Subscribe
Share
Share
Embed
"What if we Could?" A podcast exploring this question that drive us. We explore the practical application of artificial intelligence, product design, blockchain & AR/VR, and tech alpha in the service of humans.
David DeVore (02:21.448)
Hey, what's up? This is David DeVore with the What If We Could Show where every week we ask that burning question around artificial intelligence, tech alpha, potentially we talk about blockchain as well as user interface design and so forth. Super excited because this week we have all four mesh co-founders in the room. We've got Bob, we've got Calvin, we've got Kevin and myself. And this week we're going to break open a really awesome topic which is
talking about using artificial intelligence to support those autistic and people who are neurodivergent in living a better life. And so super excited about popping this open. So we're going to hand it over to you, Bob, like to kick it off and like talk about your experience and sort of how we have come to this topic and what we're hoping to do potentially.
Bob Ullery (03:38.582)
Yeah, I'd love to. And let's just start by setting the table. No topic off limits here. We're very transparent, open book. It's been a long journey with Easton, my nine-year-old. It's hard to encapsulate nine years of a brain dump into a podcast. But open book, if you got questions, let me know. I'm actually the father of two autistic children. My six-year-old's also on the spectrum.
but not as severe as my nine year old. Now I'll start with just a really kind of fun alignment to autism I think, which is in our experience anyways, both with our kids and others that we've met, oftentimes we see that these kids have one or more superpowers too. And so I wanna start there. Easton's superpower I think is creativity. He loves to draw, he paints, he loves to sing, even though he doesn't talk.
He loves to sing, especially nursery rhyme, stuff like that. And Murphy's superpower is math, like profoundly so. Savant level math. When he was in kindergarten, he's in first grade now. He's in third grade math class currently. And in kindergarten, he was reciting prime numbers up to like 500 off memory. Nothing, I taught him. I did not ask him to memorize prime numbers. He just got into them and started internalizing them and does some fairly
complex math, just mentally, which is really interesting. And he, other than that, he's a absolutely normal six-year-old, you know, he loves doing all the things that six-year-olds do, playing roadblocks, you know, run around outside, as does Easton, very physical, loves to scooter, loves to bike, loves to run, loves to do all the things that typical nine-year-olds do too. So you often hear parents of just, I think,
Generally kids with special needs say the sort of phrase, like, I wouldn't change them if I could. And that's absolutely true. I don't know who he would be if he wasn't himself. And we love him. And here we are talking about how AI might be able to improve his life. And I'm just really excited by that, and certainly very grateful to have you all helping in that endeavor. I think it's very exciting sandbox to play in, not because it impacts our family directly, but.
Bob Ullery (06:03.422)
It's interesting application of technology to improve people's lives even more broadly than just autism and theoretically makes a wonderful business too where you can do well and do good simultaneously and I'm always excited for those opportunities.
David DeVore (06:19.41)
That's awesome. I mean, I think one of the things that you brought up that's really interesting, right, and especially when we talk about autism and neurodivergence is like, rightly termed it as a spectrum. And the
the way that each individual's brain behaves and also their capabilities and ability to navigate the world is not only very different from our own, but very different from each other. And each person has some characteristics in terms of how they operate, but also can
widely vary from age and experience out to what their superpowers are. And so one of the things I think that's really interesting about this topic and merges with other topics that we've talked about is AI ultimately has the ability to create really individual experiences at scale.
And so when we think about, you know, even the myriad of, you know, apps and tools out there that have been developed for people to be able to accelerate learning or be able to accelerate behaviors of pattern and whatnot, they are all, while somewhat personalized,
they still are very deterministic, right? And so one of the things I think that's really exciting about where the technology can help is that we can think about individual tools for individual people and individual experiences in real time that are supportive, right? And so what I'd love to sort of unpack a little bit is the
David DeVore (08:33.714)
how we came to the conclusion based on our prototype, which was very different. Our prototype was taking a F1 race and providing all of the real-time data around that race to then provide a feedback loop in real-time to an announcer, right? So I'd love to sort of say, how do we get from there to saying, wow, this...
This is interesting in my own use case of, how do we help Easton, right? And so it's super fun to sort of think about the technology and how we can apply the technology in very interesting ways, but in a lot of ways, it's the same stuff underneath the hood.
Bob Ullery (09:09.41)
Yeah.
Bob Ullery (09:26.558)
Yeah. I mean, when we were so for everybody listening in the podcast, you know, I think it was Thanksgiving week. We had an opportunity to build a POC collectively as a proxy to the Abu Dhabi Racing League. And we aligned to a race that was going on live for F1 racing. And the general premise behind this was can we have an agent really observing the environment of the race, right? And an environment for in this POC was
Can it consume information from lots of different sources in real time and then form ideas about what's going on in the race? Like can it add value from an entertainment perspective or maybe historical one, statistics, like just add additional layers on top of the commentary. And so that agent was watching live blogs from F1 and social channels. It had...
some information on the history of F1 in terms of like, who has won previous races and the sort of history of success and drivers and cars and teams and all that. And then the last bit was the really, I think very interesting part, which was it was watching ESPN live. And what it was doing was consuming the audio from the stream, right? And internalizing it. So...
audio comes in, transcribe it to text, and put it into the brain, just like it came from the blog or from any other source that we were piping into it. And then it would wake up and say, is there anything interesting to say right now? And we collectively, all of us sort of sprinted to do that like two days before Thanksgiving. And I was on my way to a family event in Iowa for Thanksgiving, having worked shotgun on this thing. And when I got there, it was on my mind. So I was showing my extended family this thing.
