Conversations on Applied AI - Stories from Experts in Artificial Intelligence

Scott Brown - How Machine Learning Is Revolutionizing Veterinary Medicine

April 12, 2022 Justin Grammens Season 2 Episode 6
Conversations on Applied AI - Stories from Experts in Artificial Intelligence
Scott Brown - How Machine Learning Is Revolutionizing Veterinary Medicine
Show Notes Transcript

The conversation this week is with Scott Brown. Scott is a business focus Senior Technology executive with 25 years of experience in transforming organizations, solving complex problems, and building globally distributed systems. He drives cross-functional strategy development, continuous improvement and is committed to growing talent and developing the next generation of leaders. He's currently the CEO of Transfur, Transfur makes it easy to request, send, manage and review veterinary medical records. Unlike other solutions in the market, Transfur leverages many advanced technologies like artificial intelligence, to extract, analyze and present clinically relevant information from an animal's medical records with no software to install. Scott holds a BS in aerospace engineering and as a mentor at the Minnesota emerging software advisory where he provides sea level pro bono mentoring to emerging software companies.

If you are interested in learning about how AI is being applied across multiple industries, be sure to join us at a future AppliedAI Monthly meetup and help support us so we can make future Emerging Technologies North non-profit events!

Resources and Topics Mentioned in this Episode


Your host,
Justin Grammens

Scott Brown  0:00  

We got to do this different, you know, this is this is the equivalent of like when you're in math class, and you have a paper and pencil, and somebody gives you a calculator, or like, you know, before Google existed, you know, now you want to know something, you type it into Google or Bing or whatever. And so we think that this technology, our application of AI is going to revolutionize the way that records are reviewed with their batch effects.

AI Announcer  0:27  

Welcome to the Conversations on Applied AI podcast where Justin Grammens and the team at emerging technologies North talk with experts in the fields of artificial intelligence and deep learning.  In each episode, we cut through the hype and dive into how these technologies are being applied to real world problems today. We hope that you find this episode educational and applicable to your industry and connect with us to learn more about our organization at applied Enjoy.

Justin Grammens  0:58  

Welcome everyone to the conversations on applied AI Podcast. Today we're talking with Scott Brown. Scott is a business focus Senior Technology executive with 25 years of experience in transforming organizations, solving complex problems and building globally distributed systems. He drives cross functional strategy development, continuous improvement and is committed to growing talent and developing the next generation of leaders. He's currently the CEO of transfer, transfer makes it easy to request, send, manage and review veterinary medical records. Unlike other solutions in the market transfer leverages many advanced technologies like artificial intelligence, to extract analyze and present clinically relevant information from an animal's medical records with no software to install. Scott holds a BS in aerospace engineering and as a mentor at the Minnesota emerging software advisory where he provides sea level pro bono mentoring to emerging software companies. Thanks, Scott, for being on the program today. 

Scott Brown  1:50  

Thanks for having me. 

Justin Grammens  1:52  

Awesome. Great. Well, you're you're speaking our language, when you're talking about artificial intelligence, maybe can give us a little bit more context, I guess, in terms of where you got to where you are today. 

