Well, hello, everybody.  Welcome back to another tech doctor podcast.  I am Robert, and it's so nice to be with you here again.  And along with me, of course, as always, is Allison Meloy.  Allison, how are you?  I am doing great on a on a rainy Tuesday in Ohio.  It's been pouring the entire day.  Oh dear, we had that last week, so now you guys are getting it, I guess.  Yeah.  All right.  Well, I don't guess I should say enjoy the rain, but I guess tolerate the rain, huh?  Yeah, we do need it.  So.  Yeah.  Yeah, for sure.  Well, this is going to be a fun and exciting podcast to do, and it's one that you're kind of bringing to us here.  So if you don't mind, I'm just going to turn this over to you and ask you to introduce our guests and guide us through this process, if you will.  I will.  I am so excited to welcome to the Tech Doctor podcast.  Michael Doise and Taylor Arndt from Techopolis Online Solutions.  Michael, I pronounced your name right.  That's a, that's a very meta inside joke.  Because if you remember the 1st time you and I were on a podcast together, I absolutely butchered your last name.  That is true.  Yeah.  Well, Michael and I have you and I have known each other.  God, forever.  Taylor, I've known you less long, but been equally impressed by all that you have done.  So, I would love it if you could each kind of take a little bit of time and go in whatever order you want.  Introduce yourselves and tell the folks listening about what you do.  My name is Michael Doise, as Allison said.  Uh, I have been in the uh, assistive technology and um, accessibility industry, probably going on 16 plus years, which is really crazy to think about at this point.  Um, I've worked for many different agencies and have done development over the years, creating uh, eye accessibility and uh, we, we've transferred that off to a friend.  And uh, working on different projects like VO Starter and different apps.  Um, and, um, done those things over the years, made video starter the 1st voiceover training app.  And uh, I wanted to branch out of just uh, being in the accessibility industry because I feel like if we can make ourselves available to everyone and also bring accessibility forward.  We can improve everywhere.  And so I created my company, Techopolis, LLC.  We were online solutions and we kind of dropped that because it's a very long name.  Um, and so we're, we're just making it techopolis. And we've started making apps for AI.  Uh, and uh, we're going to that a little bit, but uh, the perspective AI apps.  Um, starting with perspective intelligence, but, uh, we have so many things to talk about there, and I'm excited to share, but, uh, I will hand it over to Taylor so she could tell you a bit about herself.  Hello everyone.  I'm Taylor and I am the operations officer at Techopolis.  So my official title is Chief Operations Officer, and I work with Michael extensively, every day, all day, working on our apps, and before I joined Techopolis in 2024, I've been doing accessibility for like 6 to 8 years, I worked for the biggest accessibility firms.  In the world.  And I've also done a lot of screen reader accessibility, testing and speaking, and also.  doing a lot of YouTube and other things.  And I've definitely been interested in AI ever since it came out.  So in 2024.  I joined Techopolis, and It's definitely become a lot bigger than when I started.  Well that is absolutely amazing.  Thank you both so much for those introductions.  Um, I think one great passion that we all have in common here is, is, this wonderful, still fairly new to us, um, technology of AI and the possibilities that it has opened us up for, for us and for our future and the, and the things that we can accomplish, um, efficiently and, and independently. Um, so I, um,  Have, I know, uh, been working with AI for, um, the past couple of years, at least since chat GPT kind of came to prominence.  Um, why don't you both tell me and we will get into discussing your wonderful apps?  Why don't you both kind of tell me a little bit about your journeys with AI?  Honestly, I've been using AI and ChatGPT ever since I could get my hands on it.  I remember in 2022 I found out about it.  And I was using ChatGPT for more things than it was probably capable of.  I mean, I was trying to like, basically have my AI, quote, run my life in the sense of, I was trying to have it, help me with a whole bunch of tasks, even if it wasn't really advanced.  Over the years I really got.  I would say, like, engrossed in the AI ecosystem, and I've been testing and doing a lot of AI work. Not just in Tacopolis, but like personally, and I would say that over the years, well, I have gotten in more advanced.  There's always something to talk about and it seems like the industry moves so fast.  So what I say today, will probably not be relevant next week.  Isn't that the truth?  Yeah, I know now it now it can run our lives, you know.  Um, but that's really cool, like, what, like, tell me in, in those early days, like, what, what were some of the things that you were trying, if, if you don't, if you're okay with disclosing that, what, what were some of the things that you were trying to get it to do that it couldn't really do yet, and, and how was your success level?  I tried to get it to help me write programs in like Python and things that I was having trouble with and the success level was absolutely like one.  Most of the time, it's uh.  It's like, oh, we're going to show these Shakespeare poems.  So when the ChatGPT 1st came around, I'd always run into capacity limits and then show these like, Shakespeare, like poems, like, ChatGPT will be online soon, you know, and I have like the style of Shakespeare.  I bet that was ChatGPT generated on us like, yeah, and I was like, oh, not another black screen.  