And that's where it really clicked, right? Cause Easton's in the room and they were spending quality time with them. And I sort of hadn't this epiphany. It's like the ability for an agent to observe an environment from many different angles at the same time is clear. Totally feasible. And we proved that. Wouldn't it be interesting if we could do the same for Easton who largely has a problem observing and internalizing his environment. And so that's kind of where we started.
Bob Ullery (11:42.494)
in terms of the concept. Wouldn't it be interesting if there was an agent could listen in on Easton's environment when he's doing things and sort of prompt him, instead of him prompting the AI, can the AI prompt him to work on some of the skills and goals that are set out in his individualized education plan from the state? It was called an IEP. All special needs kids have one. It's mandatory by federal law.
Every special needs kid in a public school has an IEP. And what that means is all the admins, the parents, everybody in his orbit comes together once a month, or depending on your school system, however long. And we establish goals and we measure his performance against those goals and we course correct so that he's getting the resources he needs to learn the skills in order to operate in the world like anybody else. And so that's where it sort of came to a head. Wouldn't it be great if
If he's asking Alexa to play a nursery rhyme, and we know he's working on what's called WH questions, who, what, where, when, how, his typical motion is to ask Alexa to play Old MacDonald a hundred times in a row. And that's the extent of the interaction. Wouldn't it be great if this agent could prompt him after that song and be like, hey, Easton, why don't you ask Alexa what kind of animals are on a farm? And though it sounds benign to us,
to him it's a feedback loop, right? It sort of helps harden those connections in his brain, just like any AI model we're gonna put together, right? That connection gets more highly weighted and he knows when he's curious about something that he can ask a what question or a how or a where question and start to work on really the bigger goal which is, you know, two-way conversation. He's not there yet, but those are the small stepping stones to get there.
I think there's a ton of value to sort of have that side-by-side proxy for his technicians that he's worked with in autism centers, sort of being there in a light mode to continue that work as he goes so he can harden those skills as he learns.
Calvin Hoenes (13:59.559)
That's really fascinating. Like, what do you say that it is basically, I mean, if you look at GPT or if you look at transformer technology in general, right, it is basically that transforming one quote unquote truth to another truth, right? Like one mapping them together in an n-dimensional space. And we say that could directly apply them to, you know, the auto, autism spectrum or the neurodivergence. Like, do you think like basically would it be evaluating or
building or training a specific model to do that, right? To see where the gaps are and then quote unquote, what should this translate to? Is that a fair way of seeing how awesome IEP would work?
Bob Ullery (14:40.554)
Yeah, you jump the shark. That's exactly where my mind goes, right? Like this initial swag, I think is incredibly valuable, right? Like bringing in IEP, that's easy to reconcile in an LLM and here's an opportunity to work on a goal that's, you know, known. Um, but man, LLMs on, on fine tune models, even not, not fine tuned, but really great at translation and transformation. Right. And I think that's probably one of the big gaps, at least from Easton, but I also have a hunch.
uh, autism spectrum disorder in general. And I'm not a, I'm not a certified expert here, so I may use the taxonomy wrong, but I have this hunch that they have a, many of them have a hard time sort of processing the world in the same way we do. It's like, if I said, Hey, uh, this picture is green. How do I know that my green is the same green that you see out of your eyes? Maybe it looks like my blue to you.
It's sort of weird to think about in that way. And so I think the future here, and I hope we can iterate towards this if we're successful here, is this notion of like personalized models. And the way that Eason views the world, can we create a model that can translate that view for him or vice versa real world into his model so that he can parse what he's seeing? An example of that is,
He spent six years at a ABA center, applied behavioral analysis, which is a form of therapy for autism, where there's a lot of repetitive tasking, exercises, drills, right? And early on, they work on things like association. An example of that is a flash card, right? So, Easton, what's this? And they'd hold up a picture of a dog. And so,
You go ask him in the real world if we're at the grocery store. I don't know if you see a dog at the grocery store, but let's go with it. There's a dog at the grocery store. Uh, he said, what's that? He wouldn't know why, because it's not drawn in black and white on the flash card, he's associated the black and white hand drawn dog from his card to what a dog is, every representation of a dog. Of course we know there's a million. I don't even know how many variants of breeds of dogs there are. Um,
Bob Ullery (17:07.062)
I think there's a future here where there's, you know, beyond audio, it's head up, heads up guidance. Like could it translate things he's seeing through a Apple vision pro or Google glass like device and say, Oh, the devices seeing a dog and then translating it back to the flash card that he knows so that he can begin to build those associations between something he knows to all the variants that might be in the real world.
