“Nothing’s Sacred” Episode 6: KUNGFU.AI’s Steve Meier

“Nothing’s Sacred” Episode 6: KUNGFU.AI’s Steve Meier

KUNGFU.AI Steve MeierKUNGFU.AI‘s head of growth, Steve Meier, joins us to explain all the hot issues related to artificial intelligence.

How do you use AI to drive sales? What are the ethical boundaries of AI? Which movies or shows offer the most accurate depiction of AI? And does he trust Amazon Alexa?

Scroll below to listen to the podcast and to read the full transcript. You can also listen to the podcast on SoundCloud, Apple Podcasts, Spotify, and Stitcher.

Listen to past episodes here.

Nothing’s Sacred Podcast · Episode 6: Talking AI with KUNGFU.AI’s Steve Meier


Nothing’s Sacred: Episode 6 Transcript

Nick Schenck: [00:00:00] This is Nick Schenck, co-host of  recruitAbility’s “Nothing’s Sacred” podcast along with recruitability CEO Nad Elias. This week, our guest is Steve Meier, head of growth at KUNGFU.ai, an AI services firm based here in Austin. KUNGFU.AI is a partner with recruitability and has an interesting backstory, which Steve is going to get into.

[00:00:21] We’re also going to learn about many of the ways that AI is impacting and transforming how we go about work these days.

[00:00:26] Just a heads up, Steve joined the podcast remotely and there’s a bit of an echo with his audio, but we did our best to minimize it.

[00:00:34] We’ll pick up on our conversation with Steve, who explains how he got introduced to AI in 2015, when he was working in consulting, primarily with companies in the Fortune 500 tech sector.

[00:00:46] Steve Meier: [00:00:46] So I had the luck to come across IBM Watson in 2015, which is really at the height of the hype bubble, where everyone was looking at artificial intelligence from the marketing machine and seeing all these lengthy promises and all this cool action with Jeopardy and in chess and whatnot. But there was really a practicality behind it.

[00:01:08] And IBM first started releasing their AI as a microservice and as an API for people to build around, which provided us a unique opportunity. So, um, I had the opportunity to start applying it to client challenges , and eventually it got so interesting, I thought there was more that could be done. So it was fun to start to apply it to someone else’s business, but started to think, you know, how might I be able to do this on a grander scale?

[00:01:34] And coincidentally, I was approached by a serial entrepreneur in town, Stephen Strauss, one of my co-founders, and had a similar vision and said, “Hey, you know, you have a lot of great services experience, and I think you can bring a lot of value in helping me start this AI services company.” So that’s what KUNGFU set out to do.

[00:01:54] It’s interesting in that this is my first venture, but I got to join three other serial entrepreneurs who are somewhere around 20 startups to their credit. And it’s a great strategy when you’re getting started, because they can show you the ropes. And these are product guys. And we intentionally came together to start a services company, because it’s just not the best climate today to create artificial intelligence products.

[00:02:20] It works, but it works very well on a tiny solution that is usually very customized business to business. So there’s always a lot of services behind, that and we thought it might be easier for us to bring goodness into the world if we were actually building bespoke or custom solutions for our clients that were made to work versus trying to pick the one problem to solve and customize it across the board, I think was more of the tougher sled for us.

[00:02:48] So we started KUNGFU.AI under the premise that we wanted to be intentionally a services company. And, you know, so far it’s been a really good move for us. We’re enjoying the work. We get to see a lot of different challenges, and whether it be providing strategy or creating custom solutions for our clients, we get to see something new and we get to learn, but also apply ourselves to new challenges all the time. So it’s a lot of fun.

[00:03:13] Nad Elias: [00:03:13] Can you talk a bit about KUNGFU.AI’s “AI For Good” program?

[00:03:18] Steve Meier: [00:03:18] Yeah, something we’re super proud of and something that we really started doing out of the gate. I think one of the interesting things about artificial intelligence. It’s very much a reality today. This is no longer the future or tomorrowland. This is our reality. Artificial intelligence for certain businesses is ubiquitous. And for others, it seems like such a moonshot. And if you kind of decompose that down to individuals who have the ability to provide artificial intelligence or have that background, the future is here, but it’s just not equally distributed.

[00:03:53] And we thought it was, you know, part of our charter to understand our privilege and the fact that we were lucky to be here to be business owners and we were lucky to have this background and skillset, and there are others that are less fortunate. So to the extent that we can give back to the community and give what can be, you know, very high-end lucrative services and give those to people who don’t have them for free to do some good in the world, or to coach people on how to get into the artificial intelligence space.

