Warm Welcome And Carl’s Return

SPEAKER_00 0:00

Welcome to another edition of GiveMen Ovation, the Restaurant Guest Experience Podcast. I'm your host, Zach Oates, and each week I chat with industry experts to get their strategies and tactics to help you create a five-star guest experience. This podcast is powered by Ovation, the feedback and operations platform built for multi-unit restaurants. Learn what's actually happening in your restaurants and exactly what to improve while driving revenue. Learn more at ovationup.com. And today we have a first because we have a third. Carl Orsborn joins us today. This is his third time on the podcast. And I am so excited because every time I chat with him, it is just new stuff. He is so ingrained into what will be happening and what is happening in restaurants. And so, Carl, welcome

Invisible’s Mission And AI At Enterprise Scale

SPEAKER_00 0:54

to the podcast. And you're joining us right now, not only as a best-selling author for multiple books, a podcast host, a thought leader, but also as the SVP over at Invisible. So welcome to the podcast again.

SPEAKER_01 1:10

Thank you, Zach. I have to say, out of all the podcasts I join, your introduction is always the most dynamic and energized. If you're not ready and excited to go, I mean, this is not the right podcast because I just love listening to your voice, man. You if you if ovation fails miserably, which it won't, but if it does, if it does, your future in radio is secure, my friend.

SPEAKER_00 1:31

Oh my goodness. If only though, I had the boyish good looks of Carl and the accent. Because by the way, we he did let me know, because I I always think that Carl, for those of you who haven't seen a picture of him, he looks like a good five to seven years younger than me. And turns out he's actually like nine years older than me. And so I asked him what the fountain of youth, and he informed me that it is just growing up without son in England, right?

SPEAKER_01 1:58

That's right. Maybe it's also the lack of children in my life. Maybe that's another factor. I guess I hear you don't get as much sleep as I do. So maybe it's the beauty sleep. Maybe it comes back to that. Not familiar with this. Is this a kind of food? Is this a you've got all you've got the whole family, and then you're running a company, and so you've got two kind of whammies right there, my friend.

SPEAKER_00 2:16

Well, that's why I do it the Utah Way, which is 44 ounces of Diet Coke every, or excuse me, Coke Zero every lunch. So that's how we roll it. So, Carl, for those who don't know, tell me a little bit about Invisible. What is it and what attracted you to this company?

SPEAKER_01 2:32

Yeah, Invisible is a 10-year-old company. It's a scale-up more than a startup. It did something like 130 million in revenue last year. It's been ranked as the second fastest growing AI company in the US. And it's known primarily for training 80% of the world's large language models. So you can think of the Chat GPTs, the Geminis of this world. Somehow these models have had to be built and trained over a number of years to be to the level that they are performing today. So our team were there right at the very beginning, the very first iteration of Chat GPT, Invisible was supporting the Open AI folks. It's a really interesting business because AI, of course, is at the precipice of what is going to change

Why Gen AI Pilots Fail To Scale

SPEAKER_01 3:15

all of our lives right now. And so my role at the business is less on the kind of AI training model side that I just explained, but more about helping consumer-facing businesses with bringing AI into the enterprise. And one of the things I often say to folks is that there's so much hype, there's so much hyperbole about this. And actually, when you look at the data, and Gartner came up with some data recently that said something like 95%, 95% of all gen AI pilots that were launched in the last 12 months have failed to scale. Oh, interesting.

SPEAKER_00 3:49

That sounds like my dating life. How many first dates ended up in second? Not very many. And so I was like, maybe we just give up on it. But it was about finding the right combination, not about just like doing it, right?

SPEAKER_01 4:02

That's it. But also in the context of the AI story rather than the Zacco's love story. Maybe it's true on both accounts, right? You got to start with the end in mind. And actually, a lot of the businesses these days have got a lot of pressure, whether it be from the investor groups, the shareholders, the boards, say, look, we've got to be doing something with AI. And so they bring something in, they deploy something very quickly. And then what happens is the thing isn't really measured particularly well. Maybe it's been built on a shaky data foundation. And there isn't really someone championing it, believing that this is going to be something that can differentiate the business. And so what we do is we're a platform organization