And that's probably a function of a very personalized model that's trained on his interactions with the world and trying to find those correlations between things that we don't see being correlated.
Calvin Hoenes (17:49.407)
It's fascinating, right? Because exactly what you said is like, how much can it augment to like say quote unquote experience more of the shared reality to make the day-to-day, the everyday life experience easier for him, right? I think that's what you touched on now is one of the big aspects that it could potentially do. How do you feel about sort of the other side of the spectrum? Because you mentioned sort of the superpower aspect of.
And what I'm curious to see is how would you balance nurturing both sides of the spectrum? Because a transformer model could essentially do both, potentially. One could be make your everyday, day-to-day translation exercise or navigating the environment easier. But it can also detect or figure out where can it help enhance where already existing.
Bob Ullery (18:35.77)
Mm-hmm.
Calvin Hoenes (18:44.403)
you know, capabilities, skills, or, you know, quote unquote, superpowers are, right? So it could do both. And like, how do you balance that? Because I think what is interesting with what you mentioned with like the ABA or the EAPs is like, they're obviously more skewed towards putting that into the more shared reality experience of the site, versus, you know, harnessing some of the exciting capabilities that are there too. So basically the question is like, how do you feel about balancing the two?
Bob Ullery (19:11.394)
I think it's an opportunity, right? And could you blend the two? Is there a concentric circle here? And what I mean by that is, like I said, he loves being creative, loves to draw as an example. Alexa not required, nursery rhymes not required, right? If he's in the moment, one exercise could be inserting WH questions against one goal in his IEP into the creative process, you know?
something we want to kind of POC here too, is this notion of, um, L C M or, or latent stable diffusion. Um, but high horsepower and fast. So like, wouldn't it be cool if he's drawing something or if he's just even working on verbalization of something he wants to see on the device and is generating in real time. Um, maybe the agent comes over the top, like what else should we add here? Eastern, you know, like there, there's plenty of opportunity to mix in those, those exercises on goal.
attainment into the activity he loves doing, and probably a huge value there too, so that he enjoys going through the process because he's actively doing something he likes doing versus, you know, having to think of it more like a drill, like I don't know any kid that would want to do that over and over. And just a little thing I'll share. He's the most
sweet human being you ever met in your life. There's not a mean bone in his body, seriously. He's got the disposition of probably a three or four year old and is very sweet one, right? And so you can think of a three year old on his best day is Easton's demeanor. And he's a work horse too. And to put in perspective, like for ABA therapy, it's a medical service.
It's not an education one. So, you know, they follow bank holidays. The kid over six years had eight days off a year and he went to ABA therapy from 8.30 in the morning until 4.30 PM every day for six years with eight days off a year. And he's still the most cheery young man you ever met in your life. So he's down to work big time. I like to say he gets it from me. I think he probably gets it from mom, but.
Bob Ullery (21:33.57)
You know, they're perfect test subject for tinkering and maybe an opportunity since I'm his dad, I can clear out some of the red tape faster than we would otherwise. But there's lots of trials to be had and lots of cool concepts we could try with him.
Kevin Nuest (21:49.638)
definitely, uh, gets his work ethic from his, from his dad. I'll vouch for that. You can give credit to mom too. We'll clip that for her, uh, to Calvin, I wanted to, I wanted to throw out, uh, take a slight detour, throw out an analogy for a moment, thinking about, uh, the question you post, right? Balancing both, uh, increasing proficiency, but also doubling down on super power. Right. So there's, there's an assessment, uh, this, this is, I'm now jumping into you.
The realm of, uh, of, of business, of, uh, career growth and using this as an analogy here. There's an assessment called strength finders. Are you familiar with that Calvin or, or Dave, you ever. So it's one of my favorites. I got introduced to, um, uh, a long time ago. And as you know, in business, uh, there's a lot of different in professional development, there's a lot of different assessments and they say a lot of different things, you know, Myers-Briggs is the easiest, most, most approachable.
Calvin Hoenes (22:31.488)
I'm not.
Kevin Nuest (22:46.914)
that people understand about, um, you know, personality, for example, uh, the, um, strength finders is more about. The, uh, you're, you're controllables and as an individual. And so I've found it really helpful to help others understand themselves and, uh, their development. And the theory is you do this assessment 45 minutes to answer a bunch of questions. And out of that, you're, uh,
There, it breaks it down into about, I think it's 36 different skill sets that breaks down, um, your professions in. Right. So you have things like communication and strategy and futurist, for example, as some of these different categories and, uh, you get back this report and it tells you your top five strengths says, what are you really great at? Right. Slash. What do you really enjoy doing as well? So these are.
really kind of one in the same merging those together. And then conversely, it gives you, you get the, you pay for it and you get the full report, you get all 36 skills ranked. So now you know what are at the bottom of your list as well. And you might think, well, Hey, I need to go double down on the ones on the bottom and increase those. Right? So if I got a hundred units of time to go do professional development, improve myself, I should take those bottom five skills and you know, 50% of the time there.