Scott Brown  2:01  

I began my career in technology, what you said over 25 years ago, is a very different world back then all my projects were client server based. I began working at Accenture, installing PeopleSoft, which is an ERP, or enterprise resource planning. And that was back in the day when large companies didn't have automation to manage their back office systems are back office policies. And so I look back on that, have you remember the y2k bug, but everybody was the year going from 1999 to 2000. And people were predicting the end of the world. And that was really before the internet area before the proliferation of the internet. And you started off Boom, around that time, until was a very different world, from there, moved into E commerce, looked at Digital River for a few years, and then spent at Medtronic building out their Connected Care System, Accenture, client, server based technology, and then web based technology at Digital River and an electronic. And then in Medtronic, it kind of shifted from web based technology to a mobile based technology. And with a connected care system, we were remotely monitoring patients, so patients with implantable devices could continue to see their doctors, you know, our use case was in older adults, with a heart condition, they get an implantable device, they lived in Minnesota, but then they would winter in Florida, and wanted to see their doctor so they could go down to Florida during the winter, interrogate their device, the doctor to get the data, continued to provide care remotely, you know, 1000 miles or so. And then XRS, you know, transforming the company from a hardware based system to mobile. And since then, you know, a lot of what I've done has been focused on mobile. And then most recently, my experience has kind of shifted towards artificial intelligence. So reveal x, we're using computer vision and artificial intelligence to prevent diabetic limb complications. So looking at the bottoms of feet, particularly with diabetics who had peripheral neuropathy, which meant they couldn't feel their foot. And we could look at the thermal imaging of their feet and determine if a wound was about form. Since they couldn't feel their feet, you know, you and I can feel if we're worried a hole in our foot flight, if you have peripheral neuropathy, get these hotspots and you could walk on your show and often you take off your sock and there's blood there. And so the computer vision was able to look at the thermal patterns, like identify the look at the thermal patterns and you could tell before it became an issue, you know, it's a multi billion dollar problem within healthcare for patient variable to prevent that, you're talking to 1000s of dollars. If they get a wound on their foot, it becomes very hard to treat because they can't feel it. They keep re injuring it. And ultimately, it leads to an amputation. And so, amputation costs hundreds of 1000s of dollars. So if you can prevent it early on, you can say tremendous money in the healthcare system. So and now working on Transfur, where we are taking veterinary records, extracting all the clinical content from those records. So the thing about veterinary medicine is that you don't have HIPAA. And you also don't have or ability standards, like HL seven. So a lot of the information is shared through PDFs. And the problem with PDFs is to review them, they're hard to search, you know, it's a it's a document, so kind of reviewing that. And if you have bills or sick animal and you're trying to extract, see the see the animal in emergency situation, or even the non billable time, you know, after hours, when you're preparing for a visit, it takes a long time. Sure, what we're working on, is being able to extract that content, understand the presenting complaint, or the reason for visit, and then pulling out just the necessary information, which is referred to as the pertinent history. So the veterinarians will receive the appropriate information just in time for that visit. Oh, that sounds huge. And I think, I guess I don't even know what the population of the animal like the the animal population in the United States, what's gonna be huge, it's got there's got to be we're out massive number of animals. The pandemic has driven the a lot of adoption of cannibals, here locally in Minneapolis. I don't know if you saw but just before Christmas time, they ran out animals at the animal shelter. And so they were on their Facebook page while on the door. And then they posted on their Facebook page that they're like, don't even come in, we don't have any more animals left. Within the United States, the households with animals went up from 67% to 70%. And there are about 70 million dogs and about 66 million cats within the United States, there has been just unprecedented demand with the veterinary space because of that, so you have this increase in animal ownership. And when with COVID, you have all these preventive procedures in place to be able to sterilize the room of people waiting in their cars, more time in between patients. So you have this decrease in efficiency. So the demand within the veterinary space is absolutely unprecedented. I recently got a puppy. And so I called my local vet to get an in for wellness check in at a scheduled three months out. Yeah, it's crazy to be able to get in so they're looking for every way to be more efficient. So when you look at like a veterinary specialist, the veterinary specialist isn't familiar with your animal. So your animal goes to like a primary. So like you go to a primary doctor, you have to have surgery, if you would go to like an orthopedic surgeon, it's the same thing. In the animal space, you'd go to an orthopedic surgeon, you go to an oncologist, someone, that doctor isn't familiar with your animals, so the records have to be transferred. So when they transfer the records, they send a PDF. And so if you have an older sick animal that that's that can be 100 plus pages to review. All that review is non billable time, so they get paid toward you read consider procedural care within veterinary medicine. And it's typically paid by the owner. So when your animal goes in, you send them the records, the vet spends 20 to 30 minutes looking through it preparing for that evaluation, that's all it's only don't get paid for they're just expected to know this. So in a specialty practice 60 plus percent of their business is going to be referral. So that means they're spending about two to three hours per day, typically, from their free time reviewing these records preparing for the next day's visit. And this is time they don't get paid for it impacts of quality wife and my other two co founders are veterinary students. And they grew up with two veterinary parents who were both surgeons themselves. And so they saw a lot of missed opportunities, Miss dinners and just a lot of time spent at work that could have been spent, you know, at home kind of enjoying each other. And so here they, when they get to bed school, they're like, there has to be a different, there has to be a better way. We've got to be different. They had submitted their idea to tech transfer center at the University of Wisconsin, which is worth they had formed this venture studio called foresee venture studio. And their concept was the first one selected for funding and for the launch through the program. And I joined in July to help them organize and to create a business around this concept. 

