Michael is the Co-Founder of RevUnit, a digital product studio focused on building digital strategies and intelligent platforms to help clients. RevUnit specializes in elite, product focused digital teams and works with early stage to mid-size companies to build, learn and iterate on innovative digital products. In this episode, Michael and Matt discuss intelligent machines.
Episode Transcript
[music]
00:08 Matt Waller: Hi, I'm Matt Waller, Dean of the Sam M. Walton College of Business. Welcome to BeEpic, the podcast where we explore excellence, professionalism, innovation, and collegiality. And what those values mean in business, education, and your life today.
[music]
00:26 Matt Waller: I have with me Michael Paladino, co-founder of RevUnit, and he is a Walton College alumnus. We are thrilled to have you here tonight.
00:37 Michael Paladino: Thank you, Matt.
00:39 Matt Waller: Michael, I'm gonna ask you just a little bit about your company briefly. Although it's not we're focusing more on machine learning and things related to machine learning because we know machine learning is something that really everyone needs to be aware of, because it's clearly going to affect all aspects of business. It is already actually.
00:58 Michael Paladino: For sure.
01:00 Matt Waller: But I wanted to point out just a couple of things about your company before we got going. Of course, your company develops digital technology for employees, for companies and their employees to make their jobs more enjoyable and more productive. And it's an area where... It's new to me, I really didn't know that area existed before I met your company, but your company is doing really well. You're up to over 80 people. I know you've been on the Inc 5000 list for two years in a row, which is really uncommon and you have amazing clients like Walmart, Tyson, JB Hunt, and others. I know you have... One of your newer customers is a company that is an e-commerce, and what is that?
01:58 Michael Paladino: Zappos.
02:00 Matt Waller: Zappos. Maybe you could use that as an example to talk about what your company does.
02:06 Michael Paladino: Sure. So we refer to ourselves as a digital product studio. We create digital strategies and then intelligent platforms and products that help companies work better. That word intelligent is new and that'll become really interesting as we talk a little bit more tonight. But when you look at what we've been able to do with Zappos, if you understand any of the history of Zappos, and what they do, they are an e-commerce company, but they're known for their customer service. And they believe strongly that in order to have great customer service, they have to have highly engaged employees. So we've been fortunate enough to help work with them to build out a digital platform that allows their employees to share notes of congratulations with each other, share their wellness goals and how they're doing against that. They have something similar to a Make A Wish platform that they can do for other employees in the organization, and so that platform has become a huge part of their company culture, and how their employees stay engaged at their job. And obviously engaged employees, they're gonna enjoy their job more, they're gonna provide better customer service and ultimately lead to better outcomes and better results for Zappos.
03:21 Matt Waller: Well, again, congratulations on your outstanding success and Michael, I know that you won the Walton College Entrepreneur of the Year Award last year. So congratulations on that.
03:33 Michael Paladino: Thank you.
03:35 Matt Waller: So let's switch gears to machine learning and let's talk a little bit and not just machine learning, we talk about machine learning and artificial intelligence and these things have different meanings, but people use the terms and very loosely these days, there's so many different variants of them. But I remember 30 or 40 years ago, when you talked about artificial intelligence, it really was almost like decision trees, if-then statements and then pattern recognition like neural networks for "Machine learning." I think it was about eight years ago, I was listening to a book, and the book was talking about how things like self-driving cars would eventually happen but not in our lifetime. And part of the reason for that was that there's so much computation that has to occur for a self-driving car. It's more of a computer programming problem than anything. So people didn't realize how computer programming speeds would increase so much, memory would become so cheap, algorithms would take even existing speeds, and make them more efficient, but now it starts to become more and more of a reality. And it's not just self-driving cars. There are so many things out there right now where we see machine learning making a difference. Is it also making a difference in your area?
05:14 Michael Paladino: Absolutely, there's a lot of low-hanging fruit in this space. So the hype cycle will have you looking out and looking at all the future of general artificial intelligence, which is robots that are learning on their own and all the science fiction movies and TV shows that you watch. The reality of what can happen today though, the ability to do natural language processing, the ability to do computer vision and image classification, these sorts of things that were, required a lot of effort years ago. Because of some of the advances in the hardware or because of some of the libraries that are now available in the open source world, these things are becoming a lot simpler to do on a regular basis, and so that's a lot of the types of things that we're able to build into our platforms.