And maybe smear 50% at the top or something like that, and start coming up with ways to say, how do I get better? I obviously need to improve the things I'm the worst at. And so I've had the chance to manage a lot of, um, a lot of different people in positions. But the one I think of the most is in product management and product management. It's a profession where it encompasses so many different skills because it is this, this amorphous job, depending on the organization, the situation, the product, the team, the timing.
It can mean a lot of things and, and regardless of exactly what it means in any organization, it usually means a lot more things than a lot of other jobs. And so I have, I've had a lot of career conversations with product managers and I've done this assessment with them. And I, and I've shared with them both in setting expectations, what I expect out of a teammate, but then also helping them grow and it's a lot of conversations around here are the different areas of the role of product management.
Kevin Nuest (25:09.898)
I need you to be proficient across these areas, right? I need you to be at the proficient level. You will not be a 10 out of 10 at what is effectively like call it 10 different jobs, 10 different skill sets. You won't be, and that's okay. I need you to be proficient. And then I wanna help you find your superpower where you crush one or two of those that you can run forward with. And I can help you get a promotion.
by using that skill to Excel and you can leverage that through the rest of your professional career. Right. So to me now to bring it all the way back to what we're talking about to, when, when I hear you Calvin ask about like, how could you, how could you balance that, that feels like just an amazing opportunity, right? Bring, uh, bring a child up to proficient at different skills and then let them run with their super power and let them go crush it. They have, they have something to offer to the world.
Uh, this differentiated and we, you know, if we could help them unlock a fraction of that would be phenomenal.
David DeVore (26:12.174)
Yeah, I mean, the other opportunity that I think, you know, I looked around and becoming familiar with some of the tools out there and there's part of the challenge, of course, is communication, right? And I think part of the opportunity is like also helping caregivers, right? And I'm interested from your perspective.
Easton's communication style. There's also folks who are completely nonverbal. And they're, you know, that can be a real challenge for caregivers to understand what they need to understand what they want to understand what they're feeling. And I'm sort of also, you know, thinking about ways that of, of helping. Well, well,
an AI could potentially, it can potentially help consume the world for, for a person's autistic, it could also translate their world out to caregiver givers, right as well. I'd love to just, we haven't really talked about this in any way, we've talked about Eastern scope pilot. But I'd love to just sort of
Like is that what do you what do you think like what would that potentially look like? for you for Easton and you know, where do you think sort of LLMs and Artificial intelligence can play a role to help on that side of the fence
Bob Ullery (27:53.55)
lot of thoughts come in my mind. I'll start with a dream of mine, right, in terms of like, one moment I have in my mind that if we can achieve this, we know that we've made a debt in the universe. This one thing, and it's very, very simple. So for since he was two years old ish, and still today, sometimes he'll walk around and he will
write something in the air with his finger. We to this day cannot figure out what he's writing. And it has impacted a lot of people in his life are also curious about this, including the teachers and techs he's had in ABA therapy. He finished ABA in December. He's now full-time in the public school system. We want to start working on academics too. But when we left, they threw a going away party for him.
And his first tech ever is completing his masters now, six years later. And Easton was also his very first client. And so made him decide this is what he wanted to do with his life as an adult, right? He's probably late 20s, so sort of starting the career. And when we had that going away party, they gave us this book. And in the book was a page for every one of the technicians
worked with him ever with a Polaroid and a handwritten note from each of them. And it was like 45 pages long. Incredible. And the last page was a tech that he had last year for, I think a few months. And at the end of this, he had a PS and he said, um, I always wanted to know what you were spelling in there. I hope one day I'm, I'm able to figure that out and find out. And to
David DeVore (29:28.926)
Yeah, it's awesome.
Bob Ullery (29:49.55)
bring this back, like I think there is a solve here, right? If we can train models on his behavior and give enough trials of him doing that, it is absolutely feasible that we can apply this technology to figure out what he is writing in the sky. And I think if we can figure that out, we'll learn a lot more about who Easton is and how he thinks internally. And I'm really excited by that. So I wanted to anchor to that. In terms of communication.
David DeVore (30:07.739)
Mm-hmm.
David DeVore (30:16.312)
It's awesome.
Calvin Hoenes (30:18.207)
That's fascinating.
Bob Ullery (30:21.134)
If you have a question, feel free. I've got other stuff I was going to toss out there.
Calvin Hoenes (30:25.139)
I was just going to say, you know, we, you remember this, there was this discussion two months ago about how we're learning to understand the whale noises, right? And how we're trying, using AI to decipher, communicating with whales, right? And I'm sitting there and thinking, you know, if we can learn to decipher whales, like we should, you know, we should learn how to apply that to like neurodivergence and solve that. And like you said, there's so much to uncover there. And I'm confident we can figure out a way to...
David DeVore (30:38.686)
100%.
David DeVore (30:51.759)
All right.
Calvin Hoenes (30:55.387)
Learn that language.
Bob Ullery (30:57.474)
Yeah.
David DeVore (30:58.414)
Yeah, I mean, we think I'm in the on the animal side. I think that the last I heard like we think that we're going to understand dolphins and whales and quickly other there's I think it I there's a recent podcast that I listened about this and we're going to quickly be able to understand animals. For sure. It's gonna be fascinating.
