Mark This! Podcast, Episode 27: Hospitality IQ
Host: Heather Dotchel (she/her), Corporate Communications
It's time to Mark This, a podcast in which we peel back the corporate curtain to reveal the cool and innovative people, programs, and projects that are happening all over Aramark's varied lines of business.
I'm Heather Dotchel. As a member of Aramark's Communications team, I see and hear amazing things that are happening across our company every day. These remarkable initiatives happen because we have remarkable people behind them. Building opportunity, building innovation, and building community.
Let's talk AI today. You've seen it in the news, you've chuckled at AI-generated photos. You may have even used it yourself, but how does it relate to Aramark? In a nutshell, Aramark is making significant strides in the hospitality industry with innovative use of artificial intelligence-powered applications to enhance the guest experience, empower its operators, and further drive client business objectives.
Meet Aramark's Hospitality IQ. Baker Smith, chief information officer for Aramark, and Vince Miller, Vice President for information technology at Aramark are joining us today to explain how Hospitality IQ is positioning the company to introduce and to explain how emerging technical capabilities will help impact the customer experience and streamline operations.
Welcome to the program Baker and Vince. I'd like to start with some introductions. Baker, can you tell us a bit about your path to your current position?
Sure, Heather, thanks for having us.
How did I get here today? Well, it's been quite a long and winding road. Aramark is my second career That's after retiring from 30 years in technology consulting. So what about that has led me to this role at this moment? It's fundamentally about my passion to take advantage of business challenges or opportunities and to drive those business results through the use of technology. Marrying technology with business results is what I love to do.
So today we're here to talk about some of the latest technologies that can advance Aramark's business objectives and quite frankly, it's just awesome, so thanks for having us.
It is our pleasure. Vince, I'm going to ask the same question of you. So our Mark This audience knows who we are listening to.
[inaudible 00:02:40]. So Vince Miller here. I lead up technology services for our US businesses, so I have responsibility for all of the technologies that we have out in the field and the accounts. But I actually have a background in cyber security. So I came to Aramark in 2020 as our global chief information security, responsible for setting our global security program and maturing us against defenses, especially as sort of Covid started to force more of the workforce mobile and more of a presence on cybersecurity.
So in 2022 I transitioned to the US CIO role responsible for technology. I think that background in cybersecurity though helps well position us to continue to be an innovative organization, but also measured in terms of data privacy and governance. So interesting perspective and balance between Baker and I as we try to lead the technology organization going forward.
I'd like to start with a meta question to orient our listeners. What is AI? I know small question. Vince, can you tackle this?
Sure. So definitely starting small there, Heather, in terms of questions. AI is not a new concept. It's basically computers doing things that historically we've recognized as human tasks and it is recognizing based upon large data sets to execute those tasks.
We at Aramark have been using AI for the better part of a decade, starting with visual check out, so helping customers get through the line faster with technologies that identify that you have a Coca-Cola and a chicken sandwich, to other more machine learning based capabilities with product and pricing analytics.
So over the past several years we've been diving deep into AI, but I think what really has caught the public's eyes and really more enterprise opportunities as well is the rapid development of two specific type of technologies or AI solutions, generative AI and conversational AI. So generative AI is the ChatGPT type of technology that a lot of us are familiar with to create pictures, or emails, or generate a specific product based upon a data set that it's reflecting to. And then the conversational AI is the historical chatbot that we've seen through many iterations has really started to mature and accept different types of media, so that whether that's the actual text, or voice, or sort of a blend of the two.
With those two technologies combining, there's a lot of opportunity for end users and for enterprise to take advantage. And I'm sure we're going to get into a few of those opportunities as we go forward.
But Baker, I don't know if you want to add a little bit more to that as well?
Actually I got a question for you, Heather. You know GenAI is all the buzz and GenAI is based on large language models. Do you know what a large language model is?
Well, let me help you. It helps demystify what AI actually is or GenAI actually is. And actually it's just a prediction. So I'm going to give you an example. This is how I learned it and it just, it's so simple to me. If you say the word "the" and you asked a model or a person to predict the next word, who knows what would come next. But if you say "the sky", you might guess the next word would be "is", right? Pretty common. That is a two-word language model. And if you said "the sky is" the likelihood is the next word, you might say "blue". That is a three-word language model.
So when large language models really became popular with the release of ChatGPT 3.5, that was a 20,000 word language model. So if you just take that simple example and add 20,000 words before it to predict the next word, all language models are doing, are doing a prediction of what's most likely to come next, not something that's right, it's just a prediction and a guess. So that sort of demystifies that it's really not about magic.
