11:06.210 parents can enter. at the beginning on the onboarding. 00:11:06.509 --> 00:11:08.649 And then we have like a baseline of how to start. 00:11:09.750 --> 00:11:13.990 But as we go, the more interactions you have, 00:11:14.149 --> 00:11:17.830 schools, with friends, after school programs, 00:11:17.990 --> 00:11:20.350 all of those, at the pool, right? All of those 00:11:20.350 --> 00:11:24.710 interactions, we learn and then we know how the 00:11:24.710 --> 00:11:26.629 kid's routines, what the kid likes, what the 00:11:26.629 --> 00:11:29.350 kid dislikes, who the kid likes to talk to and 00:11:29.350 --> 00:11:32.929 so on, what they like to talk about. Right. And 00:11:32.929 --> 00:11:36.129 all for the key to have meaningful conversation, 00:11:36.490 --> 00:11:39.710 meaningful interactions for them to be able to 00:11:39.710 --> 00:11:45.470 express themselves. I saw on the demo that there's 00:11:45.470 --> 00:11:49.350 kind of like an avatar or a caricature that is 00:11:49.350 --> 00:11:52.529 involved. Is that a customizable feature that 00:11:52.529 --> 00:11:56.799 is intended to resemble the individual? Yeah. 00:11:56.860 --> 00:11:59.679 So we started with, we start with the, yeah. 00:11:59.799 --> 00:12:02.039 So I love this question. Well, I just, I just 00:12:02.039 --> 00:12:03.679 think that would be just a really cool feature. 00:12:03.919 --> 00:12:06.799 Uh, that just brings a whole nother level of 00:12:06.799 --> 00:12:09.320 personalization versus the traditional lower 00:12:09.320 --> 00:12:13.620 tech AAC communications. Absolutely. So yes, 00:12:13.620 --> 00:12:18.500 absolutely. Cause, uh, AAC needs to be not just 00:12:18.500 --> 00:12:20.919 a tool needs to be a cool tool. Right. Exactly. 00:12:20.919 --> 00:12:24.559 Yeah. All right. So part of that it's, we need 00:12:24.559 --> 00:12:27.649 to give it a character. Right. So first we started 00:12:27.649 --> 00:12:32.090 with, uh, with Verbi. So Verbi is, is Verbali's 00:12:32.090 --> 00:12:35.169 mascot and it's basically a robot dog. That's 00:12:35.169 --> 00:12:38.450 because what it is, it's, Ma-Talk acts as your 00:12:38.450 --> 00:12:41.009 companion and we want to, and you give, you have 00:12:41.009 --> 00:12:45.889 the option to call to, for the partner to remotely 00:12:45.889 --> 00:12:49.070 trigger the listening mode. Right. You don't 00:12:49.070 --> 00:12:51.409 have to click on the button. You can just say, 00:12:51.750 --> 00:12:55.639 Hey, Verbi. And and it starts. So this is this 00:12:55.639 --> 00:12:58.879 is how we start with. We give the Ma-Talk a character, 00:12:58.879 --> 00:13:01.440 right? So we give it a mascot that's Verbi. You 00:13:01.440 --> 00:13:06.179 can see in our logo or on the website. And then 00:13:06.179 --> 00:13:10.159 we're like, OK, we have images and the images 00:13:10.159 --> 00:13:13.139 have kids playing around, running around on those 00:13:13.139 --> 00:13:17.600 images. We need to be basically those are our 00:13:17.600 --> 00:13:20.470 kids heroes, right? How our parents like. our 00:13:20.470 --> 00:13:24.809 heroes, right? So how can we make this more look 00:13:24.809 --> 00:13:32.350 alike to the kid? So at first, this is where 00:13:32.350 --> 00:13:37.509 we are now. We have six options, boy, girl, and 00:13:37.509 --> 00:13:42.009 so on, and that you can customize it based on 00:13:42.009 --> 00:13:46.769 who you are, who your kid is, right? We are looking 00:13:46.769 --> 00:13:51.149 to extend that using And this is where I go always. 00:13:51.409 --> 00:13:53.990 I say AI is so cool. It's so fun. There's so 00:13:53.990 --> 00:13:57.129 much, so many options that you can, you can explore, 00:13:57.330 --> 00:14:01.950 especially now with the latest Google nano banana. 00:14:02.750 --> 00:14:08.269 It's fantastic. So you can, you can, we can take 00:14:08.269 --> 00:14:12.409 those images and really create an avatar for 00:14:12.409 --> 00:14:16.690 the kid. And our plan, it's not now, but our 00:14:16.690 --> 00:14:21.039 plan is for that. avatar to be part of the image 00:14:21.039 --> 00:14:23.480 itself eventually, right? So it will be really, 00:14:23.480 --> 00:14:26.360 really customized. So when the kid sees the scene, 00:14:26.659 --> 00:14:29.580 the image of the scene, they basically will see 00:14:29.580 --> 00:14:32.360 themselves. Yeah. Yeah. It would be really cool 00:14:32.360 --> 00:14:35.080 if you could upload a family photo and then like 00:14:35.080 --> 00:14:37.659 those, those people are the ones that populate 00:14:37.659 --> 00:14:40.139 the app. What is the, what's the foundational 00:14:40.139 --> 00:14:43.139 model that it's all kind of built on? So we're 00:14:43.139 --> 00:14:45.340 using different models. We're not using the specific 00:14:45.340 --> 00:14:49.879 one because, uh, each model is good in different 00:14:49.879 --> 00:14:53.659 ways. So, for example, for images, we're using 00:14:53.659 --> 00:14:58.220 Flux and Google for different types of images. 00:14:59.240 --> 00:15:02.919 For our system, we're currently using OpenAI. 00:15:05.519 --> 00:15:09.460 Google helps with the translations. OpenAI also 00:15:09.460 --> 00:15:15.690 helps with the transcription. So yeah, so each 00:15:15.690 --> 00:15:20.250 model uses different things because AI works 00:15:20.250 --> 00:15:24.309 on the front end, which what you and I will encounter, 00:15:24.509 --> 00:15:26.289 we're using the app, but it also works in the 00:15:26.289 --> 00:15:29.490 back end when we're creating the images, constantly 00:15:29.490 --> 00:15:32.950 and making sure those images are the best image 00:15:32.950 --> 00:15:36.730 that we can provide. So everything, it's an ecosystem 00:15:36.730 --> 00:15:40.659 of assistants and agents that help us. bring 00:15:40.659 --> 00:15:43.259 the best experience possible to the AC space. 00:15:43.820 --> 00:15:46.879 Yeah. What were your main concerns, I guess, 00:15:46.960 --> 00:15:49.980 with the technology or how the technology was 00:15:49.980 --> 00:15:53.620 built? I know there's techno pessimists that 00:15:53.620 --> 00:15:57.379 don't love all the AI tools. Obviously, they 00:15:57.379 --> 00:16:00.039 create a whole suite of opportunities that didn't 00:16:00.039 --> 00:16:02.500 exist even two or three years ago. But have you 00:16:02.500 --> 00:16:06.700 identified any aspects of the AC device that 00:16:06.700 --> 00:16:10.879 are Yeah, maybe concerning to you or that you 00:16:10.879 --> 00:16:16.279 guys are focusing on specifically. I guess just 00:16:16.279 --> 00:16:20.460 the either the AI tool that builds your program 00:16:20.460 --> 00:16:25.440 or your software itself. Like, are there any, 00:16:25.440 --> 00:16:28.379 I guess, deviations that you're kind of focusing 00:16:28.379 --> 00:16:33.889 on? Yeah, yes. So I would say I think the answer 00:16:33.889 --> 00:16:36.990 for this is twofold. I'll start by a short answer 00:16:36.990 --> 00:16:40.970 is that from the perspective of a person and 00:16:40.970 --> 00:16:45.690 advocate for the assistive tech space, I would 00:16:45.690 --> 00:16:49.110 say that I have no concern. Actually, I think 00:16:49.110 --> 00:16:54.870 AI is a blessing for this space. AI is the technology 00:16:54.870 --> 00:17:00.159 that we needed. to help us push this space to 00:17:00.159 --> 00:17:05.940 the moon. I mean, it gives us, the caregivers, 00:17:06.880 --> 00:17:10.500 so much flexibility, and it gives the people 00:17:10.500 --> 00:17:15.859 who need that assistive technology so much independency. 