Justin Grammens  9:44  

That's awesome. Yeah, I was gonna ask you sort of how long have you been at the company and where did the idea come from? So it sort of came out of a personal pain point, I guess right from from from people that are veterinaries in the space. 

Scott Brown  9:57  

Yeah, so all my background is health Care Technology, I've been building teams and scaling systems and, you know, really focused on technology. My other two co founders are very deep in the veterinary space. And I think our skills and our experience complement each other very well. So they grew up with veterinary parents, and they've been around animals their entire lives. They've helped their parents review records, they've helped their parents with their practice. And so when they went to veterinary school, they, you know, they lived in this world, they understood it intimately. And so here, you know, this is very personal to that. And so when they, when they get to school, like, we got to do this different, you know, this is this is the equivalent of like, when you're in math class, and you have a paper pencil, and somebody gives you a calculator, or like, you know, before Google existed, think about how you found information, you'd ask what's your library, on the library? Now, you want to know, something, you type it into Google or Bing, or whatever. And so we think that this technology, our application of AI is going to revolutionize the way that records are reviewed within veterinary that's.

Justin Grammens  11:10  

Absolutely, it sounds like it. I mean, if it's a if it's a huge time saver, this feels like that's ripe opportunity for applying artificial intelligence, right? I mean, who was who likes to do them all the mundane work? 

Scott Brown  11:22  

Well, you think about it. So you go to work, and say you work eight hours a day, then let's say you have three hours of homework each night that you don't get paid for, right? You don't get paid for that you get you get paid to do your job, but then you're expected to be able to do all these other things. So if I came to you with solutions that, hey, you could get back most of that time. I think most people would take that and say, Hey, we're gonna cut that three hours down to 30 minutes. 30 minutes is a lot more manageable than spending three hours a day reviewing records. 

Justin Grammens  11:52  

Yeah, for sure. And like I said, If there, you're not even getting paid for it. So it just sort of it is purely wasted time, I was gonna delve in a little bit. It's kind of a question off the cuff here. But you know, having non technical co founders do sometimes you get the sense that I don't even know how to phrase this, per se, but there's a lot of hype around AI. Right. And so people are like, Well, geez, you know, it should be able to just plug some data in, and I should be able to get the answer pretty quickly. Hmm. Again, there's just there's a lot of stories around AI doing some amazing things. But a lot of it is hype, right? A lot of it like isn't really we're still very, very far away from this from this general artificial intelligence. Right. It's very, very narrow space stuff. Have you have you? Have you seen that? Again, I don't want to pick on your co founders or anything like that. But you know, I do you sense that I guess and people that you talk with, maybe they think it's too easy, too hard, or don't have a grasp of where we are in this space? 

