Leveraging Data Analytics and AI for Business Success with Richard Langlois, President at IT Architecture and Strategy

Data-Driven Podcast

In this interview Dave Mariani welcomes Richard Langlois, President at IT Architecture and Strategy. Richard discusses his history, being an early adopter of AtScale, and how he achieved significant performance improvements. He delves into the importance of a strong data foundation, including ontology, governance, MDM, and a flexible data analytics platform. Richard emphasizes the need for a well-planned roadmap aligned with business benefits and highlights the potential of AI and innovation in the future of data and analytics.

See All Podcasts Button Arrow


Dave Mariani: Hi everyone, and welcome to another episode of AtScale’s Data-Driven podcast. Today’s special guest is Richard Langlois, and Richard is the president of IT Architecture and Strategy. So, Richard, welcome to the podcast.

Richard Langlois: Thank you, Dave. It’s great to be with you. We have a long story and, with Covid, it was a bit difficult to keep, to keep in touch in a way. But, I saw what you’ve been doing with the podcast. It’s pretty interesting.

Dave Mariani: Okay, so let’s, let’s, let’s talk about the history. So, listeners, Richard and I have known each other since really the beginning of AtScale. So Richard, I think you, you win the prize of being the first customer to go live on AtScale, probably back in what, 2015, 2016 time kind of timeframe. Exactly. and it was back when, AtScale was, the first port that we did was Hadoop. I think you might’ve been a Cloudera shop. Is that right

Richard Langlois: Yeah, quite right. in fact, the funny thing is, I had a growing team at, at that period of time, and, we were using OOP for, for their purpose. Actually. People have expected using OOP for ai. but we were using OOP as a, as a gigantic, sorting machine, for, for, some, database appliance. and it was working well, it’s just that, you know, I was reading about a Duke and I was like, I can do more with that. And in the many years before I was using M P P technology for regular databases, when I see regular databases, I’m talking TB two or recall, for example. and, they had something, that was interesting, that was dealing with aggregates. you know, that, that they were done automatically for you as long as you define them, obviously.

Richard Langlois: And, I recall, I had a guy was quite, bright in the team, and, he was taking some vacation and he was going in California. And I said, look, because you’re going to California, just, you know, depending where you’re going, but if you hear about some technology that can give me the M Q T that or materialized views or those type of technology that I used to have, and I was using M P P again. So I had M P P what I do, but I was missing the aggregate functional team. If you hear about this, tell me. And this is somehow, somehow I think it’s up to you, Dave, and then you, we, we got in touch. And that was the beginning of, an exciting journey because we had amazing result. by using S Scale, we, we had situation where we went from, 10 times for some queries to like 30 times faster because of those aggregate.

Richard Langlois: And, that’s required, an aggregate layer. Now, today, we talk about semantic layer. We talk, we talk about data, data, fabric. It’s still the same thing. You need a definition that is, that is clear. And then you need some, some automation to help you, render good response time. And as you may, as people may, understand here beyond the lines, I, I was at Yellow Pages back then. The volume was quite high. And, so the, because of that performance was important. It was really, coming from, big data and gold, the funny thing with big data, it used to be called V L D B, very large database. Mm-hmm.

Dave Mariani: mm-hmm. . So

Richard Langlois: DB before people relay it to, to, big data. But this is somehow how we started. and, we had great result, like I said.

Dave Mariani: Yeah, no, and that was, you did have big data. so Yellow Pages is like, it was a, a great sort of first customer for us, because first of all, you’re a visionary. and you wouldn’t have, you wouldn’t have gone with a small company like AtScale at the time if you weren’t. and you were really great to partner with. so a lot of the success for AtScale really sort of rests with the part with our early partnership, where you gave us some great feedback on the product and really pushed us to innovate and in innovate really fast to deliver on your use case. So those were some fun times.

Richard Langlois: it, it, it was, in a way, we’re in the same boat, right I mean, we’re trying to achieve. Mm-hmm. you was for I think the product, a market, the new customers and so on. Me, I was really, chasing for, technology that could do the work. And, we know we were early. I mean, it’s pretty tough to be earlier than that, right With a company . But, but somehow, the, the support that your entire team gave. so yes, I do feel that we, we did contribute a bit to the, to the destination of that scale, and very happy about this. ’cause at the end, I think, both company did benefit, and it was great. I mean, I recall, I have to say it like that, you know, I recall I made, I made a presentation to the, it was the, the chief, digital officer of, yellow Pages.