00:17:16.619 --> 00:17:19.900 Because the AI, basically, it's a small brain, 00:17:20.079 --> 00:17:22.460 and sometimes even smarter than us. So I would 00:17:22.460 --> 00:17:28.970 say from the corner of assistive technology space, 00:17:29.250 --> 00:17:33.210 I would say, I think it's a blessing. Going aside 00:17:33.210 --> 00:17:36.930 to the bigger picture, I would say, I don't think 00:17:36.930 --> 00:17:39.430 it concerns the right word. I think challenges 00:17:39.430 --> 00:17:43.750 is, I think for me, it's a better word. And I'll 00:17:43.750 --> 00:17:46.730 explain why. I think the best way that I at least 00:17:46.730 --> 00:17:50.180 look at... AI is that you have like this talented, 00:17:51.740 --> 00:17:54.279 smart, super talented, super intelligent, smart 00:17:54.279 --> 00:17:59.880 and fast assistant, but it's also young and unexperienced, 00:17:59.940 --> 00:18:04.599 right? So what gives, so you can give it a task 00:18:04.599 --> 00:18:06.759 and it will do it. It will do it fast and it 00:18:06.759 --> 00:18:10.460 will do it in scale. But like every young assistant, 00:18:10.779 --> 00:18:13.519 you need supervisor. If you're not supervising 00:18:13.519 --> 00:18:17.859 correctly, uh uh our mistakes will happen right 00:18:17.859 --> 00:18:21.259 it's it's very normal it's expected right and 00:18:21.259 --> 00:18:25.119 so i think what the opportunities on this is 00:18:25.119 --> 00:18:30.099 with ai are are humongous like it's it's we're 00:18:30.099 --> 00:18:32.420 we're not i don't think we're even scraping the 00:18:32.420 --> 00:18:36.680 surface on this and i would say that ai now it's 00:18:36.680 --> 00:18:39.359 the worst that it will ever be it's only going 00:18:39.359 --> 00:18:43.660 to go up right whatever we see now it's the worst 00:18:43.660 --> 00:18:46.799 that it will have ever be. So tomorrow's will 00:18:46.799 --> 00:18:52.160 be better and moving so fast. And the reason 00:18:52.160 --> 00:18:56.779 I'm saying that is because I'll give an example, 00:18:56.960 --> 00:19:00.680 right? So we're creating images and we have image 00:19:00.680 --> 00:19:07.599 bang for all the words that are possible. And 00:19:07.599 --> 00:19:09.519 we're constantly adding and constantly fixing 00:19:09.519 --> 00:19:12.799 it and making it better and better. But at first, 00:19:12.940 --> 00:19:15.259 in our first batch, this is something that you 00:19:15.259 --> 00:19:18.000 need automation and AI. I cannot create this 00:19:18.000 --> 00:19:21.000 menu, right? We're talking about dozens of pictures, 00:19:21.400 --> 00:19:25.980 thousands of pictures. So when we first created 00:19:25.980 --> 00:19:29.180 the first batch, I was so happy with the images. 00:19:29.500 --> 00:19:31.579 And I'm like, wow, this is amazing that it's 00:19:31.579 --> 00:19:34.839 so clear, the scene. I don't need to read what 00:19:34.839 --> 00:19:39.640 it says, and I know what it is. And then my co 00:19:39.640 --> 00:19:45.670 -founder, Told me that his daughter Look at the 00:19:45.670 --> 00:19:47.849 pictures. She noticed something. She knows like 00:19:47.849 --> 00:19:53.490 all the pictures are boys and and and I mean, 00:19:53.509 --> 00:19:55.849 this is interesting because I didn't I didn't 00:19:55.849 --> 00:19:58.869 notice that he didn't notice that her his daughter 00:19:58.869 --> 00:20:03.049 is obviously female noticed that and and it's 00:20:03.049 --> 00:20:08.670 to show that The mistakes are different. It the 00:20:08.670 --> 00:20:12.450 image is still good but because of the scale, 00:20:13.029 --> 00:20:17.829 you might miss something. And so you need an 00:20:17.829 --> 00:20:21.190 AI. So this is where we're using AI now to help 00:20:21.190 --> 00:20:28.109 us QA the outcome of a different set of AI agents, 00:20:28.470 --> 00:20:31.490 right? To make sure that, for example, we have 00:20:31.490 --> 00:20:34.730 equal representation of boys and girls, right? 00:20:34.730 --> 00:20:41.140 That's for example. Yeah. So this is how I see 00:20:41.140 --> 00:20:44.099 it. Yeah. Yeah. The biases. I know when some 00:20:44.099 --> 00:20:46.660 of the initial foundational models came out, 00:20:46.859 --> 00:20:50.359 there were some errors in how it recalled certain 00:20:50.359 --> 00:20:53.140 aspects of history or how it biased different 00:20:53.140 --> 00:20:56.619 populations or genders or races. But like you 00:20:56.619 --> 00:20:58.619 said, it's only getting better every day and 00:20:58.619 --> 00:21:03.180 every week. So how do your son's educators and 00:21:03.180 --> 00:21:09.210 how do you as parents use the app, like I know 00:21:09.210 --> 00:21:12.329 obviously he's responding, but are you inputting 00:21:12.329 --> 00:21:15.190 any information into the software as well? Yeah, 00:21:15.190 --> 00:21:18.869 absolutely. We added. So, uh, maybe it's a good 00:21:18.869 --> 00:21:21.690 time to, uh, to mention my co -founder Lori, 00:21:21.690 --> 00:21:26.589 Lori Azerati. He's amazing. He did an amazing 00:21:26.589 --> 00:21:33.789 job on, on making the app so easy to, to, uh, 00:21:33.789 --> 00:21:39.369 to use. One of the things we just introduced 00:21:39.369 --> 00:21:45.049 was the ability to add niche images. I say niche 00:21:45.049 --> 00:21:47.690 because I think that's the terminology, but for 00:21:47.690 --> 00:21:50.670 us, for everyone, niche could be just a picture 00:21:50.670 --> 00:21:54.769 of you and I as a parent. It's just not a regular 00:21:54.769 --> 00:22:02.619 vocabulary. So we introduced the ability to add 00:22:02.619 --> 00:22:08.079 images and give the aliases for those images. 00:22:08.539 --> 00:22:12.339 So for example, my name is Shay, I'm Matan's 00:22:12.339 --> 00:22:18.140 dad, I'm also his father. And I say in Hebrew, 00:22:18.599 --> 00:22:22.339 that's how he calls me Abba instead of dad. So 00:22:22.339 --> 00:22:25.579 all of those are my aliases that the app needs 00:22:25.579 --> 00:22:30.420 to know when they show my picture, right? Because 00:22:30.420 --> 00:22:33.519 at the end of the day, the app understands the 00:22:33.519 --> 00:22:36.779 conversation and I don't know the answer it will 00:22:36.779 --> 00:22:38.700 bring, right? It will bring something that I 00:22:38.700 --> 00:22:41.019 know it's relevant, but I don't know what for. 00:22:41.680 --> 00:22:48.900 So we provided options for parents to cover this 00:22:48.900 --> 00:22:52.299 aspect of it, to help the app learn more about 00:22:52.519 --> 00:22:57.680 people in the kid's life about their toys, food, 00:22:58.680 --> 00:23:01.819 locations. The GPS location is huge in contact 00:23:01.819 --> 00:23:05.720 signals because the type of conversations that 00:23:05.720 --> 00:23:08.220 we'll have at school are totally different than 00:23:08.220 --> 00:23:10.279 the type of conversation that we'll have at home. 00:23:10.759 --> 00:23:12.319 The type of conversation we'll have in therapy, 00:23:12.640 --> 00:23:15.839 physical therapy, speech therapy, are different 00:23:15.839 --> 00:23:18.099 than the one we'll have at home. So if we want 00:23:18.099 --> 00:23:23.220 to have relevant conversational answers, responses, 00:23:23.220 --> 00:23:28.339 we need to take GPS into consideration. And those 00:23:28.339 --> 00:23:30.660 are also part of the things that you can add 00:23:30.660 --> 00:23:33.599 to the app and customize. So are you pulling, 00:23:33.900 --> 00:23:36.299 because earlier you mentioned weather, holidays, 00:23:36.480 --> 00:23:40.180 are you integrating with other aspects of like 00:23:40.180 --> 00:23:44.059 an iOS, like an Apple operating system to pull 00:23:44.059 --> 00:23:46.380 that information? Where is that coming from? 00:23:47.200 --> 00:23:52.359 Yeah. So some of them are from public APIs. that 00:23:52.359 --> 00:23:56.920 you can use, whether it's the public API. Some 00:23:56.920 --> 00:24:01.779 of them are just, you know, time, date, those 00:24:01.779 --> 00:24:04.599 are coming from your phone. It's not something 00:24:04.599 --> 00:24:09.880 that you need to... Your phone has GPS, obviously, 00:24:10.059 --> 00:24:16.400 as well. But the fact, GPS is different from 00:24:16.400 --> 00:24:18.720 the place that you are. GPS is just a coordinate, 00:24:18.900 --> 00:24:21.740 right, but you need to... Your corners can be 00:24:21.740 --> 00:24:23.839 Starbucks, right? So you need to know that this 00:24:23.839 --> 00:24:26.240 coordinate is Starbucks. This comes from another 00:24:26.240 --> 00:24:32.160 API that we're using. What else? We're talking 00:24:32.160 --> 00:24:36.359 about weather, speed, for example, velocity, 00:24:36.480 --> 00:24:42.500 right? We're planning on integrating a velocity 00:24:42.500 --> 00:24:46.700 sensor that's built in your phone. We just need 00:24:46.700 --> 00:24:49.960 to take the data from it because when you're 00:24:49.799 --> 00:24:53.380 When you're driving, at least my kids, they constantly 00:24:53.380 --> 00:24:59.400 ask me to DJ for them. I want to give Matan the 00:24:59.400 --> 00:25:04.200 capability to ask his songs. In order for that, 00:25:05.019 --> 00:25:07.240 the app needs to know that they're in a car. 00:25:08.400 --> 00:25:11.220 One of the ways to do it is to understand that 00:25:11.220 --> 00:25:15.500 we're driving, the velocity. Yeah, so that's 00:25:15.500 --> 00:25:17.960 one of the sensors that we're using and more 00:25:17.960 --> 00:25:20.900 we constantly looking for better ways to collect 00:25:20.900 --> 00:25:25.359 context contextual signals that will help us 00:25:25.359 --> 00:25:30.099 Just make the experience Better and not better 00:25:30.099 --> 00:25:32.200 just for the kid. I think this is where where 00:25:32.200 --> 00:25:37.019 it's important for to mention is When we started 00:25:37.019 --> 00:25:42.019 building my token, I we looked on three user 00:25:42.019 --> 00:25:49.599 personas that can benefit from the app. The first 00:25:49.599 --> 00:25:53.579 was obviously the kid who's using it. The second 00:25:53.579 --> 00:25:57.799 is the caregiver or the conversational partner. 00:25:59.759 --> 00:26:01.900 Because the caregiver would want to give them 00:26:01.900 --> 00:26:04.599 a return on investment to be as high as possible. 00:26:04.960 --> 00:26:06.920 The investment necessarily needs to be a capital 00:26:06.920 --> 00:26:09.079 investment. It could be an energy. Today, you 00:26:09.079 --> 00:26:11.640 need constantly to customize a different app. 00:26:11.630 --> 00:26:15.509 Different AAC devices used to add pictures, remove 00:26:15.509 --> 00:26:20.309 pictures, organize it. So, but at the end of 00:26:20.309 --> 00:26:23.109 the day, the AAC becomes an I want machine and 00:26:23.109 --> 00:26:25.450 that's not what we wanted, right? So the return 00:26:25.450 --> 00:26:28.029 on investment is very low. What we wanted is 00:26:28.029 --> 00:26:33.089 for the parent to have as much of flexibility 00:26:33.089 --> 00:26:36.769 or freedom as possible. You don't need to customize 00:26:36.769 --> 00:26:39.420 the app because the app... customize itself, 00:26:39.839 --> 00:26:42.720 right? Your job is just to enjoy the conversation, 00:26:43.259 --> 00:26:47.440 right? So that's the second user persona. It 00:26:47.440 --> 00:26:51.539 goes also to the conversational partner. We don't 00:26:51.539 --> 00:26:53.700 want it to be boring for the conversational partner, 00:26:53.900 --> 00:26:57.400 right? So that's part of it. And they want the 00:26:57.400 --> 00:27:01.480 conversation to be engaging. And third, the third 00:27:01.480 --> 00:27:07.059 user persona is the speech therapist. We want 00:27:07.059 --> 00:27:13.140 to give them the tools to create the best therapy, 00:27:13.359 --> 00:27:15.700 speech therapy they can for that specific kid, 00:27:16.000 --> 00:27:18.460 right? Today, from my experience and talking 00:27:18.460 --> 00:27:21.579 to other speech therapists, usually they have 00:27:21.579 --> 00:27:25.960 once a week, they have a speech session with 00:27:25.960 --> 00:27:29.599 the kid. And it's amazing. They're doing an amazing 00:27:29.599 --> 00:27:34.039 job, right? When i started i was like is one 00:27:34.039 --> 00:27:36.400 data point of that meeting a week is that enough 00:27:36.400 --> 00:27:39.660 can we create more right what if we have 24 7 00:27:39.660 --> 00:27:43.160 of those data right can we make a better speech 00:27:43.160 --> 00:27:47.359 term and that's how we approach with them okay 00:27:47.359 --> 00:27:52.119 we are the data that we're collecting is always 00:27:52.119 --> 00:27:56.869 going back to the kid. In different ways so if 00:27:56.869 --> 00:27:58.750 we're collecting data to help customizing it 00:27:58.750 --> 00:28:01.069 that's one way and then we're collecting data 00:28:01.069 --> 00:28:06.849 to help them. Analyze it this is speech capabilities 00:28:06.849 --> 00:28:10.609 for the speech there. So now they will be able 00:28:10.609 --> 00:28:13.970 to have like a dashboard that will tell them. 00:28:14.750 --> 00:28:16.710 This is where this is the baseline is when we 00:28:16.710 --> 00:28:19.650 started two weeks from now look is like now clicking 00:28:19.650 --> 00:28:24.349 really fast on the. on the images, two weeks 00:28:24.349 --> 00:28:29.690 afterwards, his switching is more engaging. Now 00:28:29.690 --> 00:28:32.250 we'll be able to see a benchmark, you'll be able 00:28:32.250 --> 00:28:35.769 to see graphs or charts of the progress, something 00:28:35.769 --> 00:28:39.250 that they, unfortunately they don't have it today. 00:28:40.670 --> 00:28:43.210 Yeah, no, that's a really interesting use case. 00:28:43.490 --> 00:28:46.269 I guess I hadn't thought of it from the SLP standpoint. 00:28:47.089 --> 00:28:50.029 So is there, if there's different user personas, 00:28:50.230 --> 00:28:53.579 is there different pricing models or is it the 00:28:53.579 --> 00:28:56.619 individual has a subscription and then They would 00:28:56.619 --> 00:28:59.500 just bring it to their their speech therapy and 00:28:59.500 --> 00:29:03.140 use it as a tool Yeah, so that's that's that's 00:29:03.140 --> 00:29:05.940 what I love this question. You know, I'll tell 00:29:05.940 --> 00:29:10.640 a story about when I went to school University 00:29:10.640 --> 00:29:12.720 my professor told me there are three types of 00:29:12.720 --> 00:29:15.140 questions. There are good questions. There are 00:29:15.140 --> 00:29:17.099 great questions. There are interesting questions 00:29:17.660 --> 00:29:20.720 So good questions are the ones that have obvious 00:29:20.720 --> 00:29:23.660 answers. Great questions are the ones that are 00:29:23.660 --> 00:29:25.420 not obvious questions, but the professor has 00:29:25.420 --> 00:29:28.960 the answer. An interesting question is that a 00:29:28.960 --> 00:29:32.759 question that the professor don't have. So this 00:29:32.759 --> 00:29:36.740 is a great question. So yeah, what we looked 00:29:36.740 --> 00:29:42.019 at it that our vision for Ma-Talk AI is for it 00:29:42.019 --> 00:29:47.539 to be a companion. You don't need to think. In 00:29:47.539 --> 00:29:51.099 order to talk, in order to speak, that's going 00:29:51.099 --> 00:29:53.920 back to natural. That's what we want. In order 00:29:53.920 --> 00:30:00.900 for that to happen, we didn't create different 00:30:00.900 --> 00:30:05.500 payment plans for speech therapists or for families. 