Bob Ullery (31:21.765)
They're gonna hate us. They're gonna be like, what are you doing? What are you guys doing out there?
David DeVore (31:27.173)
Yeah.
Calvin Hoenes (31:27.241)
There's a non-zero chance that this is...
Bob Ullery (31:27.674)
Can somebody turn the water down?
Kevin Nuest (31:30.946)
We have a lot of explaining to do, yeah, for sure.
Bob Ullery (31:31.13)
It turns nervous right off. Yeah.
Calvin Hoenes (31:35.631)
But it's interesting because just recently with like you tapped into LCM before, right? Like just if you look at the last two months, right? The technology that we now have at our hands where not only is, you know, are the AI models becoming more capable at a smaller size, but also like we are getting to a point when our real time has become feasible, right? And it's been a wild dream for a long period of time, but it has actually become feasible just in the last...
30 to 60 days and you have to think about that, right? And now we can actually get a real time interpretation and transformation and then also putting it back out. So just what you just explained before, like the ability to perceive and respond to the environment in a loop in real time is now feasible. And that unlocks a whole different universe.
Bob Ullery (32:24.032)
And for every-
Kevin Nuest (32:24.215)
Yeah.
Bob Ullery (32:27.838)
And for every individual at scale, right, I think the key word here, and we, we precinct on this podcast in terms of like what we might should or should not say applied to, to autism, but I'll get a little maybe on the edge here. Specs, when you start to think of the world and spectrums, it, it clears up a lot, and it provides a lot of empathy. Right. And so
David DeVore (32:56.769)
100%.
Bob Ullery (32:57.426)
ASD spectrum is very, very wide. I've met hundreds of kids over the last six years that are, every single one of them is totally different, right? So that's a true statement. But you think about the political strife and some other things going on in the world, and the takeaway there is, it seems to be many, if not the majority of human beings on the planet see the world in a binary way.
A or B and not as a spectrum. And so if you applied it to Easton, he's on the B side, right? He is a special needs person. They're all the same, right? They can't do stuff like we can on the A side, right? But there's other examples of this too. The one that it would be controversial to say potentially but I think it's a no brainer is sexuality, right? Sexuality is a spectrum. If you see the world in binary, I can point.
you to people who want to do things like implement policy to make it so that LGBTQ people can't marry each other. That's a binary view of the world. Not understanding that everybody is somewhere on the sexuality spectrum. You are not an A or a B. And there's intelligence, athletic ability, whatever it is, everything is a spectrum. There is no such thing as binary except in trad computing.
and ones and zeros, right? Where the term comes from. Everything else is a spectrum.
Calvin Hoenes (34:27.165)
Mm-hmm.
Kevin Nuest (34:29.45)
Yeah, which we've talked about.
Calvin Hoenes (34:30.251)
And even there we're likely to move away, right, in some point in the future, if we get to quantum.
Bob Ullery (34:33.358)
Right. Yeah.
Kevin Nuest (34:36.247)
To, to, to probabilities. Yeah. So, so prob, you know, we've talked about this before of thinking in probabilities versus thinking in, uh, in binary when assessing opportunities in front of us and, and potential risks, right. And it helps actually clarify a lot as we operate a business. It makes sense that the brain also works like that. If, you know, some of us have had the opportunity to build, um, uh, neural network.
deep learning models and what are those? They're branching probabilities applied, right? And they're, we built them as humble humans based off of our understanding of the neural network of our brains. So why, if our brains operate in probabilities, why wouldn't all of these possible skills and expressions be probabilities, AKA spectrums, right?
Bob Ullery (35:30.714)
100%. And that that's actually a good, I guess, corollary to why we made a decision to stop ABA therapy and then go direct full time school, right. And as I mentioned before, ABA is a medical practice. And so actually, by debt, by law, they are not allowed to teach academics, right. And so he's nine should be in third grade.
I sort of tell people this often, you know, this notion that maybe one day we could cure autism, right? That would be phenomenal, right? But it would probably really only apply to very young children with it, right? Two, three years old, not necessarily a nine-year-old. It's not like you could fix that necessarily if it's DNA or whatever it is, right?
fix it and then all of a sudden he's gonna wake up the next morning, he's at parody with his peers in a third grade math class because those connections in his brain were never formed. Just like a machine learning model. If you don't have the training data, there's no connection there and you can't infer off of it. And just like that, he wouldn't be able to just do that. That was the other kind of driving force behind thinking about like Easton's co-pilot was this, the other side of the equation is can we do
something to attempt at least to keep him on track with his peers. And the academic side, you have a kid going through ABA, that's great, but like, as it's working on goals or whatever, like can we introduce some learning as well to make sure they don't fall too far behind and are squarely only focused on life skills. So I'm excited by that opportunity too. I think that spans well beyond autism.
obviously too.
Kevin Nuest (37:29.742)
Can we rewind and hit a couple of questions that I don't think we hit at the top? It may be helpful context. How many people are impacted by autism?
Bob Ullery (37:41.718)
Last I saw, one in 40 children estimated this year will be diagnosed. It's a very large swath. I mean, it's a massive population.