I also try to have it right, website and web apps, not a good success.  I try to have it right Swift UI for Swift making apps.  Not a good success.  What I was able to do at 1st is like give it a transcript, for example, and have it turn my notes into or turn the transcript into good notes.  That was a good early use case.  But honestly, I, ever since ChatGPT came out.  I have not looked back and I've just become more and more into AI as the years have gone by.  Yes, same here.  I know the, um, the notes were kind of an early, early success for me too.  Michael, what about what about your AI journey?  For me, AI started?  Well, beyond, um, what we know about, uh, Jim, what we, the term is generative AI, and, uh, but as as people who are blind and low vision and visually impaired, however, we want to refer to ourselves.  We all know about AI.  We've we've been using AI since the 1970s in in some state or form.  Um so, you know, we call it machine learning.  We called it, um, you know, all of these different names, but it's basically all the same thing.  Um, when we started getting um, the modern LLM, um, that I think that changed the world.  That is kind of the newest invention computing that has changed how we look at artificial intelligence.  And uh, I think that, you know, my journey started with uh,  Uh, really using OCR and different technologies back on the early phones and and uh, using those things and then it it adapted into 2 more, I want to build AI.  And so I started building.  I had the ideas for the prospective apps back in 2019 or earlier.  And they were going to be just OCR in different classification apps, uh, based on the Apple platforms, and I created a few prototypes by hand. Um, and I've used AI to, you know, um, whenever it came around in 2022, uh, with uh, ChatGPT and GitHub co-pilot to write code and do predictions, which is all great and everything.  Although, you know, I find that predictions kind of get in the way now.  Yeah, I agree.  Um,  But we now have AI that is so smart, it can build things for us.  And and do more than just, you know, be a chat experience. And, um, It's it's amazing to see where this is gone from that.  And and now we're taking a new step into on-device AI.  And that journey, I think, is going to be more important than anything that we're doing with chat GPT or any, you know, Claude or anything else. And I think that's where my journey is now going is to determine what's possible with AI that runs on our devices.  Yeah, I just got a, uh, uh, um, the, uh, EchoVision, a giga, glasses, and they do a lot of AI processing on the glasses themselves, and I have been amazed.  I this morning.  wanted to know about the display of a radio that I have and to what extent the battery was charged, and I ask the glasses to look at the display, locate the area where the battery icon is, and to tell me how charged the battery was.  And the AI on the glasses came back with the exact information that I needed, didn't give me more than I needed or less than I needed, but told me exactly what I asked for.  And that was really cool.  So I'm guessing that maybe in a, in a sense, this is some of the kind of thing that you're talking about, when you say this, this development of the large language models and all is, is a game changer in the sense that it's so much more flexible and powerful than it used to be.  At least that's how it seems to me.  I'm curious if you all have thoughts about that.  The 1st thing I would like to know, um, I'm curious how much they're doing on device for that because, um, it was my understanding that the gigo glasses still relied on a large language model that ran in the cloud.  Um, I would always, I mean, I always want to learn more, and but whenever I spoke to them last summer, I think it was mainly cloud based.  But it would be very interesting if they are using on-device AI, like AI that just completely lives in the glasses.  Um, Because these models are big.  Um, and they take up a lot of space, but, um, Our iPhones and our other devices are starting to be where we can run some of these models on our phones and and where we can, we can take advantage of some of the things that we could do in the cloud on our devices.  Uh, the unfortunate thing is you need a newer iPhone to do these things, but it's pretty incredible.  What, um, you know, on-device models can allow us to do and, uh, um, it, it's, it's only going to get better from here. And that's what I think the, the important thing for us to realize is, you know, everybody says, well, if I can't do it now, it should, you know, it's not a big thing to even look at, but I always challenge people.  If it's doing this now, if we can do some of these things now.  2 years later, what will it do?  And that I think is the most incredible thing to think about as people that are into technology.  What can this technology do?  And then where was it 2 years ago?  Let's look at what Taylor was talking about with, uh, ChatGPT and other AIs in 2022.  And what we can do now.  Yeah, exactly.  And I think I misspoke a little bit or it was a little bit misleading.  A giga is doing a lot of the lot of it in the cloud.  Um, but it's interesting.  They're doing it, you know, between the glasses and the cloud, and you don't even have to have a phone involved, if you don't want to.  And that has pros and cons, but it's an interesting move in that direction.  Right, exactly.  There's so many options and opportunities.  Um, we, we've talked about uh, for our apps and we'll, we'll talk about those in a minute.  You know, what, what, and pair of glasses in our apps would look like, and it's very interesting, um, with wearables and things like that.  So, um, the, it's all about the journey.  It's all about the AI journey, because what we're able to do now is is going so much faster compared to where we were when we started.  