Fully and Dave interview Rob Carpenter, a serial entrepreneur and passionate technologist. Rob is the CEO and Founder of Valyant AI, the first Artificially Intelligent “Digital Employee” that works directly alongside employees in customer facing roles. Valyant’s AI “Holly” works in fast food restaurants to greet customers at the drive-thru post, answer questions, and take food orders. Rob was named one of the top 25 most influential young professionals in Colorado by ColoradoBiz Magazine in 2013 and received the Denver Trailblazer award in 2016.
Fully and Dave interview Rob Carpenter, a serial entrepreneur and passionate technologist. Rob is the CEO and Founder of Valyant AI, the first Artificially Intelligent “Digital Employee” that works directly alongside employees in customer facing roles. Valyant’s AI “Holly” works in fast food restaurants to greet customers at the drive-thru post, answer questions, and take food orders. Rob was named one of the top 25 most influential young professionals in Colorado by ColoradoBiz Magazine in 2013 and received the Denver Trailblazer award in 2016.
Rob Carpenter is Founder and CEO of Valyant AI, a company that pairs Artificially Intelligent Digital Employees alongside employees in customer facing roles primarily in fast food restaurants including Rallys and Checkers. Valyants AI Holly greets customers at the drive-thru, answers questions, and takes food orders. The company has raised$10.5M to date. Rob, welcome to the program!
Rob Carpenter:Thank you so much for having me. I'm excited to be here.
Fully Teasdale:I just road trip from St Louis to Michigan and back with my kids, which meant a lot of fast food restaurant, drive throughs, especially because like they want all eat the same thing. And I think every single place we went to there was, you know help wanted sign in the drive thru Lane including info on benefits and hourly rates which you would never have seen two years ago. So I'm guessing you're a company that has seen growth in this crazy pandemic climate. Yeah, I feel extremely fortunate, the pandemic has been really awful for a lot of people for a lot of reasons, but in terms of technology, especially in the restaurant space, it's accelerated adoption by five to 10 years. Right now, restaurants are facing a shortage, and they don't worry cite the source between 1.4 and 1.7 million people, and it was bad, even before the pandemic started there were 800,000 unfilled positions that the pandemic just accelerated that, and so you have a lot of restaurants where they are making more money than they ever had before and they're having to shut early or not be open on weekends because they literally can't find people to show up and work.
Dave Costenaro:So, you know this is a space that sounds, I'd love to hear some Rob about your your platform and your technology and what that looks like, you know startups and huge software companies are tackling things with conversational AI and interactions. How did you identify the drive thru experience as the pain point that you wanted to focus on.
Rob Carpenter:Yeah, that's a great question. So, our first product idea was to build full on holographic employees for physical locations, and within six months we'd actually prototyped a physical version of this. So, imagine a transparent OLED display, which is kind of like a piece of glass merge with a flat screen TV, and we turn it into portrait mode and we set it inside the frame of something roughly the size of kiosks, so it's about four feet wide six feet all and then we use the 3d gaming engine and the rendered a five and a half foot tall digital person and she could blink her eyes and shift weight, meaning when you started talking to her and we use off the shelf conversational AI technology to be able to talk to this holographic AI, with the idea that it could take your food order, a car, check you in your hotel room, you name it the ideas are kind of limitless. But what we found very quickly was that the actual interaction was terrible, and the off the shelf technology just really wasn't up to speed, the accuracy was really low, there could be 10 to 22nd delays before the AI responded back to you and it degraded kind of how cool it felt talking to this sort of synthetic intelligence. So we looked at if we were ever going to bring this technology to market we were really going to have to double down and focus on the actual conversational interaction and centrally interaction from almost being more of a hologram company to being more of a conversational AI company and so we had a bunch of use cases in mind and so I basically just sat down, when did you know listed out about 20 Different industries where I thought this technology could be applicable, and then created about 20 different sort of data points on basically the ease of integration, you know, and the ultimate market opportunity and so I eventually settled on QSR or quick serve restaurants. The menu sets fairly limited right there's you know maybe on average 100 to 200 menu items. A couple of 100 additional words like ketchup and napkins that you have to understand, but it's a fairly limited domain set, all things being equal, and then in the United States prior to the pandemic, the restaurant industry was an $865, billion per year industry. And historically they have extremely thin margins, and so if you can come in and make a restaurant more profitable within this.