Data Foundation And Custom Workflow Building

SPEAKER_01 4:40

with a series of modules. And we basically structure data in its first instance, with where we take structured and unstructured data, unify it into a platform that basically takes data through a medallion architecture from raw data where it doesn't make much sense to actually cleaning it up, to then a place where we actually can derive insights. And then we take that into a module called Atomic, which is a process builder. So you think of any digital workflow, whether it be invoice management, whether it be onboarding a new merchant or procurement or some form of closure to your financial books, any of those digital processes that exist, there are lots of steps. But here's the thing that's quite unique about us in the sense that we are not a platform that works off the shelf. We custom build everything around an organization's current tech stack. So we're not about replacing a CRM or an ERP or anything like that. We're about working alongside it. So Atomic is a process builder that outlines the steps necessary in an optimal fashion to work alongside the current tech stack. But then also we've got another module, which is an Radial Marketplace, which has something like 20,000 people around the world that are working to help us either train models or play a role where human in the loop is necessary alongside those digital workflows. And so that is kind of important because anyone that's selling you anything which says AI does it all and you don't have to worry about any humans, that's also one of the fallacies. Humans are still very, very important as much as anything for risk mitigation and quality assurance.

SPEAKER_00 6:10

And I love, by the way, what Tony Robbins says about that. He goes, AI isn't going to replace you. Someone that understands AI is going to replace you. Right.

SPEAKER_01 6:17

Yeah.

SPEAKER_00 6:17

Well, Tony stole that from Harvard, and they know a thing or two, don't they?

SPEAKER_01 6:20

So yeah.

SPEAKER_00 6:21

Um, Tony, do you mean Tony Robbins didn't come up with all his own original material? Come on now, Carl.

SPEAKER_01 6:28

He probably said it better than him with that with his booming voice, right? I mean, he was probably far more compelling.

Human In The Loop And Agentic Workflows

SPEAKER_01 6:34

Uh but uh so you build out these workflows, but then you've got this piece that's getting everyone's attention right now, and that's uh our module is called Axon, but it's the the agentic workflows. And this like, you know, if I was picturing this business to you right now, Zach, I'd be bringing up this demo that we have, which we we show of an insurance claim. It's just so dynamic to watch it happen in front of you. But imagine you're running, I don't know, like say an insurance business, right? And you've got uh you're doing all the claims, and we know what it's like. Everyone's had a car accident, the length of time it takes to be able to get your money back after a car accident. Well, what we're doing with the gentic workflows is helping all of those workflows happen concurrent to each other. So in one corner, you'll see the video of the car accident happening, the other you'll see the 911 call being analyzed, the other you'll see the actual website being looked at for the regulance of an accidents happening on that particular intersection. And all of these things are working together as specific agents to then be able to tell an overarching story where then these agents uh collaborate with each other and tell a human, look, should we approve this claim or not? So you think about it from that angle. We get to a place where you then say, well, if we can start to apply AI in this world, then we move it away from the conversation that many people are having right now. And it's the thing that causes most fear, and that is, is AI going to take my job? And I think it's a really interesting question because they're of course going to probably be disruptions in the same way as before the dishwasher existed, there were people washing all the dishes. And then what happened? It's not like we lost people in our restaurants, right? They went and did other roles.

SPEAKER_00 8:06

Yeah, like I was actually just watching a show that was taking place in the early 1900s or the yeah, like the 1920s, and it showed like all these switchboard operators, like rooms full of these people that were switchboard operators and like they moved on and they found other jobs. There's still more jobs to be done, and the fact that the phones created then more jobs of receptionists, right? That wasn't really a thing before.

SPEAKER_01 8:32

Absolutely. Yeah. And if you approach AI, I mean, in fact, I always say my biggest challenges in this business are either the inertia within the enterprise customers to see where this can potentially head or their willingness to take the gambling, moving in this direction, or the imagination to see what this really can do. Because if you think about this through the lens of what can I do with 10 less people in my team, then you're missing the opportunity of thinking, what would I do if I had a thousand more people in my team? Oh, oh. Right? And when you look at and when we put this to a big dairy farmer up in your part of the world, and if he'd have thought about it as to how to run his farm with 10 less people, we would have never got to where we got to. Because we said, if you can run your business with a thousand more people, where would you put those thousand people?

Jobs, Disruption, And Reframing AI’s Value

SPEAKER_01 9:20

And he said, Well, I care about the welfare of my cows, and what I'd really want to be able to do is have more eyes on my cows. So we've developed a computer vision swarm drone kind of division, which is basically assessing these cows to see are they eating correctly? Are they displaying some early signs of infection? So that way he can then turn his human team onto caring for those cows faster and stopping there from being larger issues and ultimately improving the yield of his kind of farming activity. Okay. That's the kind of thing you can do with a thousand more people, right? So that's part of the challenge in my role, Zach, is helping people just open their eyes to different perspectives of how AI really is going to transform and ultimately build differentiation into their offering.

SPEAKER_00 10:02

So I'm a brand and I look at all these other huge things that were supposed to be NFTs, I look at the segue, I look at these huge promises of this is going to be the next frontier, and it didn't quite match up. What is the difference? How do we prevent AI from being a pun intended segue into a flop versus a gateway into the future?