00:30:06.079 --> 00:30:11.049 The family is the core user. And they have it's 00:30:11.049 --> 00:30:13.029 not now. This is a feature that will introduce 00:30:13.029 --> 00:30:15.049 in the future But this is where we're going with 00:30:15.049 --> 00:30:17.710 it is that you have a profile the family will 00:30:17.710 --> 00:30:20.630 have a profile and under that profile You can 00:30:20.630 --> 00:30:23.750 call it the the kids team right you have the 00:30:23.750 --> 00:30:26.210 parents and then the parents will invite the 00:30:26.210 --> 00:30:29.410 speech therapist Right to because they need to 00:30:29.410 --> 00:30:31.990 share the data, right? They need to share the 00:30:31.990 --> 00:30:35.549 data and they need to benefit from from that 00:30:35.549 --> 00:30:39.000 data. So the speech therapist won't But in short 00:30:39.000 --> 00:30:41.839 answer, the speech therapist will not need to 00:30:41.839 --> 00:30:45.099 pay for it because their user will be invited 00:30:45.099 --> 00:30:51.200 by the parents. Yeah. I would imagine some of 00:30:51.200 --> 00:30:55.079 this vision for the product as a whole has evolved 00:30:55.079 --> 00:31:00.559 over time. So maybe you found a problem in your 00:31:00.559 --> 00:31:04.240 son not having clear fluid communication. You 00:31:04.240 --> 00:31:09.109 built it with him in mind. Did you have this 00:31:09.109 --> 00:31:13.130 like speech therapy angle in mind when you first 00:31:13.130 --> 00:31:16.710 started? How has, I guess how has it kind of 00:31:16.710 --> 00:31:19.769 evolved from idea conception into a product? 00:31:20.809 --> 00:31:27.450 Yeah. Yeah. So, so I told you how I started thinking 00:31:27.450 --> 00:31:30.609 about it, but I think, I think the bad, the what. 00:31:31.509 --> 00:31:34.829 Changed the way for me thinking about it to action 00:31:34.829 --> 00:31:37.130 to taking action on it happened about a year 00:31:37.130 --> 00:31:43.829 ago and so like we went to his IP at school IP 00:31:43.829 --> 00:31:46.829 who further Someone doesn't know it's basically 00:31:46.829 --> 00:31:49.289 it's a thing about as a meeting did you have 00:31:49.289 --> 00:31:53.349 between you and the schools? professionals about 00:31:53.349 --> 00:31:58.789 the sons your son's Special activity at school 00:31:59.019 --> 00:32:01.619 You agree together what you want to achieve in 00:32:01.619 --> 00:32:07.099 the coming period. And we were in that IEP and 00:32:07.099 --> 00:32:10.660 I'll start by saying I love his school and I 00:32:10.660 --> 00:32:17.160 love his team. We got really lucky and they love 00:32:17.160 --> 00:32:20.819 him. So then we got even luckier. But that specific 00:32:20.819 --> 00:32:25.769 meeting was evolving around AAC. his AACs, his 00:32:25.769 --> 00:32:28.170 tool. And I was like, and it was an hour and 00:32:28.170 --> 00:32:30.029 a half. And me and my wife came back and I was 00:32:30.029 --> 00:32:32.529 frustrated. I was like, why are we talking so 00:32:32.529 --> 00:32:35.069 much about something that needs to be obvious? 00:32:35.170 --> 00:32:37.490 We're not thinking about us talking. It should 00:32:37.490 --> 00:32:40.309 be effortless. Why are we putting so much effort 00:32:40.309 --> 00:32:45.930 on it? So that was, that changed the way I approach. 00:32:46.130 --> 00:32:49.710 I was like, okay, let's try. I'll try doing it 00:32:49.710 --> 00:32:52.269 myself, right? Like I had some time and I was 00:32:52.269 --> 00:32:56.640 like, okay, so. I started, I talked to a few 00:32:56.640 --> 00:32:59.940 speech therapists, I described my vision, and 00:32:59.940 --> 00:33:04.519 I noticed that there is a, they love the idea, 00:33:04.599 --> 00:33:08.200 but some of my ideas are, it's harder for them 00:33:08.200 --> 00:33:13.039 to imagine. So I was like, okay, I'm going to 00:33:13.039 --> 00:33:16.799 build a prototype. So I had a Kindle at home, 00:33:17.099 --> 00:33:21.160 a Kindle Fire at home was, we didn't use it for 00:33:21.160 --> 00:33:23.630 a while. So I was like, okay, I'm going to. Use 00:33:23.630 --> 00:33:27.309 AI. It was before vibe coding. It was before 00:33:27.309 --> 00:33:30.650 that. So I took ChatGPT and me and ChatGPT became 00:33:30.650 --> 00:33:33.829 really, really good friends. And back and forth. 00:33:34.150 --> 00:33:36.210 And eventually I was able to build a canap. And 00:33:36.210 --> 00:33:38.430 my background is, is an engineering, but I'm 00:33:38.430 --> 00:33:41.410 a product manager. So it's, it's been a, and 00:33:41.410 --> 00:33:43.490 I'm an electrical engineer. So I'm not a programmer. 00:33:44.130 --> 00:33:46.269 So I, everything I was able to go back and forth 00:33:46.269 --> 00:33:53.769 with ChatGPT and I created this, this app. The 00:33:53.769 --> 00:33:56.529 only thing the only thing it could do was to 00:33:56.529 --> 00:34:03.349 answer questions listen and and And it listened 00:34:03.349 --> 00:34:06.089 understood the question and then came up with 00:34:06.089 --> 00:34:10.429 those Answers and took it to to my the speech 00:34:10.429 --> 00:34:12.130 therapist that I know and I was like, okay, what 00:34:12.130 --> 00:34:14.329 do you think about that? And they're like, wow, 00:34:14.349 --> 00:34:17.230 this is this amazing my blog blowing like you 00:34:17.230 --> 00:34:20.070 have to do it. You have to do it I was like, 00:34:20.170 --> 00:34:23.269 okay great And then I met my co -founder Lori, 00:34:23.329 --> 00:34:28.730 which I am so lucky to meet him. And together 00:34:28.730 --> 00:34:31.889 we started designing. So we created like, cause 00:34:31.889 --> 00:34:35.650 we approached the entire conversation a little 00:34:35.650 --> 00:34:41.349 bit different. We break it into events. We, you 00:34:41.349 --> 00:34:43.869 have the event of you responding to a question, 00:34:43.929 --> 00:34:46.250 which I talked a lot about, and then you have 00:34:46.250 --> 00:34:49.170 the event of you initiated the conversation. 00:34:49.840 --> 00:34:52.059 And then you have event of you talking about 00:34:52.059 --> 00:34:55.119 your feelings and then you have the event of 00:34:55.119 --> 00:34:58.800 you of shortcuts Of from like you need to go 00:34:58.800 --> 00:35:02.539 to the restroom for example, right? So we broke 00:35:02.539 --> 00:35:05.500 those events and we're like, okay each event 00:35:05.500 --> 00:35:08.219 will have a different user experience because 00:35:08.219 --> 00:35:11.320 it requires different flows different context 00:35:11.320 --> 00:35:18.250 and different AI's so So for example, for initiating 00:35:18.250 --> 00:35:21.590 conversation, I was like, okay, go back to the 00:35:21.590 --> 00:35:26.289 real estate on the screen. You want it to be 00:35:26.289 --> 00:35:30.130 easy to be used for kids with special needs. 00:35:30.730 --> 00:35:33.949 And the interesting thing about special needs 00:35:33.949 --> 00:35:36.989 is that they're special. You cannot predict. 