Dave Costenaro:I felt that would lend itself, ultimately to a very large company. Yeah, so that's why we ultimately settled on the space. Fascinating. So okay, maybe walk us through what your implementation looks like from the point of view of a customer that, you know, drives up to a drive thru at a rallies or at a checkers or some other restaurant, and they're expecting to, you know, hey, order the number one meal, but I don't like mayo data. They're expecting to talk to a person I had said, What does it look like from their point of view. Yeah, absolutely. So, when the customer pulls into the drive thru, there's a magnetic detector that's installed underneath the concrete that senses that a car is there so it currently sends a signal to the headset system with the employees. That's what beeps in their ear and tells them there's a customer to greet. So what we do is we grab that signal, and then we install a patented piece of hardware called the MX one, a civilian is both a hardware and a software company, and the NA realizes that a customer is there, and then it greets the customers so it's Welcome to checkers, I'm your automated ordering taker, you know order when you're so then the customer will say something like, I want a cheeseburger, the audio from the customer flows through the microphone that's in the microphone speaker tower in the drive thru flows into the headset systems, the employees can hear what the customer said, and then it flows into our MX one a hardware, and then in there what our software does the first thing is we translate what they said, from speech to text and that's one of the key kind of areas of AI modules built into our platform. And in this case again it was, you know, I want a cheeseburger. From there it flows into the natural language processor, which the easiest way to think about a natural language processor or NLP is that it creates, intense, and so in this case, it's add one cheeseburger to order that then flows into the logic engine that says, Okay, do we sell cheeseburgers, is this the right time of day for cheeseburgers, are we missing any Cree key ingredients that we don't have an inventory. Should I ask if you want to make on dough. Should I ask you if it should be a medium combo or a medium, large combo, all of that, you know, decision making happens within this logic engine, and let's say it decides to ask the customer, you know, would you like to make get a medium or a large combo, the logic then outputs the text we convert it back to audio using text to speech, we then send that audio back to the employee headset system. So the employees can hear what the AI said, and then from there the headset sends that out to the speaker to be able to then communicate back to the customer, and that entire interaction happens in about one to two seconds, so it roughly feels like it's at about the speed of normal human conversation within the drive thru environment. So I'd imagine there are times when the conversation goes off the rails, maybe somebody uses slang or is hard to understand or uses a different language, what are your, you know fallback mechanisms to handle a situation like that.
Rob Carpenter:Yeah, so there's two main ways that we handle it, as does the rest of the industry that's focusing on building this type of technology. So there we refer to kind of having a human in the loop as auditing they are auditing the conversation. So we have call center backup, and the auditor's can do active auditing, and that's when they detect that there's a problem they hit a button, and they're able to just talk directly to the customer and say, oh you know I heard you say you wanted to check your burgers, but we added 22 checker burgers, you know, Let me, you know, delete those 22 erroneous ticker burgers. Is there anything else you want, and so they get the conversation back on track. And then there's another form which is called passive on, and that's where you maintain a veneer that everything is AI, but in reality behind the scenes, you still have that call center person that's watching the conversation, and in that exact same situation they might just click a button that tells the customer, please wait one moment, and then they delete the 22 checker burgers that shouldn't have been in the order, and then they fire off a new summary. Okay, I have two checker burgers in your order from the customers perspective, not knowing what's going on behind the scenes, it seems like the AI handled everything correctly. Okay, at least, the passive underneath. It allows you to maintain that veneer of the AI which maybe is a better customer experience. The benefit to the active audience tends to shave 10 to 20 seconds off your average order time, so it really just comes down to what's most important to the restaurant. Okay, so, you know, I imagine it's not a one to one relationship of like a sales rep to a to an order in so maybe that sales rep is monitoring multiple stations and you have some automated cues for that, that sales rep who's somewhere in a call center to be like oh this order looks like it's gonna have a an error so pay more attention to this one in real time while you're auditing. Yeah, correct. Yeah, it says, Oh, I think this conversation is going to go off the rails, you know, then on this one. Gotcha, gotcha. So what is the the tech stack look underneath that are your training, some of these big pre