However, when you combine that learning of a large language model with other techniques such as machine learning and you throw some human interaction in that, things really get interesting and really get exciting in what we can do. So there's no real magic to this. It's all technology and science. It just feels that sometimes when you ask ChatGPT to write you a poem to your cat in Shakespearean verse and it does it. So anyway, that's a little perspective on what it is and what it's not.
I love that. That makes it very easy to access. There is a lot of fear around AI capabilities in the general public, but it is a powerful tool with a seemingly endless array of practical helpful application.
Baker, how does Hospitality IQ responsibly use these capabilities?
Yeah, that's great. That's a great question. It's on the top of everybody's mind.
The first thing I wanted to address is the overall fear factor, the big fear factor when it comes to AI. And I heard this interesting comment about a year ago, and it stuck with me, and I probably say it a lot. No one should be afraid of losing their job to AI, all the buzz, you're going to lose your job to AI. However, what you probably should be afraid of is losing your job to someone who uses AI really well. And I thought that was really insightful and when we think about how we view AI, is we don't think about AI as a way to replace humans, but actually to enhance humans, to help humans be more productive and be more impactful. And that is directly in line with the way Aramark views our services because we're a hospitality company, we're not a technology company.
But that being said, there are a set of risks that we try to manage at Aramark. And so real quickly, the first one is legal and compliance. This is uncharted territory, unregulated territory, and there's lots of regulations spinning up in Europe, and the US, et cetera, so we have to monitor that and make sure we're staying within compliance.
Secondly though, and probably more impactful is the protection of intellectual property. The one thing everybody needs to understand when they're using AI technologies and particularly GenAI technology, is that anything you put into a public model is now public. So if you've got something you're trying to protect, I would not go to an open ChatGPT model and just put it out there because now it's available for everyone else to consume.
That actually supports the third risk that we try to protect, which is data privacy. It may not be your intellectual property or you may not be stealing intellectual property, but data privacy, we have privy to lots of private information on guests and on clients, and you do not want to take the risk of putting that in any public model, because not only is it available to the public, but it's actually available in a technology that makes the exposure of that much faster and at a much larger scale.
The fourth risk we have to manage is what we call the quality risk. Again, GenAI is based on large language models. Large language models are trained on predictions. They're inherently, based on the information they use to train, is going to be biased, so we have to be conscious that models can be biased. And at the same time, models can also be wrong. The common term that you hear a lot in the market is hallucinations. So hallucinations are errors, may absolutely be there, and then the answers you're getting may be biased. So you as the owner or the user of AI need to always be able to check the results, cross-reference the results. You see that in a lot of tools today, it's inherent in a lot of tools, which is great. It wasn't in the early days.
That's why when we initially went down the track, we took a very balanced approach to managing this evolving technology. First thing we did was we established an AI Governance committee, and that committee is responsible for setting our policy and rules to keep our use of the technology within a set of guardrails. We like to call it do no harm. But at the same time, we also created the AI Exploratory committee, and that committee's point was to promote and encourage all parts of our business to try to take advantage of these technologies, of course, within the guardrails that we set. Both are just as important so that we actually use this technology to Aramark's advantage and quite frankly displace our competitors.
I'm glad to hear all of that. I would also like to say that I say please and thank you to AI when I use it because if there's any shenanigans down the road, I would like them to remember that I was very hospitable to them.
Let's talk about the hospitality sector. Vince, what are the applications there?
Heather, I'm glad that your hospitable with the AI solutions. It's certainly adding to the hospitality culture within Aramark.
As Baker alluded to, that is the core of who we are as an organization. So there's endless applications of advanced technology and AI within the hospitality sector. But at Aramark we don't want to lose that human touch because we are a hospitality first organization. So we're very intentional in first the naming of Hospitality IQ, and actually leading with hospitality, and who we are, and then putting the intelligence behind that, and then the specific technology solutions that we curate and provide to the field.
So our view or strategy around technology in the hospitality sector is focused on three core pillars, our guest, our operators or our teammates, or our clients. So within each of those three parallels, we'll talk about some examples as we go forward, but you could think about our guest as really providing that frictionless experience with various AI solutions that get them the product that they want, when they want and how they want.
Our operators, it's all about making their lives easier, providing solutions in their hand that allow them to run the business and then be out front of the counter and actually speaking with our guests in our consumers, and making a more enjoyable and hospitable environment.