[00:04:25]But we basically just shined the beacon said, “Hey, we have open office hours. If you are looking for free advice, or if you are looking for mentoring or coaching opportunities, or just understand how to get into this game, we’re here for you.” And surprisingly all different walks of life started to show up, and we’ve had the pleasure of really parlaying some of those first encounters into pro bono projects where we’ve worked with groups that fight human trafficking. We’ve worked to help create recommendation systems that match mentors to foster children, so they can have an adult in their life that can make an impact. We’ve helped identify gunfire in the rainforest to detect poaching.

[00:05:10] It really takes on a life of its own and it’s a little bit different, but we do it every month.And it’s something that’s very near and dear.

[00:05:16] Nick Schenck: [00:05:16] That’s awesome. And I think that’s important too, because for right or wrong, I think a lot of people, the first thing they think of when they think of AI is either Skynet from Terminator 2, or they might think of job elimination.

[00:05:31] And so for y’all to really focus or emphasize AI for good and all the great applications of the technology, I think that’s important for dispelling a lot of myths as well. Do you agree with my initial assessment that when people,  if you’re at a dinner party pre-COVID, when you bring up AI, do a lot of people bring up Skynet or other things?

[00:05:55] Steve Meier: [00:05:55] Oh, all the time.  It’s funny for how  long-standing this technology’s been around, you know, AI is,  by some measures it’s almost 90 years old.

[00:06:03] This is not new stuff, but it’s very new to most people. And the opportunities that we are leveraging today are still very new. So there’s been a lot of advancement and you come across a lot of hurdles, but conceptually it’s been around for a long time. But we still are doing a lot of educating and I think that’s just the number one barrier to adoption right now is just the education hurdle.

[00:06:24] And to some extent, a cultural hurdle.

[00:06:27] Nad Elias: [00:06:27] Hearing you talk about the AI for Good shows me how many applications, you know, AI has, and it really applies to any industry vertical, right? I mean, I think it was last year Gartner came out with a report to be in their top quadrant, you need to be using AI in your business.  All these companies now are trying to figure out the what around AI, what do we need? Right.  And from a services standpoint, that’s what they’re coming to you guys for. Right, Steve?

[00:06:53] Steve Meier: [00:06:53] Yeah, absolutely. It’s truly ubiquitous technology, and it will touch everything and anything. And it’s just going to take a little time to get there, but regardless of industry, you’re going to see have’s and have not’s. People that are incredibly mature and have been doing this at scale for quite some time.

[00:07:13] You’re going to find many more who have yet to get started or even hire their first data scientist. So it’s not really an industry thing. It’s just very much a who figured it out faster type of thing.

[00:07:30] Nick Schenck: [00:07:30] When people will reach out to Kung Fu, do they usually reach out to you with like, “Here’s our pain point. We think AI could solve this.” And then they ask you guys to do an assessment? Or are they coming to you with, like, they’re just curious about AI and they’re like, you know, can you just list out 10 ways we could use AI to improve our business? Give me a sense of that.

[00:07:54] Steve Meier: [00:07:54] Yeah. I think it’s more of the former, though I think our original hypothesis thought it was going to be the latter. We thought we were going to come in and people were going to be so excited and curious about artificial intelligence and feel totally lost on where to start, but they also were so excited about it and they wanted to have a strategy before they actually started embarking in just fruitless R & D or development that may or may not work.

[00:08:21] And that’s really not the case. We find very few clients that are actually thinking about it that way. Many of them come in with a fuzzy idea of what they can and should be doing. They don’t know if they can pull it off. They don’t know how practical it is. So we’re coming in helping them de-risk different projects.

[00:08:37] And if this project you present to us isn’t the best one, well, let’s find one that makes sense to your business. And then let’s find a quick win. And once we find a low hanging fruit opportunity or something that we can build in a few short weeks or a few short months, that’s kind of the momentum builder for the business to say, let’s take a step back. Let’s think about strategy. Let’s talk about AI across our business and come up with more of a roadmap for this stuff. So it usually moves like that.

[00:09:07] Nad Elias: [00:09:07] Is there a story that was particularly challenging for you and your team, where somebody came to you with that scenario and you guys devised the right AI solution. Is there one that comes to mind that’s just man, that was a good notch in our belt.

[00:09:23] Steve Meier: [00:09:23] Oh, absolutely. You know, one of our largest and longest clients when we first got started was working with one of the biggest real estate companies here in the world. And they were really broadcasting a vision of how data and artificial intelligence was going to change their business radically, but internally they weren’t quite sure what that meant or where to start.

[00:09:49] And they needed momentum to then parlay into greater investment and scaling it across the business. So the first project we worked on was really using artificial intelligence to help them  convert contract data in real estate transactions and extract that information from the paper contract, which is a data source, right? It’s just not a usable data source. You can’t actually write a SQL query or call up information in a PDF contract. So when you want to understand patterns and predicting property prices, for example, we have to look to the past and understand how those contracts were authored and were they accepted or rejected.