trained, you know NLP transformer models, I'm assuming that seems to be kind of a standard tool in the toolbox, are you fine tuning in, in training some of those. Is that what that looks like. Yeah, correct. So coming back to our initial problem statement, we just didn't find any off the shelf technology that we really thought was viable for this type of an advanced you get use case for conversational AI, I think you know it's one thing when conversational AI is running in a call center in the background, maybe providing KPIs or things like that, it's not 100% Perfect, not a big deal when the AI is actually carry non conversation with your customer and is the source where, you know 70 to 90% of your revenue comes from every failure is readily apparent. And so, in order to be able to bring a product to market we really felt like there just wasn't anything that was as good as needed. So for example, speech to text, you know, we built our own proprietary speech to text engine, we were the first company to get conversational AI into the QSR marketplace in 2018. And we have hundreds and hundreds of 1000s of customer interactions that we've used from noisy drive through environments to train our speech to text engine. And so we traditionally benchmark ourselves against Google and Amazon. So we're right about 90% accurate or speech to text versus Amazon and Google are roughly 72 and 71%, accurate by themselves. And so I feel like you know we had to make that significantly on our own because we don't hear what the customer said correctly. Everything else after that is going to be a fail. Yeah, yeah, that no, that's great, that's good, that's good practice, no I, one question I have then is, you know, well, when you're gathering your own custom training data. You say you have these recordings, is there a data label that's kind of automatically applied or do you need to go back and say this conversation was good or this conversation was bad, we don't want to use it in training and fine tuning the models to some signal that will will tell it for you.
Dave Costenaro:You know, honestly, right now we want both. And I think there's a lot of wisdom that says, just want to use the cleanest and best audio possible, the drive thru is such an environment where you have so much more so called like audio pollution between cars and birds and radios, wind, snow, you name it. So currently we train on both. So we don't really differentiate between good and bad audio, we incorporate all of it into our corpus, and then from a transcription standpoint, we have our vertically integrated data annotation platform with our own user base on there, and so we will provide the audience snippets to our transcribers, and then our speech to text engine pre labels everything, and then the user just logs in, they fix anything. If our transcript isn't correct and submitted that becomes the corpus for our training data. Okay, okay, that's that sounds good. You know I'm hearing a lot about, you know, a data centric approach to model training, instead of a model centric approach or do you know you kind of need both. And so Andrew Yang is one of the kind of industry luminaries in NLP that I follow a lot. And he's really talking about. He has some lectures that I've watched where he's saying, hey, you know, look at your data, different categories or classes that you've got. So you mentioned like, hey here's, you know 3% of our data has bird chirping noises in in 6% of our data has the radio on and etc etc So it's interesting hearing him talk like you can synthetically produce new data, you can you can augment your data sets, just like get a track of, You know radio noise and add that to some of the clean data if you need to like, you know, boost certain areas of your data set I thought I'd throw that in there I think that's interesting stuff. I think that's interesting as well. We've tried that, and taking that audio kind of noise pollution as I referred to it and adding it into clean data and trying to train on that. We saw mild improvement. We didn't see that it significantly enhanced any of the models are definitely a data first approach for versus a model first approach. Ultimately it's a hybrid of both. I mean we probably have somewhere around a dozen models under the surface that are all used at various points in our platform. But I think in the short term, even though it's fairly arduous data gets you where you need to be quicker and fast to be able to get a real product into the market and then you just hope that ML continues to improve at a rapid pace but then a lot comes down to data, training, you know, having the server space and the capital, keep running and all of those models and trying to build new models, and then ultimately and I don't think anybody knows the answer here but we need to get to a point where unsupervised learning, catches up and exceeds supervised learning and I don't personally think we're there yet.
Unknown:So talk to us. The thing that I find fascinating about your approach is the amount of AV testing, you're able to do, which is hard to do like with humans, I have calls that are experienced, trying to improve caller, experience, and, you know, trying to do a B testing or pilots in a call center with human reps is really difficult, like the, the cost, just to test and implement is really high so like tell us what you've learned in the AV testing area.