Scott Brown  12:41  

Yeah, I think that's an excellent question. My co founders are very deep in the veterinary space, right. So when it comes to defining the problems, understanding the market, they are intimately involved in this. So they didn't say, Hey, we should apply AI to that. So we need to find a way we had the solution, we need to find a way to apply this is very much the other way around. And they understood market, they understood the need, they had some concepts on how to address this, they knew what needed to be done. It didn't know how it would get gone. And so is in look, all the informations there. We know what we're looking for, there's got to be a way to marry these together to present these results to us, right? We can go into Google and type of word and get the results back. You can take you know, you've got search, we understand clinically what we're looking for. And we want to know the the health history of this animal, is this animal, even healthy enough for surgery? Okay, well, once we understand that, when did these concerns present themselves? How long has it been happening? What else have they try? Right? These are all normal questions. So they understand it clinically. And that's, that's not my area of expertise, your knowledge. And I always say a problem well defined is a problem half solved. And what they were able to do was clearly defined this problem clearly understand what needed to be done. And so in a clinical setting, they say, this is how we operate. This is what we get this is this, but we want it to, for me to come in to this clearly defined problem and understand how it's going to be used. Well, now, I can apply a solution. So when we originally talked, we weren't thinking AI. Yeah, right. Usually when we talk, we said, hey, we had this all we want to be able to do a search on the content that's in here. And as I got into this and said, well, we want to know this and this is related this into these things are this then we want to do this. And understanding kind of that the logic and the concept is if it's too complex to write an algorithm around, but there are a lot of patterns based on that. So this is a great app with the application for for AI. So now, technically, we want to be able to extract that content, we want to search it, filter it, work with it, and so on. And as we get the content, and as we work with these different presenting complaints, or reasons for visit, and we understand how it's being used, we can continue to improve the system and being able to write logic around that would be next to impossible. Well, that, like, that's our ultimate goal you want, this is the information you want, and you want it just in time. So if this is the presenting complaint, this is what we're looking for. So tell me if you find this record, and being able to take that and make it searchable, is pretty easy. But then there's all the main effort of being able to search through that. So the A being able to apply the AI models to extract that kind of take one step further to be able to extract based on that presenting complaint, we think is a real differentiator market. And so that's where I came in. And I think, like I said before, we've got complementary skills. 

Justin Grammens  16:06  

For sure. I mean, I start up that I'm involved in is called Captivation, kind of similar to you, I have a co founder that as a subject matter expert, and and he's an expert in presentation skills, leadership training, coaching, and he came to me saying, Hey, I'm reviewing all this video of people as they present. And how can I like, again, for free, why I'm sitting here with a pencil on a piece of paper, when we should be using a calculator for this. And so we went, we sort of started the journey of allowing people to record their presentations without actually needing to have a teacher there. But now they can start getting data about it. Right? So are you looking at the camera or not? Are you saying arms and eyes a lot? What sort of vocabulary words are using? Or is it isn't it? Is it uplifting? You know? Or is it you know, down and detrimental? And all these other types of things over to you, if you just slow down your presentation skills get better? And so yeah, I was kind of in that same situation. I just wanted it, you know, pick your brain a little bit on that. But yeah, he basically had a problem. And it was one more or less like, like you it's like I'm just I'm spending a lot of time here doing stuff that we should have technology be able to solve it. Not saying AI or machine learning, or any of that stuff was the right answer. But, you know, there's just a lot of wasted time. So definitely, definitely feels good to pair up with people like that, that know that industry. And then I think, you know, as a technologist, I've been doing this writing software and code, you know, since the since the mid to late 90s. So I remember a hint on PeopleSoft and all that other stuff, those early days. But it feels good to actually be able to do to actually apply technology and solve solve the world's problems in whatever way you can. 

Scott Brown  17:40  

Yeah, I think, you know, a lot of these problems, you need collaboration, you need the subject matter expert, and you need someone who's done this thing and built these systems that solves problems, tech, technically. So you've got a solution for a real problem that exists. And I think that within healthcare, know, you have a lot of people with good intent, you have patients who want to help Wait, they see a problem, they want to help, but they don't necessarily understand the healthcare system, or how pay how care is delivered, and so on. So, you know, within within the healthcare space, I think it's critical to have a clinician involved and so, you know, at a hypnosis suite, or a chief medical officer, we can cheat radics officer, we have very strong clinical team. And typically, they would come in and, you know, close the deal. They were the credibility of the company. You know, it's the same thing here. The veterinarians aren't responding to me, you know, even though I'm the CEO, they're responding to, to my two co founders who are med students, right? They you tell it peer to peer. And when they have questions about how things work, and they're, you know, they're their peers. They're, you know what, I think that that's critical. So if I were to try to try to solve this problem by myself, it wouldn't work. It knew that subject matter expertise. And even though I own a dog, that may be a winner. It's very different. actually providing care and working in that system, understanding the practice management system and understanding medical records and how exams are things like that. You need the subject matter expertise. And then you also need the expertise around how to build and scale systems. What do we need to do? And then how do we need to do it? That's where having a successful team brings both those components together. 