[00:10:24] All of that goes into the equation. So part of phase one was really to use artificial intelligence to prepare them and extract data and create this data mode. So now they can do a whole bunch of machine learning on top of it and start to build an advantage around the capability. So we were able to do that.

[00:10:42] We were able to parlay that extraction – or data mining work – into a bunch of predictive capabilities. Everything from trying to predict which of our leads are likely to convert and worth spending our time to what is the price of that property? How can we get it sold in a good period of time without just giving it away?

[00:11:02] Nad Elias: [00:11:02] Knowing you want to buy a house before you know you want to buy a house.

[00:11:06] Nick Schenck: [00:11:06] That’s true. What about those companies like, is it Opendoor? Like they’ll make you an offer on your house sight unseen because they’re confident in the market by looking at all the data, right?

[00:11:16] Steve Meier: [00:11:16] Yeah, no doubt. And the interesting thing is it really opens up the opportunity. I mean, artificial intelligence can fundamentally help businesses change their business model, whether you want to sell the data that you’ve been sitting on for decades to another industry, because that industry needs it, or then using the home prediction model to then move from facilitating real estate transactions to actually making real estate purchases and then selling those homes. I mean, these are all very new things to many of our major real estate companies. And you know, this is not just anecdotal to one industry. I mean, there’s lots of opportunities where people look up and say, “Oh, this helps us become a slightly different business than we were.”

[00:11:57] Nick Schenck: [00:11:57] Steve, do you, do you have Amazon Alexa at your house or Google Home?

[00:12:02] Steve Meier: [00:12:02] All of them.

[00:12:03] Nick Schenck: [00:12:03] All of them? Okay. I have friends, family members who only plug in an Amazon Alexa when they are going to listen to music or something, otherwise they don’t want it on in the background because they’re afraid that their conversations will be picked up.

[00:12:20]Are you paranoid about that or not?

[00:12:22] Steve Meier: [00:12:22] I am not, but I understand that I may be in the minority of that camp. I think part of being a practitioner always getting stymied on not having enough data, I’m in the camp of the more data, the more opportunity, right? Because that’s just the reality in artificial intelligence.

[00:12:40] However, I’m also in the camp of, I feel like I’m being a good person and living a good life, not doing anything that I otherwise wouldn’t want someone to have that information. So I don’t really worry about those types of things, because I feel like the greater good is being able to have that data and then make my life more convenient and easier, and give me a better, more personalized experience. You know, I want you to get to know me and want you to anticipate my behavior, because the trade-off for me is easier time to find a product or make a purchase or make a return or whatever that might be. So, you know, it’s really kind of cost-benefit for everyone.

[00:13:19] You know, if you feel like that artificial intelligence is going to make your life easier and the tradeoff is having access to your data and you feel like the benefit is the experience. That’s great. If you don’t feel that way, I totally understand why you want to turn those things off or mute it, or whatever you have to do.

[00:13:39] Nad Elias: [00:13:39] My wife’s cousin, her daughter is named Alexa, so now they can only go with Google products.

[00:13:48]Nick Schenck: [00:13:48] Wasn’t there a Super Bowl ad that said the word Alexa, and like everyone’s Alexas went off. It was like this big controversy. And actually that’s a smart Super Bowl ad because you get a lot of publicity around it. Right?

[00:13:58] Nad Elias: [00:13:58] Yeah, that’s right. Hey, I want to talk a bit about, you know, in my world, Steve, where we work quite a bit on the interview side of things, right? Some of the predictive analytics in our world around the interview process, right? Assessments are a big part of what we’re seeing now with companies and how smart these assessments have become.

[00:14:26]15 years ago there were assessments that would take an hour and a half for a candidate to complete and might check off a couple of boxes on a personality profile fit. And now in 20 minutes, they can nail a profile around culture and skills.

[00:14:43] I’m curious your thoughts and the AI of assessments as I call it, I feel like they might be getting so smart that they could remove the bias of interviewing, right? The natural bias that comes from, “You know, hey, I really liked that guy. I’d go grab a beer with them.”

[00:15:00] Versus, you know, are they a fit for the job regardless of what they look, feel and sound like?

[00:15:06] Steve Meier: [00:15:06] Yeah. I think there’s a lot of work that is very practical and very feasible in that domain. And I have to assume companies like Indeed are all over this and have armies of data scientists to better streamline this experience.

[00:15:23] And, you know, there’s a lot going on right now in natural language processing and using newer architectures. We’re talking about certain architectures that are less than two years old and to some extent, less than six months old using transformers. But we are able to understand text in a new way. We are able to summarize and map text and match texts.