Dave Costenaro:Yeah, I mean, the way that I break it down is that I really compare kind of what we're doing to self driving cars I mean this is the technology, thankfully, you know, if we do our job, orally, you know, nobody gets into a crash and dies, and so it makes it a lot easier for me to sleep from that perspective, but there's a lot of similar challenges in terms of trying to take advantage of every piece of data available to you, and then trying to get every single individual component as close to 100% accurate as possible, so you know your speech to text engine, even in 90% isn't good enough and we have a lot of other things to do on the backside to make up for that, you know, but ideally be more at like 95% accurate, because if you look at speech to text and you look at NLP and you look at this logic engine. If each one of them is degraded by just 5% So if you're at 95% in each one of those three categories which by the way, is world class, in and of itself, you're still looking at an overall at minimum 15% reduction in the overall accuracy of the system because each one of these key kind of daisy chain areas reduces the overall accuracy of the entire system. And so it's been a tremendous amount of work trying to get to a point where every individual component is as close to perfect as possible. And we've found those live environments, absolutely critical to even come close to you know getting a product that's scalable in the marketplace because it's one thing for you know a bunch of white guy engineers in a quiet office to try and build a system, but when you actually start testing it with people that represent the real market, and all the variety of accents and colloquialisms and different ways and the different kind of audio sort of noise environments we talked about, you end up with just very, very different data. And so one of the things that, you know, proud of us having a product in since 2018 is, it's given us a lot of time to interact with a lot of different people here a lot of different ways to phrase requests to order various combinations of items, and so that's allowed us to go kind of from a data driven approach to making the platform better. It's not just our guests or our hypothesis of what we think it'll take but it's like okay somebody actually said, this sentence, we never would have thought of that sentence as a way to order food. So let's break it down, where did we go wrong, why did we go wrong and how do we improve our model. So, just say AV testing I'm trying to increase the overall accuracy of each one of these critical components has been absolutely essential for us to get to where we are today.
Unknown:And then what about the power and flexibility of things like upsells, where you've got, you know, an AI that knows automatically Hey it's 95 degrees today. And we've proven the shakes so much better when it's above 90 We're gonna offer between 11 and three, like, do you have data on the upsells, from the AI over shadowing that have humans trying to upsell.
Dave Costenaro:Yes, absolutely. So I think we definitely see kind of filling this hole in the labor market as sort of the number one value proposition for this technology right now and part of where you know tech is accelerated by pandemic. And then secondarily, see these upsells as being critical. So yes, we absolutely see examples like we talked about, but you can see big gains in areas that are even more freedom entry than that, you know with customers that we've worked with their employees on average only up sell 20% of the time. Well, AI never gets bored and never gets tired of asking that never forgets. And so we ask 100% of the time. So, right there you're already looking at a five fold increase in the number of times you try to upsell someone, and it really doesn't come down to any sort of brilliance, it's just sort of more at bats, so the more opportunities you have to hit a homerun, the more likely you are going to succeed with, you know, having the customer decide like, Oh, I do want to add cinnamon apple turnover that sounds delicious. And, yeah, along those lines like is the AI finding patterns that humans might not have recognized like, Oh, weird people want cinnamon rolls in the middle of the afternoon and we were only going to sell them in the morning, like is that happening. So, there's some of that I'd say more entry basis we're also still doing a lot of testing in thruline on, because there's a lot going on to around biometric laws data privacy, things like that and so the system can infer different things and it can infer that okay hey this is a Caucasian female over the age of 18 and under the age of 30 and at 7am You know and so that is going to give you a different kind of profile of a product, upsell, versus, let's say a 60 year old male at 5pm in the afternoon. And so all of that's viable, but we haven't released any of that yet because we still want to make sure that we're always in compliance with the local laws and a lot of that stuff is still evolving, as we saw with, you know, McDonald's lawsuits that came to the public about a month ago. Yeah, that's interesting. Yeah, that makes me think about the interfacing, like was it when you're trying to plug it into an existing system so like, say, say you're working with a restaurant and they want to install your system, what does that onboarding look like, how do you get the human labor to embrace what might seem, you know, complicated or even threatening the employee, and probably been our biggest advocates, to be honest, I mean, think customers are worried about a little bit of extra work on their hand, potentially, and the franchisees and the franchise ORS are kind of worried about it from a brand perspective, but the actual employees are big advocates for this type of technology because again, we're not talking about a situation where robots are taking people's jobs, we're talking about a situation where robots or restaurants are shutting down because they don't have enough employees. So we have situations with one of our franchisees where we have to find hours that we run it, and they will routinely turn it on. Outside of that, because they are so and this is the employees, they are so desperate for help. It may be it's a ship that was supposed to have five people but only three people showed up. And so the option is shut down the restaurant, or let's just go ahead and turn on the AI even though we're not really supposed to. So, contrary to that, we see employees is probably the biggest advocates within the industry and they are desperate for this type of technology because it makes their wives in their jobs easier.