SPEAKER_01 10:30

Yeah, that's a good question, right? Because we've we've lived through a few in in recent years, and you mentioned a few there. But I think the evidence ultimately comes from seeing it in action and seeing folks actually starting to deploy it. And we run across a number of different sectors, including the public sector, and certainly industries that you could say are probably a little further advanced than the restaurant space. But I think when you do that, you start to see some of the things that are being put in place right now and actually have been for a little bit now, which are really driving forward adoption in their space. So I don't think this is a question of if, but more a question of when. And the reason for that is that ultimately

From Ten Fewer To Thousand More People

SPEAKER_01 11:10

this is something that can allow us to do things at a velocity and at a pace and at a quality that we've never been able to do before. But look, our space is a little different. You and I were doing another video thing recently where we were talking about how does this kind of thing affect the guest experience and how does it really make customers feel about it? And I don't know about you, but anytime I answer one of those phone calls and it's one of those robotic voices telling me to press one or press two, it's not as good as when I'm actually speaking to someone. And I think ultimately where we need to think about AI in the hospitality sector at large is how do we create a better guest experience by allowing the humans that are involved in that to either be better prepared to be able to create a better experience, or indeed to be able to ensure that they have the time and the utility to be able to do just that. I I, you know, when I speak to operators, for example, in restaurant groups, I say, I and I haven't met one operator yet that doesn't say to me, the best GMs in my restaurants are the GMs that spend the most amount of time with their team and or the best amount of time with their guests. And so then you ask, well, what about the other 90%? And they say, well, they typically aren't using their time as effectively, or they're being caught up with spending their time in the office, doing tasks in front of the computer, or having to deal with lots of recruitment challenges or things like that. And that's all understandable. That's just operating life. But then that's the question mark I would then put to say, well, what can we do to get the GM out of the office? What can we do to get them with their team and their customers more, right? And that's the kind of stuff that then we allow the humans to do what they do best, and that's creating the best customer experiences.

SPEAKER_00 12:56

Because at the end of the day, it's about connection, right? We could do everything else that we want, but it's about connection. And like if there's a brand that delivers every single time and it's exactly the same,

Avoiding The Next Hype Cycle Flop

SPEAKER_00 13:10

I could trust that brand. But if I have a negative experience with that brand, what am I gonna do? First thing I do is I'm gonna go try someone else and I'm gonna see is there anyone else who can do it better? And if there's that connection, that is what creates true loyalty, more so than anything else that we can do. And the data proves it. In fact, a guest who has experienced real connection, which the best time to initiate that connection is actually right after a negative experience. If you have proper service recovery, that guest is 24 times more valuable than your average guest. They're twice as likely to come back. They spend $5 more on each subsequent visit. Their frequency is two and a half times the average guest, and they're 12 times more likely to leave you a five-star review. So that connection is critical, but now it's so it's about what can AI do to help enable that connection better. And the way that I've always looked at technology in hospitality is like the Tony Stark suit. It's the Iron Man suit that makes Tony Stark a superhero. He's just Tony Stark. He puts on the suit and he's Iron Man. And that's that sympatico right there of human and technology, I think is what we need to do to help our people become the superheroes that they want to be. And then it's about hiring the people who want to be out in the front, who don't want to be at the computer, because those people, technology can do the computer stuff for them. But looking someone in the white of their eyes and saying, Carl, I'm glad you came in, or let me fix that for you, or answering a guest when they have a unique response that a unique complaint that needs to be addressed by a human. Those are things that you need those individuals to do, and that only they can do. And that's how technology

AI For Guest Experience, Not IVR Robots

SPEAKER_00 15:05

helps, not replaces. And I love that idea of flipping it around to how can I do this with less with 10 fewer people, as opposed to what would I do with a thousand more people? I mean, what a powerful question to ask. And I think something that each of us in our restaurant should be looking at and working with our vendors, working with our partners to not only share what you're doing with AI, but also like ask them what's on your roadmap and work with your vendors. Because I'll tell you, every single vendor that you are working with right now is investing heavily into AI. The question is, are you investing into having the infrastructure to adapt and adopt that AI? Because if not, what's the good of having that vendor? You're paying them not just for the service, but you're paying them for the innovation. I've got a whole team of people that are building stuff that no one is paying for yet, and things that we're not even going to be charging for just to make the product better. So, like, I ask everyone, reach out to your vendors and make sure that you're using it because I bet you 95% of restaurants are using half of what they are currently paying for that the vendors have built.