[00:15:49] So there’s a lot of opportunity to use artificial intelligence to scan resumes and make recommendations on what type of roles are best suited for that resume, or if you’re doing these tests, I’m sure there’s going to be a lot of work in regards to Myers-Briggs and Enneagrams to better understand some of the softer skills and personality fit, too, which is just going to be a bunch of data about your business and the people in your business and how they are and act and behave. And then ultimately trying to see if there’s a fit for that individual. So whether it’s recommending jobs or recommending fit of a candidate, because that data is now processable using different forms of machine learning or deep learning.

[00:16:32] I think it’s very, very practical and very feasible. I mean, we can understand sentiment of our cover letters, for example. I think it’s good to know before you interview someone if that individual is actually enthusiastic about the job or if they’re kind of faking it, or if they’re maybe just doing it because they need a job so bad and maybe they don’t actually love this particular gig.

[00:16:54] So there’s so much to do, you know, and I think the next frontier there is going to be more about understanding some of the non-written, non-verbal communication. So, you know, trying to hire these days is probably very difficult, especially when we can’t be in person as much, and Zoom allows us to get around that to some extent, but you know, we’re not going to get everything from  a test or an assessment. We have to understand and be able to read body language. And that’s a nascent space in artificial intelligence. So it’s very possible for it to go there to using things like computer vision.

[00:17:24] Nick Schenck: [00:17:24] Natural language processing to me is really fascinating.

[00:17:28] I mean, for instance, there’s companies that tap into the APIs that Twitter, Facebook have, and they can do sentiment analysis for your brand. So they can like search your brand name and be able to detect based on how your brand is used in conversations, whether it’s positive or negative sentiment.

[00:17:48] And you can like measure that over time. So I think, and Steve, you probably know better than me, but I imagine huge consumer brands, Coca-Cola, Pepsi, are probably looking at that stuff all the time and sort of juxtaposing that with their ad spend and their ad creative. But I think that’s a huge opportunity. Steve, what are you seeing there?

[00:18:10] Steve Meier: [00:18:10] Yeah. That’s, you know, something that’s been around back when I was first leveraging IBM Watson back in 2015. So sentiment analysis and that type of natural language processing isn’t even all that new, but what’s more interesting now is to be able to understand, okay, so my brand has negative sentiment and let’s give it a score from zero to a hundred, and you’re a 53, and all your competitors are 74. It’s a negative sentiment, but what you do with that information, like how is that information actually actionable? And what’s becoming more interesting – and the newer area of artificial intelligence is now to say, okay, well help me summarize how people feel about my brand, and what exactly is getting that issue or pause or concern.

[00:19:01] And once you can get down to that level, you actually have actionable feedback that you can address versus either over-indexing on a vocal minority of opinions, or just blindly guessing. Oh, you know, I knew we were negative and I always had a suspicion it was these four things. Right. So let’s go address those four things.

[00:19:19] So, you know, sentiment analysis is a step in the right direction, but it’s really incomplete intelligence in a sequence of things you’d need to know to actually make some sort of meaningful business adjustment.

[00:19:31]Nick Schenck: [00:19:31] Out of curiosity, where do you think AI goes too far? For instance, I think recruitAbility, you published a blog on synthetic voices.

[00:19:40] So being able to create a synthetic voice based on  a source voice, so like if you provide your voice, you know, 10 seconds of your voice, all of a s udden, you can mimic your entire language.  Steve, I don’t know what you know about that, but are there  instances where you guys are like, that’s taking this technology too far and there’s too much of a risk there?

[00:20:01] Steve Meier: [00:20:01] Yeah. I mean, I think there’s a lot of concerning areas in regards to artificial intelligence.

[00:20:06] It’s very under-regulated and there isn’t a whole lot of consensus in the industry of where do we go and where do we not go. It is very much a thing that’s happening in the zeitgeist and happening more on a business-to-business level. But, you know, the synthetic voice thing is probably related to a lot of the deep fake stuff, which is much broader than just voices, you know, faking a likeness of an individual, being able to superimpose one person’s face and one person’s voice onto another person and simulate that voice. So, you know, there’s a whole lot of areas that are deeply concerning and you know, right now there’s less people trying to monetize those areas, but there’s a lot of bad actors trying to exploit those areas for their own nefarious deeds.

[00:20:49] So, you know, that is a big area  of concern. And we’ve been even approached by, you know, different government entities to say, how do we create a deep fake detector? We’re not even certain that’s possible, right? We might be so far advanced that by the moment we can detect a fake, then a new fake is created, right?

[00:21:10] And then there’s always this battle between the people who are counterfeiters and the ones that spot the counterfeits. These get more and more sophisticated. You know, an interesting approach was brought up by one of our advisors, is we might just need to authenticate real content , like our Twitter check verified.