Richard Langlois: Mm-hmm. , and I had slides and I explained a bunch of things and so on. And, I’m not going to talk about, we were using some, software as a service, you know, large company. But, you know, there was no response time. essentially it was dying or taking so much time. So, so I recall at the end, okay, now, I had like 10 minutes left. So let’s do some queries to show, because we build a, a prototype, with that scale at, for the, for the presentation of a another project. And, I recall when we did the query and the response time was like, you know, a few seconds. And, and the guy was like, why You didn’t start with that I mean, for , as soon as, yeah, as soon as he started response time, it was, so I stop talking, just show the, this sensor. That’s what we, so next time I will know, depending, if we have cases like that, you just show the, show the prototype up front, drop the slides, which is a bit what we’re doing today. I’m used to do presentation, which with a bunch of slides, I decided not going to show a single slide and tell a story with you.

Dave Mariani: Yeah. You know, it’s like, I always find that, you know, I do a lot of presentations and a lot of sort of initial meetings, and everybody almost perks up in the demo. I mean, the slides are, you, you gotta have some basic slides just to sort of un help people to understand what you’re doing. But I try to keep ’em brief because people just wanna see the product, and that’s, that really sort of helps ’em understand better than anything else. And you shouldn’t be, you shouldn’t be shy with showing the product. That’s, at least that’s, that’s how, that’s my philosophy. But Richard, you know, you are, you are a real innovator, and, you know, you are a guy who’s, you know, crossing the chasm guy. there’s definitely, an early adopter. So talk to me a little bit about your, your career and sort of how, how’d you end up being in that position. You now, you’re, you know, now you’re consulting for all kinds of companies, but talk to me a little bit about your sort of, path into data and analytics. How’d you get here

Richard Langlois: Oh, boy. I mean, you saw the color of my hair, right . Oh, it could be a long . It, it could be a long, we,

Dave Mariani: We share that.

Richard Langlois: Yeah. but I guess, I’ll, I’ll go at the beginning, real fast though. But, I started as a, as a programmer and, 1984. So it’s, almost 40 years ago. And, from there I ended up doing some database. I was with the first original database, it was called sra. And when I see first original database, because Dr. KD said that it was the first one who was above 50% of being with the original criteria of what is the original database. So again, I guess I started, with something new right there. and the funny thing with, being a D B A, back then, being a data architect, or being a D B A, people didn’t really understand the difference. It’s a bit like today, talking data governance, data privacy. Mm-hmm. , data architecture, data engineering, you know, people were putting things, together.

Richard Langlois: So, but so somehow we were very, early on that, and then I designed, multiple solution for, as a consultant for multiple firms. I, I built, I, the, the thing though, I always did it for one customer. I didn’t do this for a, a, a software that you then sell to bunch of, customers. but I’ve done it for a large, I use, the technology. I’m talking database technology with large customer in the, early eighties. And, and we built, probably one of the first C R m, in fact, before the turn, c r m was, was coined, and we were building a C R M for, the largest mm-hmm. company in Canada. and, we had to be quite, creative because, for them, each mutual fund was taking about six months to be, designed and built.

Richard Langlois: And, I recall I could not, when I was looking at the size, the number of products, the, the, the roadmap from marketing, we were involved at the executive level. So all, I had a lot of good information. I stay there. Actually, I moved to Winnipeg to, spend a lot of time there. So we, we came up with something, I, I spent so many nights trying to find a way to go from, a data model, because I believe in data driven architecture back then to mm-hmm. , I automate the most of it into a standard way. And we came up with something called query, query certain, you know, people can see a data driven way of building, windows and calling windows. And if you need a reference table, it was called a quid. And, if you need a multiple update, we, so I built like five, five, standard design, and it was supported hundreds of tables.

Richard Langlois: Funny thing is, you know, you talk with a c p, there are 50,000 tables, or even more, but with few hundred tables, we built something that was dealing with, a data definition of mutual fund where you can sell them under which registration, ities mortgage, and so on. We built, you know, similar to, investment bank type of product. We built that, and we came up with parties and group and families, and all the stuff you find in the C rmm mm-hmm. . So, I guess is when you’re stuck with a, a challenge. And I was young, and for me, the, the sky was the limit. And so I was trying that, and I always, did that since then. And the N P P was one of the first one, Oop was the first one in Canada introduction, s scale also, I’ve been doing, now I’m fastforwarding ’cause it’s too long, the, the anci here.