00:35:37.030 --> 00:35:39.849 There's no standard here, right? So you need 00:35:39.849 --> 00:35:42.949 unique solutions. Right. So what is the best 00:35:42.949 --> 00:35:45.030 way to create those things? You need to start 00:35:45.030 --> 00:35:46.809 thinking creatively. And I noticed that with 00:35:46.809 --> 00:35:49.730 my son, one of the things is that when he goes 00:35:49.730 --> 00:35:53.250 to use all the AC, he tries to click on buttons 00:35:53.250 --> 00:35:55.849 and he has a happy finger. It's like he finger 00:35:55.849 --> 00:35:58.769 dances. And so he clicks on the wrong button. 00:35:59.010 --> 00:36:03.670 And I was like, okay, how can I, what can I create 00:36:03.670 --> 00:36:09.539 that will, I would say fight that. Or protect 00:36:09.539 --> 00:36:12.340 from that behavior because it's like, if you're, 00:36:12.440 --> 00:36:14.619 let's say talk about feelings. If you're, if 00:36:14.619 --> 00:36:17.400 you have a button says mad and the button say 00:36:17.400 --> 00:36:19.559 happy, and you want to say click on happy, but 00:36:19.559 --> 00:36:21.699 click on mad, it changes entire conversation. 00:36:23.099 --> 00:36:26.280 Right. So I was okay. Initial conversation. I 00:36:26.280 --> 00:36:31.159 noticed about him that when he gets my phone, 00:36:31.760 --> 00:36:37.409 he loves my phone. He's like a hacker. He uses 00:36:37.409 --> 00:36:40.429 Netflix, he uses Spotify, and it's like he swipes 00:36:40.429 --> 00:36:42.250 and he scrolls and it's like, oh my God, you 00:36:42.250 --> 00:36:45.510 can do all of this, but you struggle with AAC. 00:36:45.650 --> 00:36:47.909 And I was like, oh, maybe that's because we don't 00:36:47.909 --> 00:36:52.809 have scrolls or swipes gestures in our devices 00:36:52.809 --> 00:36:56.690 today. So I introduced that with our initiated 00:36:56.690 --> 00:37:02.030 conversation flow. And there, instead of us The 00:37:02.030 --> 00:37:07.289 kid basically has three columns of them, of us 00:37:07.289 --> 00:37:12.010 predicting, but the kid helps us by pointing 00:37:12.010 --> 00:37:14.469 us to the right direction, right? So instead 00:37:14.469 --> 00:37:17.769 of us trying to predict the next word, what we're 00:37:17.769 --> 00:37:21.090 trying to predict is the topic. Meaning that 00:37:21.090 --> 00:37:24.010 if you're in the morning and you're at the kitchen 00:37:24.010 --> 00:37:26.130 and it's middle of the week, there is a good 00:37:26.130 --> 00:37:27.650 chance you're going to talk about breakfast, 00:37:28.349 --> 00:37:31.449 right? Or maybe the bus comes. or something like 00:37:31.449 --> 00:37:33.210 that. That's a topic that we can talk about. 00:37:33.369 --> 00:37:36.769 The kid will scroll to find the right topic to 00:37:36.769 --> 00:37:40.829 talk about. And then you choose the right... 00:37:40.829 --> 00:37:43.510 he chooses the action that he wants to take. 00:37:43.829 --> 00:37:49.110 And then he chooses the objects that he wants 00:37:49.110 --> 00:37:52.849 to take the action on. And then it clicks play. 00:37:52.889 --> 00:37:55.650 And what happens is that the AI takes those building 00:37:55.650 --> 00:37:59.650 blocks and creates a natural sentence. for you, 00:37:59.650 --> 00:38:03.210 for you and I to understand easily, more naturally. 00:38:04.010 --> 00:38:07.670 So that's one of the approaches that I did. And 00:38:07.670 --> 00:38:11.809 it was new. So I took it to my, again, my speech 00:38:11.809 --> 00:38:14.349 therapist friends and I showed them and I went 00:38:14.349 --> 00:38:19.909 to the JCC. They have a JCC here in this area 00:38:19.909 --> 00:38:23.250 has an amazing inclusive program since the seventh 00:38:23.250 --> 00:38:30.030 year. Amazing. Heads off on that. And so I went 00:38:30.030 --> 00:38:32.090 there and I talked to the director and I was 00:38:32.090 --> 00:38:33.889 like, OK, what do you think about this? This 00:38:33.889 --> 00:38:36.730 is what I did. I showed them the design for because 00:38:36.730 --> 00:38:39.630 it's a new UX design that I did and I needed 00:38:39.630 --> 00:38:43.210 them to give me feedback. So they look at it 00:38:43.210 --> 00:38:45.030 and I was like, well, I like it. We love it. 00:38:45.030 --> 00:38:46.849 I would do this. I would do that. So it was a 00:38:46.849 --> 00:38:49.469 lot of back and forth talking to speech therapists 00:38:49.469 --> 00:38:55.130 and people from from the industry to get the 00:38:55.130 --> 00:38:59.920 right to get the right would say directions, 00:39:00.219 --> 00:39:03.980 tweaks in order to, so you have a good idea, 00:39:04.440 --> 00:39:06.719 but it's a good idea is not enough because good 00:39:06.719 --> 00:39:09.940 idea is good for you. You need to be good for 00:39:09.940 --> 00:39:13.000 everyone else. So you need feedback. So I, and 00:39:13.000 --> 00:39:15.159 so you need, in order to get the right feedback, 00:39:15.179 --> 00:39:18.659 you need to help them imagine where you want 00:39:18.659 --> 00:39:21.300 to go. So that's where the prototype came to 00:39:21.300 --> 00:39:25.239 place, kept to be. You're able to solicit feedback 00:39:25.239 --> 00:39:28.639 from speech and language pathologists and parents, 00:39:28.880 --> 00:39:33.099 have you been able to assess how the user enjoys 00:39:33.099 --> 00:39:41.099 using the app? Yes. So we just launched. So before 00:39:41.099 --> 00:39:47.360 the launch, speech therapists are using it in 00:39:47.360 --> 00:39:53.639 their clinics and they're loving it. I mean, 00:39:53.880 --> 00:39:55.619 it's a little bit different for something, but 00:39:55.619 --> 00:39:58.300 it's different by definition. So it's supposed 00:39:58.300 --> 00:40:01.599 to be different. So, uh, it's like switching 00:40:01.599 --> 00:40:04.880 between windows to Mac, right? Like it's, it's 00:40:04.880 --> 00:40:07.679 not, it's different, but it's by definition, 00:40:07.679 --> 00:40:11.579 right? Uh, so, uh, so that's how we're approaching 00:40:11.579 --> 00:40:15.179 it. We got really good feedback. Uh, people say, 00:40:15.320 --> 00:40:18.619 I mean, one of the feedback I really like is 00:40:18.619 --> 00:40:21.550 that people love the images. And that's something 00:40:21.550 --> 00:40:24.610 I took, it's a different approach. Today, if 00:40:24.610 --> 00:40:27.409 you use the AC before, you notice that there 00:40:27.409 --> 00:40:31.789 is a very specific style for the images on the 00:40:31.789 --> 00:40:38.809 current AC in the market. And one of the feedback, 00:40:39.289 --> 00:40:42.489 one of the projects I took was, okay, I want 00:40:42.489 --> 00:40:47.530 this image to be universally understood. I don't 00:40:47.530 --> 00:40:50.329 want to create another language of images for 00:40:50.329 --> 00:40:52.909 the kid to learn. I want an image that doesn't 00:40:52.909 --> 00:40:55.289 matter if you're in the US, you're in South America, 00:40:56.090 --> 00:41:00.469 Australia, Asia, whatever. A kid playing soccer 00:41:00.469 --> 00:41:04.650 is a kid playing soccer, right? A kid eating 00:41:04.650 --> 00:41:06.730 a salad is a kid eating a salad, right? Or something 00:41:06.730 --> 00:41:14.369 like that. So I created an image that describes 00:41:14.369 --> 00:41:18.179 a scene. Instead of a static, it is static, but 00:41:18.179 --> 00:41:23.860 when you look at it, you see an action, a kid 00:41:23.860 --> 00:41:28.079 taking an action, right? And it was designed 00:41:28.079 --> 00:41:31.960 in a very friendly, easy to understand way. And 00:41:31.960 --> 00:41:34.360 again, I took a, you know, I took a little fade 00:41:34.360 --> 00:41:39.099 there, creating this. And I got a lot of amazing 00:41:39.099 --> 00:41:41.619 feedback about that people loves the images. 