06:02 Michael Paladino: That's why we use the phrase intelligent platforms and products because we're building, in our business, we're building these platforms and products but then we're baking in aspects of intelligence into those platforms. And you're right, that it's absolutely the increase in the volume of data that gets created. Everything we do is now creating data and analytics and getting tracked. The cost of computing and storage is coming down. And there have been recent developments in some of the different algorithms, the neural networks and things like that, that are allowing some of these technologies to become more practical, whereas it might have been a little bit more theoretical or required too much computing power in years past to do well. Now those things are an API call away or something that you can run on one processor and get it done pretty simply.
06:57 Michael Paladino: So that's changing what you can actually do. But absolutely, we integrate a lot of that types of technology into the products that we build on a daily basis at RevUnit.
07:05 Matt Waller: And some of it is, for example, even when I'm writing a text message and I write T-H and it comes up with three options, that's a form of artificial intelligence. It seems simple to us now, 'cause we use it a lot, but in reality that's not even simple. The other day I was in a... At a technology conference that you were at as well, and one of the companies that was there had a face recognition software. What it can do is when, for example, a retail store could use this, so if someone walks in a retail door and they have a record of shoplifting, for example, the system identifies them. Now, a store manager can't necessarily do anything about it, it might be that they're fine, they've already paid their penalties or whatever, but the store managers can be aware. And so a store manager could go up to the person and say, "May I help you?" And just that kind of interaction can deter a shoplifter. In fact, they say a lot of times if you do that, someone who's intending to shoplift, they'll leave, they won't even stay in the store.
08:32 Matt Waller: But I was even thinking from the university perspective, we live in, of course Fayetteville Arkansas is a safe city, relatively speaking. I remember one time I... First time I went to New York University, just trying to get in a building there is difficult because it's in the middle of a huge city, whereas anyone could drive up to the University Arkansas and walk in the building in the middle of the day, right?
09:00 Michael Paladino: Yeah.
09:01 Matt Waller: And I thought it would be nice to have this kind of technology because it could link to, for example, all of our... We have photos of all the students and all the faculty, and all the staff. We could be alerted that there's someone in the building who doesn't fall into one of the categories, and maybe that's fine, they could be a guest speaker, but we've got their picture and you could actually, with the system that I saw, they would find them through pictures on the internet, LinkedIn.
09:33 Michael Paladino: Try to match up that face with social media pictures?
09:35 Matt Waller: Yeah. It picked me up immediately [chuckle] when I walked in. It's kind of eerie, but it's so powerful. But as we were saying before, just when you're text messaging, you're using a form of machine learning, or when you use a voice-first kind of a thing like Alexa Echo. When you're talking, your voice isn't like everyone else's. The way I say "Alexa" is different than the way you say "Alexa," so it takes some programming logic and mathematics to be able to identify, "Oh, they're saying, Alexa." And of course the more times it's said accurately, the more information they have about what it means. But what are some ways when you're dealing with solutions for employees, what are some applications that could exist?
10:32 Michael Paladino: In the employee space we see a lot of different applications, but one of the most common is just recommendation engines. So think about Amazon recommending products or Netflix recommending movies based upon other movies, certain... Amazon based upon other products that you've purchased. You see this in search, you see this a lot when someone's looking for an answer to a solution inside of an enterprise, so helping people get access to information more quickly, there's a number of other applications we see when we talk about automating the ordinary. Automation is a hugely popular use of these types of technologies.
11:09 Michael Paladino: So one of the examples that we've worked on, there's documents that are getting processed inside of a field facility, but then they go back to a person back at the main office to process that particular document, but there's a number of different people it could get routed to. And so, if we're able to actually look at that document, understand what type of document it is, that can then adjust who that particular document gets routed to, saving time. Again, that's one less person that has to look at that, and one less thing that they have to do. And instead, they can... They can focus on actually processing the document itself, as opposed to just classifying where it needs to go. That's a great opportunity for these machine learning technologies to learn what a particular document is, just like you and I would learn by looking at it and seeing a number of different examples.
12:05 Michael Paladino: So we see a lot of those types of examples in the employee space. Obviously Insights is a huge opportunity. So being able to use these types of, these types of models to understand the outliers, even in some cases make predictions of what's gonna happen based upon a number of different factors. Predictions become tough because you have to think about all of the different... The different external factors oftentimes, and it's difficult to understand what factors come into play, but some of these types of models allow us to do those types of future predictions as well.