Justin Grammens  19:33  

Well, it sounds like you have a really good worldview of how to build software, which kind of leads into your experience working as a mentor as well, too. Right. at Mesa. I forget I forget how they pronounce that. I mean,  yeah, Mesa. Yeah. Are you working with any other companies right now? Well, hypnosis, right. I guess. Are you a mentor to them? Or were you there? No, I'm okay. What's this? What was the mentor there? So, fishing. So that's it.

Scott Brown  20:00  

It has Yeah, they're doing amazing things. Matt is a great leader. If you're into fishing, check out Omnia II, they are doing very cool stuff. So Matt's previous company was analyzing lakes. And so through these lakes, he could understand like the species and what was working and so on. And they sold that company. I don't remember, I don't remember who that some large fishing company, but they had all this data, all the analytics and the fishing data and stuff like that is all publicly available data, they created it and made it available into now. You can pick a lake in the United States, they'll tell you what species is there. They'll tell you what fishing lure to use, they'll sell you the the attack all and the rods and all that. And so you want to go up north fishing and have a good time. They'll tell you what's working. They've got people out there, they have a whole social aspect where you can put in your fishing reports talk about the Luers make recommendations. It's very cool. Huge fan and Matt huge fan of Omnia. I'm really happy to be their mentor. Yeah, they're doing super cool things. 

Justin Grammens  21:09  

Cool. Yeah, no, they've been a little bit of a success story here. I've been seeing them grow over the years. And they they raised a bunch of money. Is that true as well? 

Scott Brown  21:18  

Yeah, they put in a ton of work. They're doing a great job, just knocking it out of the park. And I think they have so much potential, like they have these huge players coming in. Courting them, last year, this summer, they gave away a boat, they've got these phenomenal promotions going on. So again, if you're if you're a fisherman, even if you're not, so I fish maybe once or twice a year. And so when I go, I want to catch something, and I want to have a good time. And I've no idea what I'm doing. I just grabbed my rod and go. And you know, the Omnia, you pick the lake that you're going to be and see what's working get paid, increase your chances of catching something. And so, you know, or a part time fisherman like me, it's it's a perfect solution. 

Justin Grammens  22:04  

For sure. Yeah, I just I had to connect the dots. I actually worked with Dan Wick many, many years ago, back when he was at a company called red stamp. So yeah, he looks like he's the he I knew he was one of the C's over there. And looks like he's the CTO. So it's been good, really good. And actually speaking of funding, I mean, how are you guys funded it you can't kind of came out of this incubator. 

Scott Brown  22:26  

So technically, it was a venture studio, as opposed to an incubator. So the difference is that, so a venture studio, RC venture studio was a collaboration between Worf, which would be the tech transfer center at the University of Wisconsin, and high alpha. And so high alpha is a is a venture studio. So high alpha innovations works with companies and universities to build venture studios. What worth saw was that they had all these concepts and ideas with the students, that IP within the university, they have a massive endowment from IP that they've licensed past. And they wanted to support their students in their faculty, you know, more than just licensing IP, but they wanted to support them in I think, a really forward looking way. So they created this Varsity Venture studio had a call for concepts. And so faculty and students could submit these concepts. And through this process, I think that you know, 120 150 different concepts that were submitted. And through High Alpha's process, they were able to work with the students and the faculty on these concepts narrowed down to about 20 that were promising. Although those 20 They went through some due diligence, it's a market analysis and so on in the narrowed that down to a group of about four that went through what they call a spring week. And in this sprint week, they work with the team pretty much 24/7 for a week, to look at the competition to develop a concept and market analysis to essentially create a pitch based on the business analysis and the opportunity. And from there, they do the report out. And transfer was the first concept that was selected for support through that process. So a venture studio has the tools to be able to curate these ideas or vet these ideas into what has the most potential and then from there, work with the entity, the business or the university to develop the concepts. And then they can choose to fund all of them or on one of them or whatever they want to do with it. They don't have a horse in this race, but what their strength is is to build businesses. So this point of the tech transfer center is developing ideas and concepts. The venture studio is helping curate vet and determine the highest potential concepts from that cohort. And then work provided funding and launch support. Kyle for innovation has provided resource support as well. And so that's kind of the difference from an incubator. And so we that funding we received, we called our precede funding to get us started. 