00:41:42.380 --> 00:41:45.320 The and you know this is this is great because 00:41:45.320 --> 00:41:47.820 now we're we're creating more images and now 00:41:47.820 --> 00:41:51.039 i know this is what they want and one of the 00:41:51.039 --> 00:41:55.179 feedback that we got was like that images are 00:41:55.179 --> 00:42:00.000 great but some kids my not like background to 00:42:00.000 --> 00:42:03.699 be a would like to see the background white. 00:42:04.780 --> 00:42:08.239 Because it's sensory wise it's too it's too busy. 00:42:09.309 --> 00:42:12.849 And so it gives us the food for thought for me 00:42:12.849 --> 00:42:16.530 and Lori, my co -founder. Like, how can we make 00:42:16.530 --> 00:42:23.550 our Ma-Talk AI even more customizable? So we are 00:42:23.550 --> 00:42:26.909 working on ways to increase the setting page. 00:42:27.610 --> 00:42:29.449 So parents, and this is again, going back to 00:42:29.449 --> 00:42:33.489 AI, AI gives you so much flexibility. So parents 00:42:33.489 --> 00:42:37.530 who knows, they know their kid the best. So they'll 00:42:37.530 --> 00:42:41.250 be able to know, okay, my kid understand the 00:42:41.250 --> 00:42:43.829 image of a kid playing soccer, for example, right? 00:42:44.030 --> 00:42:48.349 But the background of that image of an other 00:42:48.349 --> 00:42:50.309 kid in the background playing, it's too busy 00:42:50.309 --> 00:42:54.690 for it. So I would like to remove it. We'll eventually 00:42:54.690 --> 00:42:59.869 give the option for that parent to just remove 00:42:59.869 --> 00:43:02.250 it in a swipe of a button. And not just for a 00:43:02.250 --> 00:43:04.190 specific image. It will be obviously for the 00:43:04.190 --> 00:43:10.730 entire database. So the image will be even better 00:43:10.730 --> 00:43:15.670 for their kids, right? Because it's unique. Yeah, 00:43:15.730 --> 00:43:19.050 absolutely. Yeah, there's so much diversity, 00:43:19.090 --> 00:43:22.230 obviously, within the diagnoses that use AAC 00:43:22.230 --> 00:43:25.789 that it has to be individualized and user -centric. 00:43:26.530 --> 00:43:30.349 So AI is really the only way to do that in a 00:43:30.349 --> 00:43:36.840 prompt and scalable manner. Yeah. I work predominantly 00:43:36.840 --> 00:43:41.219 in fitness and physical activity and I kind of 00:43:41.219 --> 00:43:45.360 gave you a heads up beforehand to maybe see if 00:43:45.360 --> 00:43:48.579 we could talk about how verbally could be used 00:43:48.579 --> 00:43:52.280 in something like a gym setting. So your son 00:43:52.280 --> 00:43:56.900 comes to my gym, how can he and I communicate 00:43:56.900 --> 00:43:58.900 in a way that makes that experience more enjoyable 00:43:58.900 --> 00:44:05.079 for him? Yeah, so I Yeah, some things that I, 00:44:05.179 --> 00:44:07.699 so one other thing I noticed with the ASC today 00:44:07.699 --> 00:44:11.860 is that they're bulky. They are usually come 00:44:11.860 --> 00:44:14.980 in the form of a tablet. And it's great if you're 00:44:14.980 --> 00:44:18.960 in a school environment, but what happens if 00:44:18.960 --> 00:44:20.539 you walk in your hallway? What if it happened 00:44:20.539 --> 00:44:23.980 if you're in the playground or at the gym in 00:44:23.980 --> 00:44:26.940 this case, right? You cannot use a tablet. And 00:44:26.940 --> 00:44:29.239 if you're using a phone, it's maybe good for 00:44:29.239 --> 00:44:32.519 the hallway, but it's not good for the gym or 00:44:32.519 --> 00:44:34.260 the playground because it's not safe because 00:44:34.260 --> 00:44:38.079 of the string around your neck, right? So this 00:44:38.079 --> 00:44:41.340 is where Verbali comes to play because our vision 00:44:41.340 --> 00:44:44.000 is to make it, again, I'm going back, I know 00:44:44.000 --> 00:44:45.800 I mentioned it a lot, is going back to natural. 00:44:45.940 --> 00:44:48.099 The same way that you and I speak, I want it 00:44:48.099 --> 00:44:50.880 to be the same way for them, for everyone who 00:44:50.880 --> 00:44:54.610 needs AAC. And the way to do it is not just to 00:44:54.610 --> 00:44:57.449 introduce a tablet, an app for the tablet, or 00:44:57.449 --> 00:45:00.110 an app for your phone, but it's also introduced 00:45:00.110 --> 00:45:05.230 an app for your watch, right? Because the conversations 00:45:05.230 --> 00:45:07.210 still happen, but they might be a little bit 00:45:07.210 --> 00:45:12.530 different. It might be not as rich as your school 00:45:12.530 --> 00:45:14.150 environment, right? Because you're in the playground, 00:45:14.309 --> 00:45:16.210 you're going to write on this, I'm going to write 00:45:16.210 --> 00:45:18.010 there, I'm going to write there. You still want 00:45:18.010 --> 00:45:20.210 to have a conversation. You'll need a different 00:45:20.210 --> 00:45:23.750 UI and UX for it. And the hardware for this, 00:45:23.969 --> 00:45:27.050 the same way Verbal is envisioning it, is an 00:45:27.050 --> 00:45:31.809 AAC on your watch. Today we have amazing technology 00:45:31.809 --> 00:45:35.849 that Apple, Samsung, Google all provide with 00:45:35.849 --> 00:45:40.489 their, in German, provide with their watches, 00:45:40.769 --> 00:45:44.849 smartwatches. And the sensors are there. The 00:45:44.849 --> 00:45:51.010 connection to Wi -Fi or 5G is there. And so it 00:45:51.010 --> 00:45:55.030 means that the only gap is for verbally to create 00:45:55.030 --> 00:45:59.730 the right UX for the kids to be able to utilize 00:45:59.730 --> 00:46:03.769 this watch to communicate in situations, in scenarios 00:46:03.769 --> 00:46:09.050 that they're on a kayak or at the gym, or the 00:46:09.050 --> 00:46:12.530 playground, or place it, right? So yeah, that's 00:46:12.530 --> 00:46:15.429 where we're looking at it. We're actively looking 00:46:15.429 --> 00:46:18.789 at it and designing the best UX for it. Yeah, 00:46:18.929 --> 00:46:20.949 yeah, that's a really interesting use case. Is 00:46:20.949 --> 00:46:23.329 there any way that it could be used to build 00:46:23.329 --> 00:46:26.809 a social story? So if your son's coming to my 00:46:26.809 --> 00:46:31.070 gym for the first time, can you lay any groundwork 00:46:31.070 --> 00:46:33.190 for what that experience might look like to him 00:46:33.190 --> 00:46:35.230 to make him more comfortable? Or is that not 00:46:35.230 --> 00:46:39.869 really the technology? I can you I'm not sure. 00:46:40.590 --> 00:46:42.429 Yeah, I guess. Are you familiar with what a social 00:46:42.429 --> 00:46:47.090 story is? No. So like for some kids maybe with 00:46:47.090 --> 00:46:49.789 autism like before they enter a new environment, 00:46:50.050 --> 00:46:54.150 it's a series of descriptions of what they might 00:46:54.150 --> 00:46:56.210 encounter. So you're going to walk through the 00:46:56.210 --> 00:46:58.250 door, you're going to meet Brendan, you're going 00:46:58.250 --> 00:47:01.190 to put your water bottle on the shelf, you're 00:47:01.190 --> 00:47:03.510 going to grab your program, and it's to alleviate 00:47:03.510 --> 00:47:06.070 some of the concerns or the anxiety that might 00:47:06.070 --> 00:47:08.730 come with new environments. So I guess I was 00:47:08.730 --> 00:47:11.710 just curious whether maybe it's just a conversation 00:47:11.710 --> 00:47:15.190 that you and he have beforehand as opposed to 00:47:15.190 --> 00:47:18.820 a laid out sequential pictorial schedule type 00:47:18.820 --> 00:47:23.360 of thing. So you were talking, I was thinking 00:47:23.360 --> 00:47:25.059 about something specifically that might be really 00:47:25.059 --> 00:47:27.780 helpful. So I'll be honest, I haven't thought 00:47:27.780 --> 00:47:30.039 about it. Obviously, I wasn't aware of what the 00:47:30.039 --> 00:47:33.280 social story is. But you were talking and I was 00:47:33.280 --> 00:47:36.400 like, oh, again, AI can be a very good tool in 00:47:36.400 --> 00:47:39.260 this. Because Google, for example, really just 00:47:39.260 --> 00:47:43.800 introduced stories, bedtime stories feature. 