[00:21:25] Right. So instead of trying to say that’s fake, that’s fake, that’s fake, you would go back and say, well, this is authentic, this is authentic, and this is authentic. And you have a certification process for that. That’s probably where that may be going. But other areas are really facial recognition. You know, there’s a lot of concerning areas around the privacy behind facial recognition. Using it for policing. You know, a lot of that is, very much in the mind’s eye right now. And there’s even pledges that we’ve taken called the safe face pledge to say, you know, we will not use facial recognition for policing. If our client asks us to do that, we will politely decline those opportunities.

[00:22:01] But you know, there still isn’t a whole lot of regulation or consensus even there.

[00:22:05] Nad Elias: [00:22:05] In my interaction, uh, with you guys, I think one of the things that impresses me most is how important the ethics of AI are to KungFu AI. It’s something, I mean, you guys aren’t afraid to say no. Right. And that’s something that I think is an important characteristic in the value that you all provide.

[00:22:29] Nick Schenck: [00:22:29] So earlier in the conversation, you mentioned that, you know, education was a big obstacle in terms of the growth of AI. It’s not necessarily new technology – you mentioned – but educating people on it is something that you guys are emphasizing.

[00:22:44]And I imagine another obstacle for y’all is finding talent that has the skillset to work on projects in AI, machine learning, etc.

[00:22:54] Steve Meier: [00:22:54] Yeah, I don’t think the education will always be an issue. You know, we’re very much crossing the chasm when it comes to the industry of artificial intelligence, where we have now proven the technology. There are enough early adopters that there’s going to be this gap before kind of the early majority jump onboard and say, okay, now we’re comfortable moving forward with this technology.

[00:23:20] And at that point in time, people are going to be smart. They will continue to get smarter around artificial intelligence. More importantly, they’re just going to start hiring the right people who are experts in this domain. And they’re going to have good access to those individuals. Right now, the talent economics for trying to hire someone who’s really adept in AI machine learning is not in the favor of the buyer, meaning that there is far more demand than good supply, and there’s still role confusion on what exactly is it that I need, where there are many people who can claim to be a data scientist. And there are many reasons why you would want a data scientist, but there are many reasons why you would fall short if you were to over-index on a team of data scientists thinking that they could do certain types of machine learning.

[00:24:07] So it does require the business to be well-informed enough to know what they want to work on. Right. So step one, isn’t just hire a team of data scientists and find them work. It’s what do we want to work on? And what projects will lift our business? And then let’s go hire the skill sets that are very versed in that area. And when you get that far, you’re still going to run into a hurdle of this person I’m trying to hire probably has four to five other companies begging them to join their company. So you always have to pitch the machine learning engineer on why they should work for you and not someone else.

[00:24:46] So there’s salary requirements behind that. You have to have a great culture. You have to have a great business, and you have to have really interesting work to compete for these people who have many suitors, which isn’t great for many businesses, right? Maybe they don’t have a sexy product or a sexy industry, or maybe they’re working on improving their culture and they need people, too. And it’s harder for them to compete.

[00:25:10] And that’s why companies like us still make great sense. Right? Well, let’s help you bridge that gap. But yeah, it’s a good day to be a machine learning engineer/data scientist.

[00:25:19] Nad Elias: [00:25:19] And one of the things that we’ve noticed in recruiting in this space is that there’s just not a lot of people that do it yet because the education hasn’t necessarily been there.

[00:25:32] So recently, UT was selected as the home for the national AI Institute for Machine Learning. I mean, great to see, you know, my alma mater stepping up to the plate on that. But you see a lot in this data science world, and data engineers as well, where it’s still a new profession where it wasn’t, you know, sure you had a computer science degree or a double E and then you kind of fell into it.

[00:25:59] Or I think we talked about this in our last podcast, but we were seeing, um, you know, data scientists coming out of school with six months of experience, one or two projects under their belt, getting base salaries of $130,000-$150,000, you know?

[00:26:16]It’s because there was this sort of gap between you have your PhD data scientist that has 10 to 15 years experience, and then you have your 0-5 years experience, and then there’s nobody in between. Right. Yeah.

[00:26:34] Steve Meier: [00:26:34] Yeah, it’s a, it’s a great point. And academia is only catching up to provide both undergraduate programs, as well as masters and PhD level programs. There just isn’t anything that makes you a PhD in AI.

[00:26:47] Nick Schenck: [00:26:47] I have quick question about AI and business development. So for listeners here who maybe they run a company or they’re head of business development, chief revenue officers, you know, especially during COVID, everyone is trying to drum up new business.

[00:27:03]What are the best ways to implement AI or machine learning to help or facilitate business development? Or is it already built into Salesforce and popular CRMs that people already use, and they don’t even necessarily realize that it’s the AI that’s driving a lot of value in those products?