Justin Grammens  25:12  

Okay, and you haven't needed to take any more along the way. 

Scott Brown  25:15  

Not yet. Yet. Take us a little while, we just kicked off a bridge round. And so we're just starting to talk with some more investors to raise a little bit more money to get us across the threshold to bring an enterprise product to market. 

Justin Grammens  25:30  

Yeah, sure. Are you guys hiring today? 

Scott Brown  25:32  

We're looking at a few strategic hires. So we're looking at reinforcing our clinical race. Awesome. We've been talking with some veterinarians, also potentially bringing on a Chief Veterinary Officer. And so you've got a few opportunities that we're looking at, to really reinforce our clinical credibility to understand the clinical use cases helped develop the product, and so on. 

Justin Grammens  25:56  

Cool. I'll have liner notes and stuff like that as a part of all the podcasts, I'll be sure to include links to you guys. 

Scott Brown  26:02  

Yeah, that'd be great. 

Justin Grammens  26:03  

You know, I and this varsity venture. So this is this out of Wisconsin, says it's not local here, right. 

Scott Brown  26:09  

Yeah. So that didn't Madison was born out of the University of Wisconsin Madison. And so varsity Metro studio is part of the university as part of worth ventures in Madison. 

Justin Grammens  26:19  

Yeah, I don't, I don't know if there's other you know, there's a company called Great North labs familiar at all here, you know, I It feels like we could do better here in the Twin Cities around some of this stuff. Not sure if you know, of very, very, very many. There's also another guy named Nick teats here in the in the cities, Kyle, he runs a thing called IoT studios, it's very similar to that. But again, what you are reminding me of is, is people have these great ideas and they need they need mentorship, right? They need help, and you could sit there and do Startup Week, or Startup Weekend and stuff like that all day long. But, you know, you need a venture company to come in and actually say, that is a product that can actually go to market, right. And a they're, you know, a they're experienced with it. So they, you know, kind of weed out a lot of these other ones. And then sort of like, again, like put all your resources behind, you know, the one that you think is going to have the most impact for you guys, you know, as you were talking, yeah, you know, I'm my mind initially went to pets, but I mean, I'm guessing there's just general farm animals that might, you know, be able to be benefited from this technology, right. 

Scott Brown  27:19  

Our products are targeted in what they refer to as companion animals. So dogs and cat. So if you have an iguana, or parrot, that'd be an exotic animal, you have a low Lazarus, or a zebra that would be a zoo algae or, you know, animal Zoo. And then there are large animals like cows, horses, things like that. So they're managed in very different ways. So you manage a herd of like, large and so you bought a lot of cows, or a lot of horses, you have food producing animals, you pigs. And so those are managed in very different ways. So if you have a dog that breaks his leg, you want that dog to heal, and to get better. And so you're going to look for care, let's find a way to get that animal out of pain. That's not necessarily the case with other categories of animals, their food producing, maybe you make food out of them, rather than fix them in different ways to approach it. So we're we're targeting the companion animal space. So again, kind of dogs and cats and animals that you have around your house, and kind of a small animal veterinarians in the specifically the specialist, so surgery, oncology, neurology, internal medicine, medicine. 

Justin Grammens  28:36  

Makes sense. I mean, with the with the population numbers that you're telling me, you still have a huge amount of animals, you can deal with just only focusing on that area. And it's, it's also probably the high dollar ones, too, like you said, it's like, if a cow has cancer, there's, they're not going to do anything about it. 

Scott Brown  28:51  

They may not, you know, that that time really is ever focused on because they they provide care in a different ways you immunize the herd, you will provide care to the herd, and so on. And so it's not kind of the individual medicine that you practice, companion animals, they're all learning with the herd. Typically, it's a company as opposed to a pet owner.

Justin Grammens  29:14  

Well, one of the questions I do like to ask people is, and I think you've touched on it fair amount, but, you know, how do you define artificial intelligence? 