00:47:44.039 --> 00:47:46.460 with them. So basically you give them a, an idea 00:47:46.460 --> 00:47:51.639 for a story and the output is like, as you would 00:47:51.639 --> 00:47:54.239 think bed time story is like, it looks like a 00:47:54.239 --> 00:47:56.380 book, just a digital, but it looks like a book 00:47:56.380 --> 00:48:00.519 has like 10 or 15 pages and with images and everything. 00:48:00.960 --> 00:48:04.219 And so you can do the same thing with going to 00:48:04.219 --> 00:48:06.599 your gym. Like you pretty much going to say, 00:48:06.820 --> 00:48:08.840 okay, you're going to tell the AI, this is something 00:48:08.840 --> 00:48:14.719 we can adopt and incorporate. In verbally animal 00:48:14.719 --> 00:48:18.360 and I'm gonna say you're gonna this you got quickly 00:48:18.360 --> 00:48:21.119 described to on the fly I'm gonna quickly disrupt 00:48:21.119 --> 00:48:23.920 to my talk What you're gonna encounter go to 00:48:23.920 --> 00:48:26.300 a we're going to Brenda's gym. We're gonna need 00:48:26.300 --> 00:48:30.280 to put the water bottle on the countertop and 00:48:30.280 --> 00:48:34.920 We're gonna do One two three Exercises, whatever 00:48:34.920 --> 00:48:37.139 right you can tell that quickly today. I have 00:48:37.139 --> 00:48:40.719 will immediately will come up with a story like 00:48:40.719 --> 00:48:47.239 images book that You can see Matan as the kid 00:48:47.239 --> 00:48:50.320 in that image. And you can do that and you can 00:48:50.320 --> 00:48:53.800 do it. Yeah. Yeah. So you can do it. Yeah. Yeah. 00:48:53.980 --> 00:48:57.869 That's another. effective use case I guess of 00:48:57.869 --> 00:49:00.349 something like this that can be individual centric. 00:49:00.769 --> 00:49:04.510 Are you able to integrate to like eye gaze technology? 00:49:04.670 --> 00:49:08.050 I guess I'm thinking of someone who's non -speaking 00:49:08.050 --> 00:49:10.349 maybe because they have a spinal cord injury 00:49:10.349 --> 00:49:14.530 or a TBI. Is there a way for them to use a device 00:49:14.530 --> 00:49:18.389 through eye gaze as opposed to like the dexterity 00:49:18.389 --> 00:49:23.429 of a finger? Yeah. So not today, but definitely 00:49:23.429 --> 00:49:26.429 in the future, because Verbali's mission is to 00:49:26.429 --> 00:49:29.969 help anyone who needs to create a meaningful 00:49:29.969 --> 00:49:33.929 conversation. And so today we're focusing on 00:49:33.929 --> 00:49:37.570 kid who can master their hands capabilities. 00:49:39.150 --> 00:49:41.769 But eye gazing is definitely on the roadmap. 00:49:42.170 --> 00:49:45.699 We are experimenting with what built -in eye 00:49:45.699 --> 00:49:50.800 -gazing. So your Mac, your iPhone already has 00:49:50.800 --> 00:49:54.400 this capability. We are experimenting with it 00:49:54.400 --> 00:50:00.880 to see how powerful it is. But yes, we are going 00:50:00.880 --> 00:50:05.659 to create multi -capability to integrate with 00:50:05.659 --> 00:50:08.639 Ma-Talk AI. Yes. Because I wonder if even for someone 00:50:08.639 --> 00:50:11.980 who does have dexterity and fine motor skills 00:50:11.980 --> 00:50:15.460 with their hands, If I gaze in a way could be 00:50:15.460 --> 00:50:17.840 faster in any manner, or like, I don't know, 00:50:17.940 --> 00:50:20.099 even in like a futuristic standpoint, like, if 00:50:20.099 --> 00:50:22.079 there's a way for it to integrate with your thoughts, 00:50:22.079 --> 00:50:25.099 like, as soon as you think something, it articulates 00:50:25.099 --> 00:50:28.480 it. I don't know what technology bridges that 00:50:28.480 --> 00:50:34.719 gap outside of implanted invasive, invasive technology, 00:50:34.719 --> 00:50:39.610 but it would be cool for for someone who. has 00:50:39.610 --> 00:50:43.329 been unable to communicate due to various barriers 00:50:43.329 --> 00:50:46.349 of, if you could kind of tap into what their 00:50:46.349 --> 00:50:49.329 thoughts are. Absolutely. Absolutely. And this 00:50:49.329 --> 00:50:52.010 is definitely on the roadmap. We are, again, 00:50:52.349 --> 00:50:57.050 we, one of Verbali is a pillars or, and I talk 00:50:57.050 --> 00:51:00.210 pillars is superb user experience. We're not 00:51:00.210 --> 00:51:03.849 going to introduce something that is just good 00:51:03.849 --> 00:51:11.920 enough because we need kids and parents to be 00:51:11.920 --> 00:51:16.260 able to rely on it and The way to do it is to 00:51:16.260 --> 00:51:22.000 have a user experience that is as good as possible 00:51:22.000 --> 00:51:26.920 So we are experimenting with and part of it is 00:51:26.920 --> 00:51:30.800 going back to the eye gazing, right? We're trying 00:51:30.800 --> 00:51:33.539 first to see what the current technology that 00:51:33.539 --> 00:51:37.920 you have on your phone has Or your tablet what 00:51:37.920 --> 00:51:40.980 it's good for, because then you don't need to 00:51:40.980 --> 00:51:44.500 add anything, and then it's as light as possible, 00:51:44.760 --> 00:51:49.219 right? And if it's good enough, or sorry, if 00:51:49.219 --> 00:51:55.340 it's good as it should, right, then we can incorporate 00:51:55.340 --> 00:51:59.480 it this way to Ma-Talk AI. So this is where we are 00:51:59.480 --> 00:52:02.079 now. We're experimenting to see what is the current 00:52:02.079 --> 00:52:04.760 technology that you're already hardware devices, 00:52:04.840 --> 00:52:08.280 so you won't need to buy any other additional 00:52:08.280 --> 00:52:12.019 technology. Inclusivity is not just the ability 00:52:12.019 --> 00:52:14.539 to be part of something, but it's also financial 00:52:14.539 --> 00:52:19.079 inclusivity. And this is where part of our mission 00:52:19.079 --> 00:52:22.260 at Verbal is to give the ability for everyone 00:52:22.260 --> 00:52:27.599 to... have AAC without the thought about how 00:52:27.599 --> 00:52:29.920 much does it cost? It's too expensive. That's 00:52:29.920 --> 00:52:33.019 why we're introducing it in Apple. We're introducing 00:52:33.019 --> 00:52:35.219 it in iOS. That's why we're introducing a talk 00:52:35.219 --> 00:52:39.019 with Android. And we will introduce it with Kindle 00:52:39.019 --> 00:52:42.280 Fire very soon. Remember, the prototype started 00:52:42.280 --> 00:52:44.380 with Kindle Fire. And that's where we're going 00:52:44.380 --> 00:52:48.840 because we know a lot of homes have those Kindles. 