Unknown:That's fascinating. And it's such a great point in terms of being able to you know, essentially, if somebody calls in sick you are you literally have someone on hold. That can be turned on if you need it. Right, well
Dave Costenaro:I mean, you have to remember to mean within a quick serve restaurant in 90% of restaurants the order taker is being asked to move on simultaneously so maybe half of their job is taking orders, but the other quarter is processing payments and another 20% is putting food into the bag and another 5% is when soft drinks or coffee, you know, I almost think of this like, you know octopuses with their own different places trying to do everything simultaneously, and it's one of the positions that's hardest if bill that will often have some of the highest burnouts and turnover, And so really what we're talking about is coming in and being able to take 50% of the task load off their plate, that allows them to a take a breath and not be quite as stressed, and then their time and energy and all the other things that they still have to do the now this is just one less thing that they've got to worry about. Think about it like fry cooks, you know, you're now starting to have kind of these automated fry machines coming in, I mean out onto the market so somebody starts to put stuff in there and pull stuff out, but the fry machine just makes their job, a lot less manual and a lot less sort of memory intensive of knowing every single thing that you have to do along the way.
Unknown:And are you seeing any pushback on the consumer side where people are like, I want to talk to a human.
Dave Costenaro:So we do see that but it's pretty rare. I would say it's probably like 1% or less where they're asking for employees. I think for the most part people just kind of go with the flow of the environment. And, you know, I think there is definitely or can be a bit of a learning per the first time somebody interacts with one of the systems, not too dissimilar from the first time any of us, talking to Siri or, you know, got an Alexa in our home. We do see significant increases in efficiency from consumers, the more times they interact with the system, and we are obviously along with the rest of the industry doing everything we can to lower any of those barriers are hurdles, the first time people interact with these systems, it ultimately just comes down to I think customers needing to be sort of a little more clear and a little more direct. And I think where we see customers having the biggest issue which I think is where frustration and things like that come out is when somebody is being indecisive, or speaking very quietly, or they change their mind halfway through an order, and those are all things that are hard for humans to manage, and as a result it's understandable that it's also hard for computers to manage this situation, but when a customer drives up and they're like hey I want ABC items and that's it. You know you're talking about an order that takes 20 seconds, and there's no human that can match that speed because we and the computer by extension are just so fast and so immediate in terms of printing customers, adding those items, submitting the order to the point of sale system, and closing out. So, what are your next steps Rob like what are what are the next things that you're looking at are you going to come out of the drive thru lane and into the dining room anytime soon are you looking at different cuisines or what what are the things that are on your to do list. You know where to start right you know definitely in self driving cars you know what's the next thing you're focused on, you're like, Well, how many hours do you have. Next steps on my task list. In terms of value and kind of where we're focused. I think we're still definitely in the process of working out, proving out the viability of the system right now most importantly to the corporate entities, because from then for them this is a sort of brand issue right if you're a franchisee, you're feeling the labor pains, if you're the franchise or you tend to be more of a marketing and real estate company and you're very focused on customers perspective of your brand. And so I think getting, getting the franchise or is more comfortable and familiar with this technology is really critical. I mean we're, you know, actively