Kevin Nuest (37:47.83)
Wow, that's a lot.
Kevin Nuest (37:53.742)
That's crazy. And then you talked about getting certified as special needs, which then led down the path of by federal mandate, being able to get, uh, an IEP, which if you could define IEP again, I missed it. Uh, I don't want to get the definition wrong, but the question being like, how, how did you get certified? What does that, what does that mean? Like you should, you know, you go to the equivalent of the DMV for autism and you're like, I don't get my number in line and wait, like, what, what is that?
process look like? How'd you even know they need to do that? Right? Like what? Can you walk me through that?
Bob Ullery (38:28.982)
Yeah, first question for IEP individualized education plan. Good on the federal government. You know, this has been around for a very long time and it's a mandate across all 50 states to make sure that specifically kids that maybe can't advocate for themselves are still getting the services that they deserve and that we deserve to give them. In terms of how we went through the process, you know, I think the...
Kevin Nuest (38:33.41)
Thank you.
Bob Ullery (38:54.626)
The first is probably a lot like the seven stages of grief in a lot of ways, in a fine way. Not that this is negative or anything, but in terms of how we processed it. Eason was our first kid, and so we didn't really have a great corollary to look at necessarily and say, oh, he's on track or necessarily there's something wrong. But right about two years old, we knew.
very little interaction and was kind of always in his own little lane, right? And so the process was a medical one again, right? Like doctor referral to a psychologist and then the psychologist meets with them for a couple hours and does some assessments. And, you know, these folks are trained and go through a traditional medical school. And I think within 10 seconds, they were ready to give the label.
Part of the challenge, I think, and we never necessarily struggle with this, but I think a lot of parents do, which is the wrong word to use would be denial. Fear. You know, like, what does this mean for us, for them, for downstream impacts, you know, stuff you never thought about, like creating a special needs trust in case you die so that he can have stuff and services and...
and feed himself or whatever. So the first stage is really around like the humanity of it and like processing that and getting over the fear to say, we're gonna do whatever we need to do to make sure he gets the services that he needs. And getting that autism diagnosis opened up a number of doors around Medicaid, a lot of additional programs, and then of course access to ABA therapy, which by the way, you cannot get without a diagnosis.
had a candid conversation with a friend from the Web3 space the other day. And they're experiencing that in Virginia where they sort of want to delay the diagnosis, just they're not sure about the implications of that yet, but they wanted to get them into ABA therapy and just pay out of pocket for it. It turns out you can't. I thought it was just very strange. You can go walk into it.
Kevin Nuest (41:15.342)
Mm.
Calvin Hoenes (41:19.927)
Do you know what the reasoning behind that is?
Bob Ullery (41:23.222)
I don't, I'm not sure it's policy driven. So something in the T's and C's, not sure if it was on purpose or just a byproduct of that law.
Kevin Nuest (41:35.278)
I was going to ask also, did you have to pay out of pocket for any of the assessments to get the certification or was that all covered through insurance? Did you even have insurance then?
Bob Ullery (41:48.194)
Yeah, I had a full-time like trad job at the time with good insurance and it just kind of all ran through that. And we spent a lot of money to give him the things he needs, right? Insurance does help a lot. You know, anecdotally, and I don't have numbers to back this up just talking to therapists at ABA centers.
if insurance wasn't there and they did allow you to pay out of pocket, it would be many hundreds of thousands of dollars every year. You know, it's pretty nuts. But rightly so. They have many people working there full time. They're well trained. They've got deep educational backgrounds, all versed on BF Skinner. You know, they're pros and so they're worth the money for sure. But asking any family to have to pay that would be really impossible for almost everybody.
Kevin Nuest (42:43.586)
Being in the.
Calvin Hoenes (42:43.771)
Yeah. And now think about one in 40 being impacted and now try and extrapolate that into, you know, the amount of, of service and care and infrastructure around that it currently needs to fulfill those needs. Right. And now going back to like, going back to the potential of augmenting that or helping with AI to do that, I'm thinking that would probably allow a lot more people, you know, to get the, the care needed.
Bob Ullery (43:12.618)
It's one in 40 now too. I mean, I would bet good money in the next couple years, you're gonna see that number go down to like one in 20 and then one in 10. I mean, just a couple years ago, it was like one in 150. Literally a couple years ago, it's accelerating, right? And so that...
David DeVore (43:13.019)
I'm in.
Calvin Hoenes (43:27.802)
That's interesting.
Calvin Hoenes (43:31.771)
Do you reckon it's different because of the way it's, is it because it is more commonly diagnosed now or has the experience of what the spectrum is and in that way the diagnosis expanded with it? Do you think there, you know, are there also other changes? I mean, I just, I'm asking because I just quite frankly don't know enough about it yet.
Bob Ullery (43:55.062)
And I don't either. I think that's plausible, right? Better detection number goes up. That means though that there's, maybe it's a consistent number. It's always been one in 40, but back a couple of years ago, one in 85 didn't get diagnosed, which I think is interesting. I've got my own thoughts on how and why this may occur. I don't have any scientific backing.
Calvin Hoenes (44:14.283)
Mm-hmm.