And then our clients really, our clients want to know in the moment the performance of their business and what specific components they can pull to make sure that they're driving the outcomes that they want. A lot of our clients are looking to increase population within their office environment. How can we provide specific analytics to help them drive their results? Others may be financially driven. We have specific solutions that help them achieve their goals, and that's really the goal of Hospitality IQ, curating technology solutions to solve each of those three audiences.
One thing I'd add, Heather, when you think about what are the applications of this technology in hospitality, the first thing is the game all changed with the introduction of GenAI and the large language models, but it's really important to realize that none of this is a magic wand. It's not that special.
But the fact that GenAI is really, really good at summarization, correlations, and natural language processing to a degree we've never seen with all the technologies that have come before, and you put that together with machine learning or classic predictive analytics, with some degree of human intervention, it gets really, really impactful.
And the way we look at pretty much anything we do in our business, you think about there's a process, any given process has multiple steps. Typically, AI in this technology is good at specific steps, but they're probably not going to replace an entire process. So when you think about how to use it or where to use it, think about a holistic process, break it down into the steps, and then attack the steps. I guarantee you there's spots in there where this technology can help make you as a knowledge worker more productive, as a hospitality guest, really drive those experiences, and take some of the drudgery of some of our jobs away from us so that we can focus on the core of what we're about.
So I'd like to talk about a few of the AI solutions that we highlight as part of Hospitality IQ. The first one I'd like to talk about is Mosaic, our proprietary supply chain AI platform.
Baker, who can access this and what does it do for them?
Yeah, let's talk about Mosaic. So Mosaic is one of the early entries into the use of the combination of these technologies that we've done at Aramark, led by Avendra International and their leadership in the way that they use and look at data.
So the first thing is Mosaic starts off as a correlation of all the data of everything we buy. We buy lots, and lots, and lots of stuff from lots, and lots, and lots of different sources. And it was just not humanly possible to correlate all that data on a regular basis. It was at a one-off event. The combination of these technologies has allowed us to correlate that data basically on a daily basis. That gives our teams and supply chain an unbelievable set of information that now can be applied into every step of the process, the steps of the process where they're negotiating contracts with vendors, the steps of the process where they're responding to requests of specific clients or proposals, and particularly the requests of our clients, of our GPOs and how they're making the choices on what to buy.
So Mosaic is actually a correlation of our procurement data and that is now being applied in multiple points of context pretty much in every process that we do in supply chain.
So Baker, what I would add on to it from a Mosaic perspective is the way that it's helping our clients is it's actually helping us achieve some of our ESG goals. It's helping us understand the exact sort of ingredients and products that we're purchasing. And if you think about some of the goals of our clients, a lot of our clients have publicly state ESG goals or they're looking to buy local ingredients from specific suppliers. With insight into the complete supply chain, we're able to be a part of that journey, and it's really helping us differentiate ourselves in the marketplace as well.
Thank you for that further explanation, Vince. And I want to ask you more about another one of our solutions, Culinary Co-Pilot. Culinary Co-Pilot in my understanding is a menu creation and optimization tool. How does it work?
Yeah, sure. So practically speaking, one of the most important things that we do as a global Ford Food service organization is setting the right menu. Running thousands of cafes and food service locations, if we don't have the right quantity, quality, and selection of products, I would say that's very bad for business.
So working under the direction of our business counterparts in the corrections business first, there was a strong belief that we could use advanced technologies to drive better precision in creating menus. Basically what we're doing here is helping the business understand what are the key parameters to set a menu, whether that be financially driven, whether that be nutritional driven, or consumer feedback from surveys that we're constantly doing, and then using various technologies to create "the perfect menu".
I put perfect in quotes because it means different things to different accounts. So when we're creating this technology, the goal was to be very account specific on helping them achieve their goals, not something from the center to standardize our menus across all 2000 locations. Actually quite the opposite, to understand what is important to that account and then drive that. So that's the practical component of what we did. We basically used technology to help our accounts create menus based upon large data sets from supply chain, from weather-based items, from voice of consumer, etc.
Now, technically, how did we actually execute this? There's two core components of technology. First being machine learning. As I mentioned, we have just tons and tons of data across these thousands of locations. We can understand consumer preferences from sales data, we understand supply chain outages, we understand product pricing, etc. So we created a model that takes all that data and makes sense of it so that we understand analytical patterns that we can help make decisions based upon.
The second piece is what I referenced earlier, generative AI. So the generative AI portion is once we have the analytical decisions from the menu preferences, we now can create a menu, something that was taking users hundreds of hours a month to actually understand how much of this specific product do we need to serve our counterparts, etc. We now with the technology, have the ability to create that ideal menu.