12:39 Matt Waller: One type of artificial intelligence that I really like and appreciate, and the university has adopted something that's really good, and that is identifying spam. I see so much less junk mail than I used to. The university must have started using a software that has a really good artificial intelligence algorithm, 'cause somehow it's getting filtered out. It used to be, a long time ago when they first started doing that, we'd wind up getting things in junk mail that we shouldn't from people that we... We've even received emails. These things are being refined to the point where it's making a big difference. When I think about a company like yours though, how would you go about deciding where to apply it next?
13:39 Michael Paladino: That's a challenge we deal with on a regular basis at RevUnit. One of the first things we recognized as we started building out this practice about 18 months ago, was that it takes more than just having machine learning engineers on your team. It takes more than just having developers that are well-versed in using some of the third party products available. It has to be a company-wide effort, to look, to identify the opportunities to apply these types of technologies. So at RevUnit, we have product owners, we have designers, we have developers and our machine learning engineers. So early on, we created a curriculum at RevUnit that allowed our team to get to basically conversational knowledge of artificial intelligence and machine learning. So that they understood our viewpoint on the technologies, the types of examples they could use, and we had about two-thirds of our company go through that curriculum. Primarily to have that conversational knowledge.
14:38 Michael Paladino: So when it comes up in a meeting with a client, whether the client is specifically asking for that or whether they just see an opportunity for automation in a product or opportunity to provide a recommendation engine, or something of that sort, they're looking for those opportunities. So that for us has been an entire company effort. And we're starting to really see our team cash over the last few months has really responded to that and so we're starting to see more and more of these opportunities to apply these technologies to the products that we're building.
15:11 Matt Waller: I remember when they first started coming out with automated online assistance. And they were terrible. [chuckle] Right? I'd used it and I think, this is a waste of time. And so, I think for a few years, I just, I would bypass it, if I saw one, I just wouldn't use it. But I remember, I think it was probably at the beginning of this summer, I was at some website and it came on, an automated online assistant and it was so helpful. But it showed me that just over a short number of years, they've improved them dramatically, and in a lot of cases, with these types of things, the more people who use them, and the more frequently they're used, the better they get, like Google search engines and other types of search engines. Another place that they've been used, it's being used a lot is in healthcare, for prostate cancer, a real common surgical methods, the Da Vinci method and it's a robotic type of surgery but it uses lots of artificial intelligence and it's extremely effective. And now it's becoming the standard. Where do you think are some of the biggest opportunities going forward in the area of employee types of applications, say five or 10 years out?
16:52 Michael Paladino: In the employee space, one of the areas that you can't ignore is automation. And there's a lot of talk about this, there's a lot of talk about jobs having to be changed, jobs being eliminated, a lot of scary discussions there. The reality of what we see on a regular basis is, there's endless amounts of work to be done in just about every organization I have ever interacted with, and any time we can save effort, like that example that I shared about earlier of automatically classifying documents, if we can save effort that individual can be used somewhere else in the organization to go take on other work, and so, for us, absolutely, we're looking for any opportunities, and I think in the industry in general, any opportunities to automate those mundane tasks, that, we call them automate the ordinary, those things that don't require a whole lot of human intelligence to take something on, if we can automate that, and then free up that person to go do higher level tasks, those are huge opportunities. You see that any time there's documents being read in, photos being analyzed by human, those are all opportunities just to take that process and turn it into a digital process, and in some cases, use machine learning as a tool to enable that.
18:16 Michael Paladino: One of the other more interesting areas that you see a lot of talk in, right now, but it's not quite there yet is in talent management, so you see a lot of conversations right now of people trying to figure out, can I identify high performers earlier in their career? Can I identify individuals who may be interested, or may be thinking about leaving the company. You shared the shoplifting example earlier, if someone's able to talk to them, maybe they'd deter that activity. So what if you were able to know that an employee was within a few weeks of leaving the company based upon seeing similar patterns? And we're able to then go talk to that employee, figure out what was going on and keep that turnover from happening. That has big implications from a financial scenario, and from just employee satisfaction.
[music]
19:07 Matt Waller: Thanks for listening to today's episode of the BeEpic podcast from the Walton College. You can find us on Google, SoundCloud, iTunes, or look for us wherever you find your podcasts. Be sure to subscribe and rate us. You can find current and past episodes by searching BeEpic podcast, one word that's B-E-E-P-I-C podcast. And now, BeEpic.
[music]