Richard Langlois: But, and fast forwarding, one of the thing I’ve, created, I call it the real time lakehouse. And the thing with Lakehouse is when people hear about Lakehouse, they hear about, the, the goal is Google, you’ll see is the merging of a data warehouse with the data lake. Mm-hmm. . Mm-hmm. . so why use multiple technology if you can do everything That’s the thing. And by the way, that was my thinking also, when I went with that scale, give me some definition layer. I’m going to use it for ola mm-hmm. , which nobody was doing, I do for lab back, back then. That’s the other thing, right We’re doing all that. Instead of doing ai, then we did ai. So we were a bit reversed, but we were using the capability anyway. So the point with the real time Lakehouse is that as the real time is important here, Lakehouse, they will say that all your feeds, they could be batch, micro batch, they could be, streaming and, and so on.

Richard Langlois: And technically it’s, it’s, right. That’s the lakehouse. What I want though, is something different. I want a realtime lakehouse for one purpose. You recall about when we started with, with, with the slides, we said slide creates a context, but then you need to show the product. Mm-hmm. thinking here, I need the context. Mm-hmm. . So yes, I could have answer I T P L C feeding me in real time. But if I’m on a manufacturing, if a manufacturing company and I production line, I want to know a few things. Yes, I want to know the speed, pressure, temperature, and those things. This is the i t sensor and so on. But I also want to know who’s operating there, the unit. Mm-hmm. . Mm-hmm. . If it’s haw, it’s six foot one, a bit more than two 50, and maybe have a tendency to move to the left when this product is being manufactured.

Richard Langlois: Okay Now, I need to know what product is being manufactured. This is, let’s talk about n e s, right So there is some m e s in place, and you need to know what’s happening now, not after it’s done. When you put it back in SS a p I want to know, now I want to know it’s Vishal, because guess what, I actually learned that when we do this, there is a 75% chance that Vishal may be hurt because he has a tendency to do, to move this way. And he is a bit tall, let’s say, or too big, or whatever. So the point is that I’m adding sensor on me. I’m adding sensor on the production lines. I know what’s happening. Now, if the flow of data arrive not in real time, and you know, it was haw and you know, it was this product, then you can explain why haw is now, is now unable to work, and you’re paying for that, right

Richard Langlois: So mm-hmm. , it’s after the fact. It’s a bit like doing fraud detection after it happened. You want to stop while it’s being done. I want to stop the machines by sending P L C. So it’s not, oh, just going one way, right It’s also interacting back to your operational system and your manufacturing system to stop. This is one use case. Now, what, what happened with the real time Lakehouse is that the more I did that, the more I create what people will call, well, I call it contextual richness. Mm-hmm. . So contextual richness could, could lead you in different thing. Because when you do ai, you need to do, you need to select the, the features that are relevant, what you’re trying to optimize. If I use AI in a general optimization, problem, but that’s the thing, richness allows you if you do the proper feature selection, to have some use cases that you could not do in a traditional lakehouse.

Richard Langlois: So I’ve been doing that for multiple years, and I call it real time lakehouse. I did present that. And, this is the, so, but all the innovation, starting back with the first regional database. So I’m an early adopter. I don’t know about innovator, but without any doubt, I’m an early, adopter, and I use this technology to solve business problem. And I use the term use cases, but like anything, use cases is one thing. But what is the business case beyond all these use cases So one thing that I borrow, I’m a former, chief architect in, well, I’m still an architect and chief architect in many places, but, I used to do this, as, as a, as a role as an employee. and for me, what was super important was tying up what business wants to do with their benefits, trying to have the right architecture, the right product, the right solution, putting it together and build roadmaps.

Richard Langlois: Well, if you reuse this enterprise architecture concept in the con in, in today’s world of data and ai, I do exactly the same thing. I have a problem, a business, a use case that could be solved using technology or not. I did meet situation where they don’t have data, try to mesh machine learning when they don’t have that. Well, yes, I can still do constraint based operational research, right I, I can put rules to, minimize or maximize something based on constraint that still exists. So we, I know we talk a lot about M L D L and generative and so on, but, pure optimization is also possible. So we use whatever technology using, reference, of an approach, right An enterprise architecture roadmap. But one thing we make sure is that it’s all tied with benefits. And the other thing we make sure is that we know the cost and we line up based on the business benefits.