Yeah, it's like week to week and month to month almost, there is new, um, there's new capabilities.  I mean, I know that I, uh, I tried as someone who is not a coder, um, by trade to use, uh, chat GPT to teach me some basic things even a year or 2 ago and I didn't get very far.  But now I'm developing an app in X code, just using clawed on the website.  Um, and and just learning Xcode, um, along with it, and it's just, it, it's come so far. So far in, in just that time.  And there's even a coding assistant in X code, the accessibility has a bit to be desired.  But you could use Claude completely in Xcode as well.  Yeah, I I'd heard that and it keeps like reminding me of that.  It's like, oh, we could do this in Xcode that I had read, and it may have been you or Taylor who talked about it, that, um, the accessibility was a little wonky, so I've just been pasting into, into Xcode from, from the website.  I like that though, because I can just look at all the lines of code.  before I paste it anything. Right. And even sort of understand some of it because I've looked at enough code over the years that I kind of know what the, what the gist is.  Um, But half speaking of apps, by gosh, I have been playing, uh, with perspective intelligence in more recently, um, perspective.  Transcribed.  So tell me a little bit about, you know, what the ideas for those and kind of what what you would like, someone who's never heard of these to know about, what about what they do.  Last summer, Apple released their Apple foundation models framework.  And it was part of iOS and all the 26 updates.  And what that is, is a built in language model that runs on every iPhone that supports Apple Intelligence.  And we as developers have access to that.  And and,  So the insane, absolutely bonkers thing about this is.  It's so easy to learn.  is 3 lines of code.  3 lines.  You got an LLM and 3 lines.  How can you beat that efficiency?  How, can you beat how effective that is to implement into an app?  Apple has taken what they've done well over the years and they applied it to an LLM, and yet people still, Say that Apple Intelligence isn't great.  They say it's not good.  However, it is the fastest.  A large language model in any app you'll ever find.  The limitation is, yes, it doesn't hold a lot of context and you can only use it once at a time.  But.  It's there.  It's built in.  You can even have it run tools.  You can give it context about something you want to do.  And you can have it look at your calendar and give you information about that.  You can have a look at your reminders.  You could do all those things.  You can build it in what are called tools, and it can understand anything, even information from a website.  You could have it go and look at, oh.  Um, if, let's say top take tidbits at an RSS feed.  You could say, give me all the AI stories from top tech tidbits.  It would look at the RSS feet, give you those, and It would summarize.  And that's built in.  That's not running on the cloud.  That's not sending your data anywhere.  It's all built in.  So I thought.  Well,  Why don't we build an app around that because.  Um,  Apple.  Has not showcased anything.  I mean they have writing tools and other Apple intelligence features, but they don't showcase what they've done.  They haven't showcased how amazing the Apple intelligence features can be if you use correctly.  And so we, our goal was to set out to do that, to build an app that would run, um, Apple intelligence and foundation models to do things.  And we built that.  It was wonderful.  But, um, and we've been improving on that core tenant ever since November.  But that quickly became, okay.  Well, we have this thing, it's working.  What else can we do?  And the other thing is, how do we make things work for people that don't have iPhones with that technology?  So, we added, um, other models that you could use that may not have as much, um, may not be as smart as Apple foundation models.  And we had some shaky releases.  Uh, the models would crash because they would still take up too much memory on older phones. Uh, so we said, well, let's try other ones and, you know, it was a research project.  Um, and so we've made it where there are some models that I'm, I believe, will work pretty well on almost any phone. And then we've added image capability.  So you could describe images in chat, and we have a whole vision book that has those old perspective AI features like OCR and classification.  We even have a, I would say one of the coolest cool color identifiers.  But we said, you know, there's all this on-device AI out there.  What if we could put it in the app? and we did.  And so you could chat with almost any model that's out there that supports your device.  And I think my favorite feature that we've added, you know, in the past 3 weeks is video mode.  It is a live video mode that will read whatever it sees in front of the camera as you do things with your phone.  And I've been told by so many people.  Video mode is the best I've ever seen.  For AI-based video mode.  It may not be as good a quality as as Gemini or some of the others, but the difference is, uh, it's all on device and it's getting better.  Uh, you know, we can't prompt it.  We can't do certain things yet, but it's just amazing when people tell me, oh yeah, I went through my, um, uh, uh, spices in my in my in my cabinet with this thing.  And it was right on the dot every time.  Uh, I think the only time where there was a problem was something was upside down.  And just hearing that feedback, I built this mode is like, let's see what's possible. And, uh, there's a lot more that has to be done to test.  But it's just incredible.  to see where we've come from and where we are now in since November with this app.  That is pretty amazing.  