And then what's probably the most interesting component is this menu runs every single day so that as there's changes in the landscape, whether it be weather, or whether it be supply chain, or anything else, we're constantly updating that menu to one, serve the client better, to serve our consumers better, but also make sure that we're forecasting accordingly to have the right amount of product out there.
That makes a lot of sense. We've had our corrections people on as well as student nutrition, and they're so highly regulated with the food that they serve. I can see where a tool like this would just be fantastic and really allow them flexibility that they need to do their job the best they can.
Yeah, and Corrections was a good business for us to start. We're fastly moving into our other businesses as well, because as I mentioned, this is not only financially driven, there's multiple other methods or sort of components that we want to take into the model. So next we're going to be rolling this out into our education businesses, and then our workplace and healthcare locations as well.
So lastly, let's highlight Aramark Connected. My understanding is that Aramark Connected is a suite of advanced technologies that focus on streamlining consumer interactions and operational efficiencies, and that our international business uses it quite significantly.
Baker, can you dig into this for Mark This?
Sure, sure. I love Aramark Connected. It was really born out of necessity for a lot of our international markets. And so if you think about it, we have lots of different experiences that we try to give the consumer and we operate in lots of different countries. And one of the challenges we had early was that the same type of experience and the same type of piece of technology would not operate in all the markets we were on.
So we had to create the idea of an ecosystem that was fundamentally focused on the guest, on the consumer. And the idea is as simple as whatever the experience we were delivering to the guests, we wanted it to be a seamless experience for them, that they felt they were getting the best of Aramark, even if they were at the vending machine, or the autonomous store, or the self-checkout, or the cafe, or the Bodega, you name it, all those different experiences, we wanted them connected, hence Aramark Connected, in a seamless integrated environment, and regardless of the technology.
So it's an ingenious set of architecture that our international accounts have spearheaded to bring together the best of the technology, the best vision technology that we have today, the best autonomous store technology we have today, the best conversational commerce technology we have today, but bring that in a unified experience for the guests so that the guests can interact with us the way they want to on the platforms that they want to engage in, and receive product and experiences the way they want. It's really about our way to create that unique differentiated experience for the guest.
And it's just core to what we do because again, we are a hospitality company and we want those experiences to exceed the guest's expectation. And we needed solutions like Aramark Connected that actually takes advantage of everything we know about the consumer and applies it regardless of the channel in which we're delivering services.
One more thing, Baker that I would add to that, and it's probably a secondary benefit, but equally as important to the Aramark Connected concept is look, it's really hard to run a food service location. And oftentimes when we had layer technology at the accounts or in the field, so you have a different autonomous store as compared to a standup checkout and a mobile application, when those solutions aren't talking to each other, it adds additional burden onto our operators or our teammates to actually manage the products, to manage the menu, and manage the pricing.
As a benefit to having this connected ecosystem, now our operators are going to one location to manage their environment as well. Through their mobile phone, they're adding items to their menus that are then propagating out to the various technologies that we have in the field. One, to help from a consistency of experience from the consumer, but also to the operator. Again, back to bringing full circle to the goal of getting our operators out front of the counter to talk to our guests every single day.
And it's been an immense benefit that I think we're getting compliments on regularly, Baker, that are really important and going a long way to help make life just a little bit easier out in the accounts.
So when we look at these solutions we've just explored in a little more detail and the other solutions that come under our Hospitality IQ umbrella, how is this a differentiator? How does Hospitality IQ set Aramark apart from other companies, Baker?
Sure. So again, at the heart of what we are, we are a hospitality experience company. We give great experiences. We want those experiences to be differentiated and meaningful. And our approach toward technology, which is encapsulated in Hospitality IQ, is to use technology to help our people deliver those great experiences. We're not trying to be the next Google or build our next ChatGPT model. That's not who we are, but we actually are laser focused on using all those great technologies to create a differentiated experience.
So I do think we're different from other hospitality companies. I do think it is not about the specific piece of technology, the one-off mobile app or the one-off piece of technology. It's actually how we bring it together, and use intelligence behind each and every one of those experiences to make each interaction with Aramark to be meaningful to our guests, and meaningful to our clients, and meaningful in the way that we support our operators.
So I do think we are different. I know we're different. I'm very proud that we are different. And I think we are helping our operators knock it out of the park every day for our clients and for our guests.
Want to know more about Aramark's Hospitality IQ platform? Visit our newsroom on aramark.com to access more information.
I'd like to thank Banker and Vince for joining us. And thanks to our Mark This listeners for tuning in.