[00:27:23] Steve Meier: [00:27:23] Yeah. I mean, to some extent you’re absolutely right. You know, companies like Salesforce have Einstein. And there are plenty of analog software companies, including HubSpot, that offer some sort of machine learning that can give you some predictive capabilities when it comes to forecasting or understanding likelihood for leads to convert.

[00:27:46] And I think what we’re finding is that for the same reason we decided not to productize, is it’s hard to do any sort of machine learning thing well for a wide market of different customers with different types of data and inputs, and make high-level predictions. So what happens is a lot of the out-of-box  solutions are fine. They may be a lot better for less sophisticated sales organizations to get by and to better engineer themselves. But then what you also find for the more sophisticated sales organizations that they’re forecasting is still probably done in some sort of linear regression model or just done in Excel.

[00:28:24] And that works well enough, but all of those individuals, regardless if you’re using the out-of-box solution or your homemade solution, are feeling like the crystal ball just isn’t very accurate and it’s highly variable. And there’s still a lot of tribal knowledge going on to handicap that final number, and you’re not using any sort of automated output and showing it to your board or showing it to your CEO and saying, that’s what we’re going to do this year. And the robot produced it, or the Excel produced it, or Salesforce produced it. We’re not there yet. And that’s where sales forecasting still is a great area for machine learning is, you know, let’s use some of the more advanced models, let’s take into account all different types of input. So not just numerical input or categorical input, let’s take textual input from market reports from some of our key clients to better understand their market performance. So we can give better predictors on that account performance, because that’s what good account execs do. We don’t have a way to actually model that out until recently. Right? So let’s take in all these different types of data sources that may live outside our sphere.

[00:29:32] Let’s plug in the data inside our CRM, and let’s start doing more robust forecasting using some of these more state-of-the-art models that can take advantage of more, better, bigger information. And all of a sudden, we start getting some really interesting output. So we’ve worked on a handful of sales forecasting and demand forecasting type of projects.

[00:29:52] And that’s a great place for sales to start, but beyond that, it’s still need prioritization, trying to understand who’s likely to convert. There’s a lot of admin that sales does that can be automated through machine learning. Let’s automate meeting notes and call transcriptions, all the time we spend filling out the CRM, and there’s a lot of opportunity to use machine learning to automatically drop that input and keep your CRM very clean and crisp, which is like the biggest bane of most CROs’ existence is just trying to keep a clean, tidy Salesforce or whatever you’re using.

[00:30:30] You know, trying to understand next best actions. So we had this call with a customer, now what should I do? Let artificial intelligence make a recommendation. There’s a software company called 6th Sense that is trying to crack that code and listen to the actual meeting and then say, okay, based on what I understand, here’s what I think you should do next to move the customer closer to close.

[00:30:53] And then just sentiment, just understanding, you know, did the customer enjoy that pitch? Did that call go well from their perspective? There’s another company called GONG.io that is basically listening to SCRs give their sales presentations and their phone calls and giving real-time coaching.

[00:31:12] You know, based on what I’m hearing right now, try to offer this incentive. That might get them over the goal line. So you know, these software companies are probably more earlier stages, but that’s kind of where this thing can go in the future beyond the forecasting, the lead modeling, etc.

[00:31:27] Nad Elias: [00:31:27] There’s a company we’ve talked with, and we’ve actually demoed Kronologic.ai.

[00:31:33] Have you heard of them, Steve?

[00:31:35] Steve Meier: [00:31:35] I have not, no.

[00:31:36] Nad Elias: [00:31:36] Kronologic.ai. These guys are claiming, you know, using obviously AI, that they’re just gonna put meetings on your calendar, right. T hey sync up with your CRM. They use their algorithm to go out and start reaching out to prospects.

[00:31:53] And then all you do is sit back and get meetings put on your calendar. I’m like, this is great. We haven’t pulled the trigger on it because it almost seems too good to be true  Maybe this is more of the future of AI, right? This is where we’re going.

[00:32:04] Nick Schenck: [00:32:04] I’m thinking of like, if you’re on a sales call and you’re getting real-time feedback from a third-party on like, what incentive you should offer this and that. It makes me think of this is way back, but Steve Martin was in this movie called Roxanne where he would like give dating advice through an earpiece to another guy who like, was like an attractive guy, but he didn’t know how to talk to women.

[00:32:25] Nad Elias: [00:32:25] Yeah.

[00:32:25] Nick Schenck: [00:32:25] Imagine being on a sales call and you pause because the AI isn’t fast enough and you’re waiting to find out what to say next. Right. It’s just, there could be so many opportunities for big-time fails, I feel like.

[00:32:37] Nad Elias: [00:32:37] Yeah, totally.

[00:32:38] Nick Schenck: [00:32:38] And then what if Match.com uses this type of technology for dating and like how people conversate using Match.com or, you know, any other dating app? Think about the applications there. It’s crazy.