SPEAKER_01 16:20

If not more. Yeah, if not more. I agree. It's and also to what degree are those vendors actually listening to you to be able to adapt their product appropriately to fit your business? One of the big question marks right now that's quite central, I think, to all of this, is we've been blessed in many ways as an industry in many industries that are customer facing to have had SaaS these last 10, 15 years or so, right? Which has enabled a lot of smaller businesses to be able to get access to capabilities that have allowed them to do things that before just impossible for a company of their size. But what's happened also as a result of that is that we've created a situation whereby a lot of these SaaS systems are not talking to each other. And so therefore, what we have in our arms now is a tech stack that isn't

Freeing GMs From The Back Office

SPEAKER_01 17:10

really working in cahoots with each other. One of the big use cases around AI is demand forecasting. But you ask any restaurant executive how many demand forecasts actually exist in your business right now? And many of them will say, Well, I've got one working with my labor thas tool, I've got another thing happening to help me make sure I've got the right inventory in place, my business development team are thinking about this forecast, and the finance team are building the projections from this way. And so you suddenly say, Well, hold on a second, you're operating your business off all these different forecasts. And it makes sense because they're all operating off different systems at different iterations. But the one thing that's gonna make AI fail is that if the data isn't right, and if you take bad data into a fight with good AI, the data will win every single time. And so one of the challenges that I think we're gonna run up against is how much we're gonna find our ability as executives in this industry to really make sure our data can work across channels. And this is more than just APIs, by the way. This is about having a holistic interdependent system which allows a way for these data points to collaborate. So let me give you an example. I have this vision of us building this tool, which I'd love to call something like Smart GM, right? Something which allows us to really build on the hustle of great GMs out there. And today the great GMs have the ability to see, to hear, to have the pulse of exactly what's happening in their operation and make adjustments at that moment. But there are lots of things that they might not be aware of. So let's give you an example. Let's say you're running a restaurant tonight, Zach, and two of your cooks have just called out. One of your refrigerators and one of the predictive sensors in it is suggesting that the fridge might actually go down and fail within the next week. And also there's a snowstorm coming through this evening. Now, those three different pieces of intel are things that the GM sometimes has to adjust to. But imagine

Connection, Service Recovery, And Loyalty Math

SPEAKER_01 19:05

if there was a smart GM AI assistant that can just say, hey, get out of the restaurant industry.

SPEAKER_00 19:11

It's too hard. Is that what your system would do? If it could just say that would be good, right?

SPEAKER_01 19:16

Yeah, no indeed. But imagine if it was able to say, Don't worry, I've hired the technician within the budget constraints that you're after. They're coming tomorrow during your off-peak period. I actually think you need to move Susan across to be double exfo tonight because typically you have higher delivery transactions when we get this kind of weather through, and you'll probably have less people needing to be at the front because it'll be less traffic coming in. And also, I sent a message to the three people in your neighborhood store two miles away who've got three part time workers that have made a request for extra hours. And actually, if you bring them over, that will cover for your shortfall. Imagine that GM having that type of capability and are going, yes, yes, yes, that's really helpful. Or be able to be involved and engage with that saying, actually, I don't want that person. Last time they didn't work well with this person because they were an ex girlfriend. Or whatever. That that type of interaction then, so it's not just a prompt, but it's actually an ability for the GM to be out to engage and to ask for recommendations. That's the world we're heading towards. But you can only do that when you start to tie these data threads together, where these things can start to play good with each other.

SPEAKER_00 20:17

Well, that is so powerful. And Carl, there's so much that we could talk about here. We've just barely scratching the surface here. It's a spoonful out of your ocean of wisdom and knowledge on AI. And so I appreciate you coming on and chatting with us for a little bit. How can people find and follow you if there is by chance someone who doesn't know you?

SPEAKER_01 20:35

Well, uh, always LinkedIn is my home, as many people that do know me know. So reach out to me on LinkedIn. Be mindful of that rather odd CERN I move mine. It's O R S P. And of course, you can uh if you haven't read the books yet, then uh they're both on Amazon, delivering the digital restaurant uh to help you get some ideas as to what's happening in the world of off-premise and technology.

SPEAKER_00 20:54

Awesome. Well, Carl, for making the future where robots take over the world a little less scary, today's ovation goes to you. Thank you for joining us on Given Ovation. Thank you, Zach. Nice to see you. Thanks for joining us today. If you like this episode, leave us a review on Apple Podcasts or your favorite place to listen. We're all about feedback here. Again, this episode was sponsored by Ovation, a two-question SMS-based actionable guest feedback platform built for multi-unit restaurants. If you'd like to learn how we can help you measure and create a better guest experience, visit us at ovationup.com.