Scott Brown  29:21  

I think of artificial intelligence as a system, being aware of it to being able to assess, and then use probability to achieve that goal. So being able to create these models, where it recognizes its situation or its surroundings, understands different options, then picks the best approach to, to do something to achieve a goal to be able to accomplish a task, and so on. We're gonna argue, like an algorithm will if then statements, and kind of like a decision theory to be able to get a point. You know, I think what we're trying to do, we're taking very complex logic and, you know, it shouldn't be too difficult to program and understanding all these nuances and interactions, and then being able to extract that and then present all that's necessary. We're using natural language processing, which is essentially extracting the text, and then applying natural language understanding. So you can take that text and then understand the context of it, but meaning behind it understand, you know, not just the terms, but the context of those terms. And then understanding the meaning of that, in the extracting, you're printing that executive summary. 

Justin Grammens  30:37  

Are you guys building your own NLP models? And everything? Is there stuff you can use? It's already pre built stuff off the shelf? Are you kind of really starting from ground zero? 

Scott Brown  30:46  

Yeah, and, you know, there are a lot of tools out there today, you know, like, it's constantly developing. But we are, you know, we use a lot we use as much as we can off the shelf, we do a lot of experimentation, different models and different tools. You know, when we began, we tried a couple different tech stacks to find good results, good processing times, this last week, we, we applied a new model, it completely crushed the servers, in a relay for signs, you know, the results went way up, and just crushed the servers. And so you don't know until you try. And so now we know. And so to optimize the bottles, we look for it, you know, we can always throw more hardware at it. But in my experience, you know, even at reveal x, we began with some assumptions. And you know, there are things like TensorFlow, right? It been out there for a long time, a lot of tools, a lot of usage. It didn't work for us, you know, it worked, it worked fine. But there are a couple of other things that we use that we got 10x, the performance 10x you know, the results, and they're easier to train and more responsive, there are a lot of resources that I follow, just to kind of stay on, on top of things. And so I think the first thing is just to understand your use case, like what problem are you trying to solve? What are you trying to do on the tools out there, lots of open source stuff, you know, new tools and bottles coming out all the time, I think the first thing is to understand your problem, and then evaluate the solutions that are out there. And then try a couple of them. See how they work? 

Justin Grammens  32:25  

That's awesome. No, that's great. That's great advice. Because otherwise, you're just you're basically a hammer walking around looking for a nail. Right? You're just you're starting with the technology first. And you really need to understand the other side of it. Yeah, I just got a couple more questions here as we start to maybe think about winding down. But I mean, as you got in as you think back 20, some years or so like, what are some advice, I guess, advice you would give to people, whether it be getting into the AI space or technology in general, you know, are there classes that you've seen, or books or conferences or places that you found, as people are exploring new things and technology? 

Scott Brown  32:58  

Like I said at the beginning, you know, my career began in client server, then move to the web, and then move the mobile. And now the AI, I think everything is going to be ai ai is going to be absolutely everywhere. 10, 15 years ago, it was hard to get your hands on AI tools, right, the processing time building models, universities, and large companies had access to that, but not kind of everyday programmers. You look at the capabilities that are AWS, and Google and IBM and Microsoft, you know, it's incredible. And so very powerful tools. I would say that any programmer out there, that doesn't have experience in AI, pick up a project, find something to work on, I think I think it's gonna be every aspect of our life. I've always been curious, and low on problems and kind of diving into these things. And so if you go in there a lot of resources out there where you can find projects, you know, for example, like if you're, if you're an AWS user, they have deep lense, which is a little camera with, with artificial intelligence, and they have a couple projects out there to do in one of them. I don't know if it's still out there. But one of the projects was went deep lense, you could train it to recognize your cars, pull in the driveway, it recognizes your car that opens the garage car pool, you know, so do you want a place to begin? They've got some really cool tools, you know, it's practical, you can use it. So that's kind of a fun way to begin you there. I think one of the the other products they had was, you know, hot dog or not a hot dog from Silicon Valley. Exactly. And so that, you know, so that things like that. So Mark Cuban, I don't know if he founded this or if he supports it, but he has this group called AI for anyone. It's like a bi weekly newsletter. But like, it starts at the ground level. So if you're interested in AI, you have no concept where to begin. Follow their newsletter, they've got events, you know, they talk about all these different applications, different tools, different ways that it's been applied. And I think it's a great, great on ramp to artificial intelligence, you know, before you didn't get into using pytorch, and detect on TensorFlow and things like that. 