Bob Ullery (44:23.99)
my mind goes to hereditary genes. And I, you know, maybe, probably, but also I think the environment has a huge role here. Probably, right? This notion that we're, we don't take good care of the earth, our water supply, our food supplies, you know? Are we paying the consequences for that? What's that?
David DeVore (44:47.226)
And a lot of times not people either. I said, I did not end in a lot of times we're not taking great care of people either. Right. And, and especially those with mental health issues or other other, you know, like you said, like sort of a binary world, people who are not like us.
Calvin Hoenes (44:50.938)
Yeah.
Bob Ullery (44:53.603)
Yeah.
David DeVore (45:08.55)
you know, it's a, you know, love humans. At the same time, like we've got, we've got some problems when it comes to helping each other. I'm interested, and I know just as we sort of why and down here at the top of the hour, I'm interested in talking about interfaces a little bit, right. And we at this point,
you know, we have all the interfaces available to us, which is, you know, whether it's whether there's VR, AR, audio, you know, tap screens and apps, and potentially, you know, implants, you know, brainwave measures, I don't even know what we call them. Neuralink, right? You know, and I'm
Bob Ullery (46:00.334)
Like, Neuralink? Yeah.
David DeVore (46:05.978)
It'll be interesting. My question is a little bit like as we start to sort of think about interfaces, like what are the best interface? What do we think the best interfaces are right now? The interface really is tablets and, and the screen that tends to be where the, the apps exist today. Um, where do we think the interfaces are in the next five, 10 years?
Calvin Hoenes (46:33.303)
I just think I...
David DeVore (46:35.672)
Actually, let me take it back. Where is the interface that we think that we can have the most impact as we drive towards Eastern's co-pilot? What does that interface look like to sort of accomplish the things that we've talked about in terms of helping people?
Bob Ullery (46:56.942)
Good question.
Calvin Hoenes (46:57.267)
And maybe putting that to Bob, like, what is he responding best to in your experience so far?
Bob Ullery (47:04.458)
Well, I think the answer for Easton is rooted in the skills he's looking to acquire, right, which is communication. So I lean towards audio with him, which is part of the reason we positioned the POC in the way it is, right. A, it's sort of the cleanest path to test this out. And then also think it aligns really well to him, this notion of audio cues to him so that he can internalize what's being said, process it and then practice.
outbound communication in the same mode. I don't know what the answer to the broader question is other than I think the opportunity could be the right interface at the right time given the individual. So long as the experience sort of follows them from device to device, from interface to interface and is consistent, I guess we'll see. I don't know, Calvin, you're the UX whiz. What do you think?
David DeVore (47:45.06)
Hmm.
Calvin Hoenes (47:59.667)
Yeah, I've been thinking about it quite a little bit. I think I'm on the same page with you in terms of it does depend on the person, right? Like we said, it is a spectrum and like there's a variety of maybe call them impairments that, you know, where it makes sense to act differently. I mean, I do have the same feeling though that audio gets a very long way because it is comparatively.
less intrusive and is easier to get to you than trying to focus on a certain visual. A visual requires a lot of focus attention, which might not be ideal for some. And on the other hand, so that's what we said, I think in the near term, audio is probably a very good mechanism, is my hunch. But going from there, going to the other end of the extreme, what might this look like in 20 years?
you know, whatever time, however many time it takes is if we're looking at implant, I have just it's an interesting question, right? Like what if you start to use that transformer to actively change your perception on the fly in a multi sensory experience, just as we're thinking about multimodality in the AI transformer world today, right? It's a same scary thought as well as we might get there, right? If you look at the current trajectory of
the implant world just the last year as well. So do we wanna be there and how intrusive do we wanna be I think is a very big question to ask that we quite frankly don't even know the right questions for yet. So, but like I said, in the short term audio feels good.
Bob Ullery (49:41.11)
Yeah, I'm not in love with it.
And I'm not in love with the notion of an implant and only because I'm a little bit of a, even though we spend all day building AI as a business, I'm kind of a doomer when it comes to it, honestly. If it's an implant, what happens when I tell the AI agent, all right, let's turn the session off? And it's like, nope, I'm good. We're gonna leave this going. And you're stuck in inception, in the second layer.
David DeVore (50:07.843)
Yeah.
Bob Ullery (50:07.874)
At least if it's a neural link on the head, like mom and dad can come over after now, like take it off, right? Like, Hey, are you okay? So that kind of worries me. But I think neural link is very interesting. It's the common denominator across all modes. Right? Like some kids can't see some kids can't hear some kids can't, you know, walk or use their hands. There's there's a lot of, let's say hurdles across the board.
Calvin Hoenes (50:08.311)
Definitely a scary thought.
David DeVore (50:11.094)
Meh.
Yeah.
Calvin Hoenes (50:16.139)
For sure.
Bob Ullery (50:37.09)
So right, being able to go multimodal, I think is really the answer to your question, like have it be a little bit agnostic.
Calvin Hoenes (50:41.883)
Mm. Yeah.