[00:32:49] Nad Elias: [00:32:49] Yeah. Um, do you see AI, and again, from a business development sales perspective, is it bigger in the B2C world, the B2B world, or are the applications sort of the same, or is there one over the other that we might see growing quicker as a result of AI?

[00:33:11] Steve Meier: [00:33:11] You know, I think the B2C world will always be a little bit behind the B2B world relative to consumer data versus business data, where there’s a whole lot of protection around consumer data, and AI needs great data to be effective.

[00:33:29] So, you know, the B2C world is always going to be a little bit behind because there’s going to be troubling aspects to leveraging artificial intelligence on certain data sets that will make the capability. Not to say that it can’t work and isn’t working, but I think just across the board, you will see that limitation keep the B2C world a little bit slower to adopting some of these more advanced capabilities.

[00:33:53] I think generally speaking, both sides, B2B, B2C are always going to have success applying it to back office tasks. So anytime that we can automate a manual task that allows us to do our business in the first place. Any opportunity that we can do to provide some sort of predictive intelligence so we can better understand the health and growth and trajectory of our business, regardless of what type of business you’re in.

[00:34:17] Those things are right and available to everyone. And don’t have a lot of restrictions, but the moment you want to try to interface your artificial intelligence with a real customer, and that interface will make or break the experience or make or break the purchase behavior. You know, that’s where things get a little trickier and there’s less margin of error, or less tolerance for margin of error.

[00:34:40] So, you know, for those reasons you might see different industries spiking, where other ones are maybe slow adopters.

[00:34:46] Nad Elias: [00:34:46] I’m almost surprised to hear you say that. Cause I feel like there might be, it’s just easier to determine consumer behavior, right? I mean, just to make the predictions.

[00:34:55] You know, like Amazon knowing what you want before you want it. Right. It just arrives on your door. And you’re like, okay, maybe I do want these pair of pants. If I don’t, it’s easy to return right. Just send it back. I see how it’s applicable to B2B. It just would seem to me, the behaviors are just easier to predict with consumers.

[00:35:15] Steve Meier: [00:35:15] Yeah. And I think, again, I would put maybe part of that, what you said as a back-office task, how do we better understand our customer? That’s a back-office task. That’s invisible AI, and I’m going to go off and service my customer in a better way. And they don’t know where the intelligence came from, but when it comes to then creating a recommendation, that would be more consumer-facing AI.

[00:35:38] And that’s a great example. But when you think about all the different examples that you can apply, you know, that’s just a very easy, a one-off type of thing. And that’s a great example for B2C companies. Today I think that it’s just from our perspective, I think it’s easier for us to play around and find good use cases for the companies that either want to do back office or more B2B. When it comes to B2C, we have to say, okay, what kind of protections do we have around the data? What type of ethical issues do we have around using this in front of customers? What type of threats, if we get it wrong, like the stakes are just higher, I think, for a lot of reasons. So therefore there’s going to be slightly less opportunity.

[00:36:17] And I think eventually a lot of that will change and the culture will shift and all of a sudden, you know, we’re seeing tremendous application and we can’t get away from it. Right. From a consumer standpoint.

[00:36:29] Nad Elias: [00:36:29] Now you used to be a football coach, right?

[00:36:30] Steve Meier: [00:36:30] I did. Yes. When I was teaching, I also coached.

[00:36:33] Nad Elias: [00:36:33] What thoughts do you have on AI in sports? Is it giving coaches the ability to make, you know, predictions? Is it something, I mean, yeah, I know in the NFL, you’ve got the Bill Belichick’s of the world that have been using data for years.

[00:36:48] Steve Meier: [00:36:48] Yeah, I think, you know, I’m a big fan of the next gen stats that are being provided during our football games. I think that makes the fan experience more interesting. And it’s really compelling to say like, wow, that team has a 2% probability of winning this game, and they come back and win the game. I think that’s kind of fun. And we’re just scratching the surface with things like that. But, you know, a lot of the advanced analytics I think are somewhat new to sports. Um, you know, the NBA has been using a lot of these advanced analytics back in 2015. You know, there was, um, a lot of IBM Watson work that was being done with the Toronto Raptors.

[00:37:27] And I don’t know if you guys are sports fans or anyone listening, but you know, in the early to mid 2000s, the Raptors were not a good team and they have consistently put a quality product on the floor over the last, you know, 5-10 years. And it’s probably not a coincidence that they were using artificial intelligence for roster construction, trying to understand what type of player they needed in a certain threshold of salary band.

[00:37:56] And all of a sudden, it’s all kind of working for them, you know, not to say that was the only thing that they did, but it’s probably not a coincidence. So, you know, we’re seeing AI impact sports on the soccer field. There’s a startup here, I think they have an office here in Austin, but they’re basically using artificial intelligence to monitor tape for soccer games.