Justin Grammens  35:10  

So yeah, there's a project out there called Teachable Machine. Have you already familiar with that, it's a website you can go to, I'll put a link to it. And in the notes, but it's it is like computer vision, well, it does audio as well, learning one on one. So basically, you can, it's all browser based. So you go on in your, in your browser, you essentially, you can do it, you can do whatever you want. But then it allows you to basically attack the data, right? So it's like, you know, raise your right hand, raise your right hand, raise your right hand, and you tag a bunch of those, raise your left hand, you know, whatever you want. But it's it uses your computer camera to do whatever you want, right. So if you wanted to show a picture of a dog, or a cat, or whatever you train it. And you can do the same thing. Same thing with audio, but it generates a model for you. You can actually download it, it's all it's backed by Google. So it's basically a TensorFlow model. And you don't even need to even know programming at all. It's It's seriously like that simple. And it starts training the concept. Again, it's not unsupervised to be supervised learning. But it's like, hey, you know, this is kind of what it is at its root. And if you want to start playing around with it, all you need is a browser really, really fun. Well, how do people get ahold of you, Scott, if they you on LinkedIn, Twitter, like, what's the best place for people to find you? 

Scott Brown  36:17  

I'm on LinkedIn, I kind of got off Twitter, there's just too much hatred. I was a huge Twitter user loved it. But now, too much hatred out there, LinkedIn, and Instagram. So we've got Instagram account for transfer. We've just started using that. And put some fun stuff out there to kind of entertain people like creative things for with animals, and so on while we're getting ready to launch. I'm on LinkedIn at Project Brown, or Scott at transfur dot vet. So it's transfur "f-u-r". And so since I joined Transfur, I spell fill transfer wrong all the time.

Justin Grammens  36:59  

Yeah, that muscle memory.

Scott Brown  37:02  

But yeah, LinkedIn, you know, or Transfur. 

Justin Grammens  37:05  

I love the name. That's awesome. I mean, it's, did one of the founders come up with it? 

Scott Brown  37:10  

Yeah. So through sprint week, when they're developing the concept, they're doing the market analysis. And so we're looking at things. And so one of the things that they do is they name the project. And so they were looking for names that didn't exist in the market, but that applied, and so that was kind of the creative working title in it, and it stuck since then. I love it. I think it's great. I think it's very creative. 

Justin Grammens  37:34  

It is really, really good. Good. Well, cool. Is there any other topic or any other things that you wanted to share at all, before we sign off? 

Scott Brown  37:42  

I think these discussions are great, you know, kind of getting out in the community and exchanging ideas. There's so much cool stuff that's going on, like I said, you know, I'd follow you know, AI dot google and AI, dot Facebook, on Twitter, all these large tech companies have their, their engineering blogs, you know, sharing information and sharing their open source code and so on. And so I I think it's great that you're doing this, and thanks for having me on. 

Justin Grammens  38:09  

Oh, yeah, absolutely. Yeah, anytime, and I, you know, I hope to have you on again in the future. So just to see how you guys have progressed. And, again, it seems like an awesome product with a lot of potential and, you know, very, very clear cut, like, here's where we're going after, and here's the value that we're bringing to the market using this new technology. So I wish you guys nothing but the best I'm sure it's gonna be a fun ride here over the next couple years. I think for all of us in this AI space, there's, you know, there's many applications we haven't even trumped up yet. So that's what I think is so fun. Every time I have a new person on I learned something new. So yeah, hopefully intend to do these conversations around artificial intelligence on this podcast here for many more, many, many more years and chance to meet some really awesome people. So I thank you for being on the show.

AI Announcer  38:56  

You've listened to another episode of the conversations on applied AI podcast. We hope you're eager to learn more about applying artificial intelligence and deep learning within your organization. You can visit us at applied To keep up to date on our events and connect with our amazing community. Please don't hesitate to reach out to Justin at applied If you're interested in participating in a future episode. Thank you for listening