in the process of working or only now with, let's say, three of the largest 15 QSR is in the country and talking to six more about actively starting pilots so this is a some, something where every one of the restaurants in the industry are actively interested in this technology, or they're actively testing this technology and so for us it's starting to move beyond initial tests and actually starting to roll out the technology, which is not a simple task. But I do think we're making good progress there, we are converting our franchisee customers to paying customers which I think anytime you're bringing new technology to market, working through, you know what can be a very painful and long process of establishing product market fit. Once you start to get customers pain, then you can start to feel like I'm kind of getting to the other side of the chasm or product market fit so we feel like we're sort of arriving at that place, customers are wanting to pay we're starting to roll out to more locations we're getting the franchise ORS on board, you know, I think, especially in the drive thru environment, you know, we expect to see probably 50 ish locations where can we live by the end of the year and I think next year that'll probably be in the high hundreds. So I do think we're kind of at the phase of sort of aggressive route. So for valiant that's something that we're hyper focused on, we are talking to other restaurants outside of just kind of the burger category which is just sort of lucked into a lot of interest for consumers. So we're looking into that. I really like and I'm excited by the idea of integrating voice AI into the mobile apps and into cars, because a lot of people they're driving to a fast food restaurant to go in order. It's his way where you can just push a button in order through your car, or if you're walking up, push a button and just order it through your mobile app. All of those things help from a efficiency and convenience standpoint. Man, that I have thought that a number of times, and with that extra 20 seconds I would be able to watch that many more tech talks or Instagram stories so much more. Step by Step mobile menu, you just tell it what you want your day, no But in all seriousness I would I would I would be your first customer for that I would love to, on my way. Just order, it'd be safer to right like I wouldn't be trying to do it with with my fingers and looking at it, you're not You're not trying to glance down at a quick red light to try to, you know, quickly get to the menu to order ahead to be more efficient to just hit a button and then just order, you know through voice, while keeping your eyes on the road and your hands on the wheel, right, right, well let's uh, so we have a segment on our show called we got jokes, we have a few stupid stupid jokes. Were you warned about this Rob, do you have any that you've brought with us was painfully warranted warned about it Yes, excellent, excellent. Well I've got a zinger here and I just saw it a couple days ago while I was walking around on the street outside of the restaurant. So I'll start out it's pretty. I enjoyed this one. What's the difference between a good pizza joke, and a bad pizza joke. I don't know, the delivery. Nice. Alright well right back at you. Here's mine. How can we beat the AI invasion. I don't know how. We just were stoplights. Be the AI, I don't know if I get it. I don't get that. Oh, now well back to work on my delivery then within, within machine learning, AI recognizes a stoplight and knows that it needs to stop and do anything, are coming for you and you have a stoplight on attack you, because they've been trained to stop and wait. Nice, nice. Well I need to work on my gosh good i.
Unknown:We learned something. So that's the value right there. All right, I've got one. Why don't robots like apples. I don't know why, because they're androids.
Dave Costenaro:Good stuff. Well, they're all three jokes on a similar caliber to every joke time we have this segment, it's just all for guffaws and I roles. Because stuff, fully you want to go to the magic wand.
Unknown:Yeah so, we'd like to ask our guests, if you had an imaginary magic wand and you could change one thing that would make you know the world better for the consumers and customers that you're serving, what would that one thing be.