00:52:49.070 --> 00:52:53.090 and they're cheap and they are very safe for 00:52:53.090 --> 00:52:58.650 kids and they're kid -oriented. So it will help 00:52:58.650 --> 00:53:04.250 the kids get the AAC that they need faster and 00:53:04.250 --> 00:53:07.159 a more accessible way. Do you envision having 00:53:07.159 --> 00:53:11.039 a suite of products or like on, on the five year 00:53:11.039 --> 00:53:14.559 vision, do you think it's my talk AI with different 00:53:14.559 --> 00:53:17.079 features? Do you think it's different apps or 00:53:17.079 --> 00:53:22.300 different software within verbally? Yeah. So 00:53:22.300 --> 00:53:26.179 I do think there'll be a suite of products. Definitely, 00:53:26.380 --> 00:53:28.460 we're constantly adding more features, constantly 00:53:28.460 --> 00:53:33.820 adding more features. One of the first features 00:53:33.820 --> 00:53:35.880 that we're introducing now today, it's actually 00:53:35.880 --> 00:53:38.659 we got the approval from Apple today, was a free 00:53:38.659 --> 00:53:42.920 trial, seven -day free trials. When we launched, 00:53:43.139 --> 00:53:46.139 we got amazing feedback. We got, this is amazing, 00:53:46.239 --> 00:53:53.000 so we got... The International Society of AAC 00:53:53.000 --> 00:53:55.440 president reached out to us. She said, this is 00:53:55.440 --> 00:53:58.980 amazing. Parents said, and we understood that 00:53:58.980 --> 00:54:01.980 the technology is legit. Our technology is legit. 00:54:02.780 --> 00:54:06.760 The hunger for death technology is there. We 00:54:06.760 --> 00:54:09.119 just need to make it more accessible. So our 00:54:09.119 --> 00:54:11.320 first feature that we put on the roadmap was 00:54:11.320 --> 00:54:15.099 to introduce the seven day free trial. So more 00:54:15.099 --> 00:54:18.639 people will be able to use it easily without 00:54:18.760 --> 00:54:21.019 thinking, oh, I don't know, I don't want to pay 00:54:21.019 --> 00:54:23.179 first month or something. Just use it, download 00:54:23.179 --> 00:54:27.599 it, enjoy it. And so today it's available already 00:54:27.599 --> 00:54:30.960 with Apple. Android will probably be this week 00:54:30.960 --> 00:54:34.920 or next. We're waiting for Google to give the 00:54:34.920 --> 00:54:40.119 thumbs up. And so this is from a feature perspective. 00:54:40.380 --> 00:54:44.460 We're constantly adding more. From suite of products, 00:54:44.860 --> 00:54:49.619 yes. I look at it more like an ecosystem that 00:54:49.619 --> 00:54:55.800 goes back to how much Ma-Talk is customized, how 00:54:55.800 --> 00:54:59.500 much it can help the kid. So AAC helps the kid 00:54:59.500 --> 00:55:02.860 communicate. But it's not just the only way because 00:55:02.860 --> 00:55:08.769 the kid grows. want to do more things so why 00:55:08.769 --> 00:55:13.250 not helping them talk for example not just communicate 00:55:13.250 --> 00:55:16.829 but actually speak right today you need to do 00:55:16.829 --> 00:55:20.050 you say sight words or stuff like that but what 00:55:20.050 --> 00:55:22.869 i guess a lot of speech therapists will tell 00:55:22.869 --> 00:55:24.750 you you need to find the right motivation for 00:55:24.750 --> 00:55:29.190 the kid to to cooperate but what Ma-Talk will know 00:55:29.190 --> 00:55:32.650 Ma-Talk will know what the kid likes so if Ma-Talk 00:55:32.650 --> 00:55:36.880 knows what the kid likes then why not use this 00:55:36.880 --> 00:55:41.179 data in order to customize also the teaching 00:55:41.179 --> 00:55:44.199 process of sight words, find the right words 00:55:44.199 --> 00:55:49.239 that excite the kid to say it, right? So we can 00:55:49.239 --> 00:55:53.820 do that. We can use the data also to customize 00:55:53.820 --> 00:55:57.840 learning material specifically for the kid. Again, 00:55:57.840 --> 00:56:01.480 it goes back to what the kid is motivated by. 00:56:02.820 --> 00:56:06.059 So I mentioned in the beginning about the data 00:56:06.059 --> 00:56:11.840 dashboard for the speech therapist. This data, 00:56:11.980 --> 00:56:15.039 it's a collective system that you have a lot 00:56:15.039 --> 00:56:21.460 of different corners that feed each other. So 00:56:21.460 --> 00:56:23.800 you will be able to enjoy it in different scenarios 00:56:23.800 --> 00:56:27.800 in your life. Also, smart keyboard that you can 00:56:27.800 --> 00:56:32.079 use tone with, voice tones, right? And customize. 00:56:32.400 --> 00:56:37.400 Yeah, we, yeah, we have, the sky's the limit. 00:56:38.079 --> 00:56:39.440 Yeah. Well, I'm looking forward to seeing how 00:56:39.440 --> 00:56:42.059 it develops. I know you guys have a wait list 00:56:42.059 --> 00:56:46.099 to join the app, but if people want to learn 00:56:46.099 --> 00:56:48.900 about the software, if they want to try out the 00:56:48.900 --> 00:56:52.880 software, where can they go? We can include the 00:56:52.880 --> 00:56:54.719 exact links in the show notes, but what's the 00:56:54.719 --> 00:56:58.160 best way to follow or support your work? Yeah, 00:56:58.160 --> 00:57:01.559 so obviously download the app now it's there's 00:57:01.559 --> 00:57:03.679 no waitlist now not just can go ahead and just 00:57:03.679 --> 00:57:07.599 don't the seven -day Free trial just go ahead 00:57:07.599 --> 00:57:11.679 use it try to like it amazing If you have and 00:57:11.679 --> 00:57:15.719 we also have on our website We will incorporate 00:57:15.719 --> 00:57:18.159 also eventually in the app itself but in the 00:57:18.159 --> 00:57:21.880 website if you have Goes back to unique solutions 00:57:21.880 --> 00:57:24.800 if you have something that you would love to 00:57:24.800 --> 00:57:33.260 see on your kids' AAC. Or, my talk AI is amazing, 00:57:33.340 --> 00:57:35.599 but it's missing this for it to be perfect for 00:57:35.599 --> 00:57:38.619 my kids, right? We have a space on our website 00:57:38.619 --> 00:57:41.760 that you can actually write us specific feedback 00:57:41.760 --> 00:57:45.159 with feature requests, language requests. And 00:57:45.159 --> 00:57:48.019 we are actively reading this and we are actively 00:57:48.019 --> 00:57:51.619 exploring ways to incorporate all of those requests 00:57:51.619 --> 00:57:56.690 in our future offering. So that's another way 00:57:56.690 --> 00:58:02.949 to reach out to us. I'll add my email address 00:58:02.949 --> 00:58:06.929 to the podcast afterwards. So if someone wants 00:58:06.929 --> 00:58:10.630 to reach out, ask questions, anything, yeah. 00:58:12.809 --> 00:58:16.230 Yeah, no, I love the technological innovation. 00:58:16.710 --> 00:58:19.949 I love products that solve real -world problems, 00:58:20.530 --> 00:58:23.349 especially when the origin story is deeply personal 00:58:23.349 --> 00:58:26.710 for the guests. So yeah, I really enjoyed learning 00:58:26.710 --> 00:58:30.869 about it, talking about it, and I'd love to see 00:58:30.869 --> 00:58:34.489 it widely adopted in schools and gyms and different 00:58:34.489 --> 00:58:37.590 environments. So Shi, I appreciate not only your 00:58:37.590 --> 00:58:39.469 time this morning, but your mission and the work 00:58:39.469 --> 00:58:41.840 that you've done. Thank you very much Brendan. 00:58:42.000 --> 00:58:45.739 This was amazing. Thank you very much Thank you 00:58:45.739 --> 00:58:48.179 for listening to the adapt X podcast Our effort 00:58:48.179 --> 00:58:50.000 to amplify the ideas of our guests and create 00:58:50.000 --> 00:58:52.420 more inclusive and accessible industries is futile 00:58:52.420 --> 00:58:54.360 unless these episodes reach a larger audience. 00:58:54.800 --> 00:58:56.780 If you enjoyed our discussion today, please leave 00:58:56.780 --> 00:58:58.739 us a rating or a review on whichever platform 00:58:58.739 --> 00:59:00.800 you use. And if you would like to learn more 00:59:00.800 --> 00:59:02.860 about AdaptX, the course that we teach to health 00:59:02.860 --> 00:59:04.619 and fitness professionals and the projects that 00:59:04.619 --> 00:59:07.059 our organization is working on, you can subscribe 00:59:07.059 --> 00:59:10.079 to our newsletter through our website www .adaptx 00:59:10.079 --> 00:59:11.739 .org. Until next Monday.