I was playing with it.  Um, Just today, actually, and I, I discovered a video mode.  I was like, oh, this is great because I was just able to kind of wave it around my living room and get like, beautiful descriptions of like, you know, our, at one point I was on purpose, pointing it at the ceilings, we have nice beams.  Um, in our ceiling and they were like describing the woodwork and describing um, Dusty the dog as I was pointing to him and, um, I did use the, um, I'll have to use video mode for checking my mail next time.  Um, because I was using uh, um, the text recognition and it was pretty awesome. And I was able to find out that what I had was just a, another pre-approval for a credit card and uh, And to realize, oh, that's junk.  don't need to.  I don't need to keep that.  But yeah, like, and it was fast, and the voice, the quality, what really, um, what really got me, was the quality of, of the voices that are available, and it seems like they're very small to download.  Um, but they sound darn near to human.  Mm-hmm.  I mean, I was just so impressed by that.  We're looking to customize those to make them better.  I'm finding that some numbers may not read perfectly with those, but you know, always room for improvement. And that's, that's our goal with this, is to take what's there out there on the market and say, how can we make it better?  Exactly.  Exactly.  Taylor, perspective, transcribe, literally just came out, and I was happy because as a, uh, as an all access subscriber, um, for a perspective intelligence, I already have a subscription now to, uh, to perspective transcribe.  So that's a wonderful thing you've done, um, for, for your customers.  Um, but tell me, uh, a little bit about, um, Michael, this is your project that I've read some, you know, about your reasons, but for, for folks who don't follow you guys on social media like I do.  Taylor, tell me kind of about, um, about why perspective intelligence is important to you and why you developed it.  I love the app and honestly, I had the idea.  I don't know how many months ago.  There's so many transcription apps on the market.  But if you remember last year, there was these pins, like the B devices and the plods.  I like that.  However, most of them now require a subscription, and most of them will send your data to Neverland.  You don't know where it's going.  And, I transcribe a lot.  I mean, I accidentally had a recording going for 10 plus hours.  Could you imagine?  Could you imagine that on a cloud service and then you get billed per the minute potentially?  That can get expensive.  I know, and the thing is, is that for, I, I need transcription because oftentimes I've got a lot of thoughts and trying to, you know, write them down or other things, I just get to sidetrack because of my ADHD and other things.  I'm like, why don't I dictate them?  But then of course I think about, oh, well, the dictation, you know, there's not a lot of good options.  Then I'm like, okay, well, I'm going to make my own. And so I really have one core principle, and I really want people to know that perspective transcribe is not just your other transcription app that, you know, is on your phone.  It's actually very different.  There's a few different capability to actually make it pretty different and I tried to mimic the B and the plot devices because trust me, those were my lifelines, especially the B. But I didn't want to have to pay for hardware, other things.  I didn't have to pay for subscriptions.  So prospect to transcribe.  Um, it, it starts off where you can use a lot of it for free.  There's a few paid features.  I'll get into that in a 2nd.  But this is really great for people who need to dictate actually now I write all my substack posts with that.  I just start dictating and then I have AI polish and that's my substack.  I start, you know, yammering like, okay, I need this, this, this, this, this, and I need, you know, talk, sometimes for 20 minutes straight and then have it make a post.  I also use it for meetings.  Michael and I have a lot of meetings.  You know, being executives at a small tech startup.  We have a lot of meeting and sometimes I want to be able to keep track of all of this stuff so I record it.  The transcription is fully on device.  It uses the Apple transcription by default.  And then there's something called Whisper, which is like, I've like open AIs model that they made, but there's also uh, local variants of it.  And a lot of other apps such you guys may have heard of like MacWhisperer others use it.  And we use the same technology, but it will still run on your phone.  Um, it's all on device.  There's nothing that goes to the cloud, and then the other feature, we have a summary feature, where when you're done transcribing, it will make a summary, and it uses the Apple Foundation models, like we talked about intelligence, and then it also uses other models for people who don't have those on their phone.  Like if they don't have the newest stuff.  So it works for anybody, not just people with the foundation models.  And then also there's another feature too that is, uh, this is a paid, all access feature.  This one is really cool because let's say you just want walked out of a meeting, but you want like action items or tasks.  There's a way under this summary where you can go action items and you can look at all the action items that the transcript has gotten out for you and you can check it off like reminders.  This is actually something that the bee had and I really loved it.  And I really want to be replicated here.  Uh, I will say that, um, it does actually work extremely well and I've got a lot of other features planned.  So hopefully, that is uh, usable for everybody, and I also have another differentiating feature.  This is actually the biggest differentiator.