Dave Costenaro:Um, well I would take all the carbon out of the air and hopefully reverse global warming, so much happier and better off in life, and that would directly impact our customers. Nice, good use of magic wand. Thank you. I'm like I'm going, I got a magic wand I'm going big with this nice, cool. Okay, so our last segment here, we call it smorgasbord. It doesn't really have to do with tech or the industry or anything that we've talked about, just kind of some questions to get, get to know you a little bit better. First question, What's your weirdest phobia. Your dust phobia been any good phobia well do I suppose. I've been buried alive. I am a bit claustrophobic, and I hate that idea of being stuck in some small confined space that you can't get out of that, I would probably be the worst thing. That is terrifying. What is one book podcast TV show or movie that you'd recommend. Well, there's some things I disagree about it, but I do think it's very insightful from an AI standpoint, and I like to KaiFu Lee's AI superpowers book talking about kind of the rise of AI and artificial intelligence and sort of juxtaposing between the United States and China, and saying hey here's what one country does well here's what the struggle is that a country does well, here's where they struggle. And I think most importantly, you know what really interested me is that he talks a lot about the idea of AI or Artificial Narrow Intelligence and it's just staying super focused on doing one thing and doing it really well, and that in the next 10 to 15 years that's probably what we're gonna see the most successful AI. AI companies come out, versus trying to be everything to everyone and kind of probably disappointing everybody in the process. That's, that's a really important perspective. I really like thinking about things from the you know the worldview of the West to worldview of the east and kind of like looking at it both ways, I think that's really valuable. That's good. I only ask it to me. It's gonna be important to the future too so I think it's important to be aware of, for sure. Let's see, what's the best piece of advice you've ever received. Don't quit. I think I don't know that I can necessarily attribute this to any one source, but you know I am an avid consumer of books and podcasts with highly successful entrepreneurs, and, you know, especially when we're talking about beings in sort of the billion dollar valuation or exit range on luck. Absolutely plays a role and you'll you'll see a lot of people still reference luck. But I think generally, with a lot of the most successful entrepreneurs, they just kind of repeatedly state like I didn't quit. I never gave up, like I kept pushing and I kept pushing and yes it was hard and yes there were problems and yes there were barriers, but I just kept going and going and going. And eventually, you know that led to successful outcomes of one kind or another. I do what I can to get back and mentor other entrepreneurs that are coming up the ranks behind me. And that's just one of my most consistent pieces of feedback is just don't quit. Just keep going, just keep going.
Unknown:And one. And then our last question. I really like this one, if someone were to gift you $100,000 in shares of any company on earth that you could not sell for 20 years. What company would you want it to be.
Dave Costenaro:I would like to, you know, think I'm smarter than everybody else, but given the rate of things. Think I'd probably just take apple with a sit on Apple for 20 years because their, their meteoric rise over the last 15 years you know is one for the history books especially ironically an Android PC, windows person, but their ability to go, you know, especially from work Tim Cook took over to the first trillion dollar company is nothing short of amazing. And you know I think the focus on products and really trying to bring, you know, beautiful, really high quality technology to market has served them well. I think technology is getting harder, you know I think things like the smartphone and the iPad, you know, it took a while but I think we're seeing some of the new deep tech stuff is taking much, much longer. But if Apple continues, you know their same style of not always being first, but generally maybe having the highest quality product on the market. I think they will continue to do significantly well over the long term. This has been an excellent conversation Rob, I know I've learned a ton. I appreciate you taking the time to be with us, we like to give our guests the last word. Is there anything that we didn't cover or anything that you want to say mentioned or promote. Before we go, I would just like to say that, you know, we need more brilliant people out there working on these types of technologies and these problems. I think that there's a lot artificial intelligence can do over the long term to significantly improve the quality of life on Earth, and so I would just like to encourage anybody that's you know younger, maybe trying to figure out what they want to do with their life or they're considering a new career path, you know, really consider software, consider artificial intelligence. We can use anybody possible who's interested and, as a final plug following on that valiant is actively trying to hire for six different positions so anybody's looking for work, please, please contact us and you can find us through our website@valiant.ai, and we'd love to talk to you.
Unknown:Awesome. Well thank you so much. This is such an interesting conversation and we're excited to see what valiant does in the next year or two, we'll have to keep in touch.
Dave Costenaro:Thank you, appreciate it. recording stopped. All right. All right, great. That was great. Perfect. Fascinating. Perfect, perfect. Yeah. Your questions are good, it's hard to like think on your feet, you know about the magic wand or a book or upon, things like that and it's like, quickly and it's got to be insightful and reason why I see that you're a type A person Rob. Perfect. All right. That's funny. Cool, cool. Um, yeah sorry go ahead. Yeah, I
Unknown:was just looking, I think this will probably post in like four weeks, and we will give you all heads up, send you like some promotional graphics somewhere. Yeah, so probably the fifth October 5 And maybe two weeks after that but let's say the fifth for now.