[00:38:19] And try to understand basically on-field analytics, how fast is that player moving? You know, are they engaged in the entire game? Is their spacing proper for that particular play? You know, all of this stuff that you have a lot of game film on and coaches spend a lot of time trying to figure that out, and they can do it only to the extent that their human eyes and brains are capable. We now have software that can do that, and that’s a big part of the Amazon recognition product, doing those types of things for sports, for example. So, yeah, there are tons of examples of how AI is working in sports.

[00:38:54] And I think Amazon has just made a huge partnership with the NFL to do a whole bunch of interesting stuff that I’m sure we’ll see rolling out soon.

[00:39:00] Nick Schenck: [00:39:00] Just to wrap up, um, I’m curious when you watch shows, movies, is there any one in particular that you look at and you’re like, I think they do a great job of representing AI or what AI could become? And I hope you don’t say Westworld or Ex Machina.

[00:39:20] Steve Meier: [00:39:20] Yeah. You know, Hollywood generally does a poor job depicting practical AI, because to be honest with you, the AI that’s really driving businesses today is again, invisible and very unsexy. You know, some of the most business impact things that we’ve done is that example prior is that we’re using computer vision to extract key data values from PDF and paper documents. Right? Big business benefit. Not sexy, not going to make a movie anytime soon. You know, another one is, you know in the CSI show where you’re basically, they’re looking at the computer at evidence. It’s kind of fuzzy and they go enhance and then the image is a little clear and they go enhance. So they can actually see what’s in the image, that little piece of clue or evidence.

[00:39:59] You know, that’s something that we did on another project using machine learning and, uh, you know, and that’s not something again that anyone would have acknowledged is like really exciting artificial intelligence, compelling cinema type stuff. What’s interesting is everyone’s trying to paint the future vision.

[00:40:16] And a show that does it really well is a show called Upload on Amazon, which was a bit of a COVID sleeper pick if you were really behind and looking for something new to watch, you know, check out Amazon Prime Video and the show called Upload, which is basically, you know, set 20 years into the future where we can upload our conscious to a hard drive.

[00:40:40] So when we die, our loved ones can still interface with us, right? So it’s a sense of digital immortality, but that’s the central plot, the world they create around it on how artificial intelligence applies, where like convenience stores, there are no people working in convenience stores and it’s all robotics, which actually makes a ton of sense in a COVID future, right? Where we want less contact with people we don’t deem necessary. I could see that happening in the short term. So they do a really good job kind of painting a picture of like practical applications. The conscious they follow around actually has a human form in an actor of the show.

[00:41:16] And they walk into a room and like three pop-up ads pop up on exactly what he was thinking on, what he wanted to eat that day. And he had paid for the pop-up blocker, you know, his family didn’t give him the upgraded plan where there’s no ads to his consciousness. So those recommended ads are all driven by AI, so those are all kind of clever depictions, but, um, you know, surprisingly, the show that no one’s ever heard of is the one that probably does the best.

[00:41:41] Nad Elias: [00:41:41] I remember a movie when I was a kid that, uh, that always comes to mind when I talk about AI is a movie called Short Circuit. It was a talking smart robot and this talking smart robot is Johnny-Five. Johnny-Five was alive.

[00:41:53] Steve Meier: [00:41:53] Johnny-Five.

[00:41:55] Nad Elias: [00:41:55] If you were a kid, it was like the coolest thing ever.

[00:41:57] He got electrocuted. Right. And then he became like, you know, the epitome of artificial intelligence in the 80s. Right. He was just a robot that could think, speak, feel, cry. And then it was so good, they made a sequel, which sucked, but, uh, yeah.

[00:42:13] Nick Schenck: [00:42:13] What was the sequel? Just Short Circuit 2? The sequel?

[00:42:16] Nad Elias: [00:42:16] The sequel was called Short Circuit 2, yeah.

[00:42:18] Nick Schenck: [00:42:18] Okay.

[00:42:18] Nad Elias: [00:42:18] It was all 80s stars like Allie. I can’t remember Allie’s name from The Breakfast Club.

[00:42:22] Steve Meier: [00:42:22] Oh, very 80s.

[00:42:23] Nick Schenck: [00:42:23] Alright, Upload and then Short Circuit. I gotta watch those.

[00:42:29] Steve, it’s been great talking to you and thanks so much for joining. If people want to learn more about KUNGFU.ai where can they get more information?

[00:42:39] Steve Meier: [00:42:39] Absolutely. Just come visit our website KUNGFU.ai. Pretty easy to remember. And if you have any specific questions, you can outreach directly to me, just Steve@KUNGFU.ai.