David DeVore (50:44.022)
Does like the perfect, does the perfect, so in terms of devices available to us right now, right? And when you think about, you know, Easton drawing, right? Or you think about other environmental stuff, we can get a little ways with, you know, inbound audio and translating what's going on in the environment. But at this, like, I'm wondering, does this, but, you know, video,
capture, real-time video capture goes a long way as well, like using, and then you have the potential to use Vision, Google Vision and other tools to translate that. But does this look like rabbit? Does this, or like human? I know there's also the pendant that sort of consumes all day long. What do you guys think?
Bob Ullery (51:38.298)
I think you'll hopefully it's us, right? I think you're gonna see a lot of optionality out there. Maybe it starts in neurodivergence slash healthcare. It may not, it may be something else that's repurposed and fits well in this world, like the lapel pen or whatever.
David DeVore (51:39.79)
What does a device look like?
David DeVore (51:59.334)
Well, I'm kind of asking the question in terms of
Calvin Hoenes (52:00.167)
Yeah, and it might have different requirements. Sorry. It might have different requirements compared to the, the standard consumer tech that is out there. Right. Just in the, just in how it needs to be. Like if you look at a lot of medical devices and speaking of what you said before with sort of the development skills too, they used to say, like, if you want to be a really good programmer, that doesn't, that gets so well that you don't actually produce bugs, then start in medical development, because that's what definitely where you're going to be trained the hardest to not do that.
and B, also punish the hardest if you do it. But same goes for the actual physical devices. So medical has some of the highest requirements on standards to actually build a device. So we might as well see, because it needs to be more rugged, for example. It might fall down a hundred times more.
David DeVore (52:49.938)
Mm.
Calvin Hoenes (52:50.387)
And there might be a lot of other things, then it might shake a lot more. So, you might need a different type of software to kind of get the shakiness out of it. And so there might be a lot of different requirements where, which is part of the reason why I don't know if the consumer devices will be the first to help in that area, but we will see.
David DeVore (53:08.882)
Mm-hmm. So, I mean, how do we...
Kevin Nuest (53:09.246)
Yeah, Dave, I, my mind kind of goes like with yours, right? Is like, there's a lot of there's a, if there's a proliferation of tech that's out there now that's in people's hands that could then be used and purposed. Like how many more people can you help and how quickly, uh, verse needing to get specialized hardware in, in place, right? And just, we've watched iPhones change people's lives. Even that, uh, aren't, you know, on the spectrum, for example, I saw a, I saw
someone that was in a fender bender car accident and they happened to be deaf and they're typing on their phone and holding up to the police to like make their police report. And that was the fastest way for them to communicate as an example. It's just like, yeah, they have that right in their pocket. And of course you get to extend that turn audio on and do lots of things. So we've seen the power of having computers in our pocket extend that and maybe it is like a little bit more extra device, Calvin, right? Maybe it is like, great, your iPhone, your phone plus.
Calvin Hoenes (54:02.376)
Maybe.
Kevin Nuest (54:04.01)
your earpiece that happens to be the special earpiece and we jammed a whole bunch of tech in it as well. Your phone plus these special version of the Ray-Bans that take the meta cameras pointing out and actually doubles them, right? Like what, you know, maybe it's combinations like that.
Bob Ullery (54:20.474)
peripherals. Yeah, they're probably right there. Like pick the peripheral that meets the need around the accessibility or issue to solve. But ultimately, we all have the supercomputer in our pocket. It's also really exciting that we can run models on those devices now to like, get around the HIPAA issues, right? Like
David DeVore (54:20.758)
sense.
Bob Ullery (54:40.418)
Maybe we don't store any of this data anywhere. It just lives on the device that they're using, right? Give them the ability to offline it, back it up and bring it back in if they get a new phone or whatever, but just sort of, you know, not worry about things like data mining on pieces like this, right? Other than for the things you might need for like UX improvement, but in terms of the individual, I don't know.
Calvin Hoenes (54:54.187)
Hmm.
Calvin Hoenes (55:02.479)
And maybe even we will get to a point where we can do the training. I mean, we're very far away from that, but like, if we get to the point where we can do training and execution on smaller devices, then you're at a point where you can really get to what we said at the beginning of this cast, which is truly personalized inference for, you know, any type of use cases.
David DeVore (55:22.704)
Yeah, amazing. So we're, we're at the top. Thank you guys. This has been an awesome conversation, you know, awesome to awesome to learn about Easton and your life with him and really unpack this subject because I'm excited to work to
Bob Ullery (55:29.358)
Yeah.
David DeVore (55:40.45)
you know, put this into practice over the next few weeks and start to think about prototyping and sort of getting to a demonstrable state of where we can make a difference here. So thank you all.
Bob Ullery (55:59.134)
I appreciate your support and thanks for bringing the subject up. I think it's worth talking about. And so I appreciate you guys being there for us and, and having the convo.
David DeVore (56:08.578)
100%. It ends up being such an awesome arena to solve really interesting problems. Awesome, thank you. So this has been the What If We Could Show with Mesh Mesh, and you all have a good one.
Bob Ullery (56:23.93)
See ya.
Calvin Hoenes (56:25.323)
Bye.