Richard Langlois: And the other thing, we decouple those, recommendation or changes. Because at the end, a roadmap is a series of, let’s call that projects. Mm-hmm. , the thing with projects is they need to be financed. Sometimes they fail. Sometimes you wait for another project, right When you’re in a large company. So the whole thing with this is you need to tie the change with the benefits you will obtain and the cost. And you need to tie the change with the project. So if a project is down, if a project fails you, your architecture still exists to reach the benefits, but somehow the, the story still exists. If you can, a project, the story is kind of altered, and therefore your plan is, is wrong. I also came up with this and the, you know, project management office, like the idea of a big time because they liked that I was aligning with the business.

Richard Langlois: And then there was a relationship from project programs to that. So the P M O also became a good friend, of us because we were, helping them to do their, the annual budget, right So, so tying everything, enterprise, architecture, planning, use of, technology, but always, always keeping in mind, we’re doing this forge generating value. Value. It’s not just about, well, although it’s important. and improving, revenue could be also cutting expense. We all know that. But it could also be to do, new products, right New product engineering. It could be improving the customer experience. We say this all the time, how about the employee experience How about team experience It

Dave Mariani: Could be mm-hmm. ,

Richard Langlois: I have tons of, examples in, in innovation. When we talk to inno innovation people, they are not necessarily architects. And when we talk to architects, they are not necessarily innovation people. And when you talk to governance people, they are not necessarily architect, and vice versa. So there is a lot of specialties today and lining up, aligning these people. This, this is really what I, my sweet spot this is trying to see. so I guess when you see innovations about that, and I do recall when we compare a bit our history, Dave, you were telling me what you were doing and, and it, there were many things that were similar. So the, the, the, I guess one big difference, you’re in California, means versus me in Montreal, but still, at, at least on the ai, on the, academic ai, Canada is, is, is much better. So I guess if we can find a way to, to be up our, investment, it could be an interesting place. I’m not saying stuff that we just rank three cities in the top, 10%. Well, they’re ing obviously, but the point is it’s not a bad place. And we’re not that far from you guys, , but Joe, well,

Dave Mariani: It’s definitely not a, it’s definitely a great place, but, you know, you sort of, you sort of combine that you, you’re an innovator. you’re an architect, which is really key, because that means you’re a planner and you’re, and you’re, and, and you know how to, how to think ahead when it comes to technology and how to put it to, to use. And it sounds like you are really, you’re, you, you’re almost like a finance guy. You’re a, you are able to put your, your finance hat on because you’ve been talking about R o I and tracking cost of projects and value for projects, and a lot of other people are just talking about that now in terms of, of finops and that whole discipline. It sounds like that’s something you’ve been doing for a while, for a while. So Richard, you know, for, so for, for sure. And so, talk to me a little bit about, you know, to make all this work, you need a good data foundation, right Yeah. to be able to innovate at scale. So talk a, talk to me a little bit about what goes into building a, a good data foundation to enable you to, to, to deliver all this value to, to the business.

Richard Langlois: Yeah, good point. for me, it’s implied, mm-hmm. , but you’re right. if you don’t have foundation, you want, because it’s one thing to innovate, but how can you innovate at scale And for me, innovating at scale is companies have hundreds of processes. Business processes could also be thousands of processes. It depends which business you’re looking. Some business are quite, quite large, large. So, first do you know all the processes That’s one thing, you know, how do you organize the processes Same thing as how do you organize the data and do you understand the relationship within between data and processes at the end process requires an input to be triggered and produce, data as an output. And it creates read, update, delete data when you’re running the process. So there are, tools that I learned in information engineering back in the 90 about how to have a, a capital architecture process architecture as the same time as doing the data architecture.

Richard Langlois: The funny thing, again, you have data expert, you have process expert, somehow there is a wall in between, and it’s, it’s pretty bad. Mm-hmm. So one of the thing, I, I learned in information engineering, this, this is when I was at Texas Instruments, j Martin’s, type of approach. And anyway, so when I, when I learned that, I realized, how, couple they were. So when we talk about the foundation, if you want to do it at scale, it means you need to have an understanding of the way people talk. A lot of people talk much more process slash procedure versus the talk data. It’s still the case today. So you need to be able to go back and forth. So, but in order to do this, and I’m not saying it’s Isha this time. I’m going to say, it’s McKenzie.

Richard Langlois: McKenzie did a store, a study in 2018. What nstitute, great data foundation, only seven, 8%, if I’m not wrong, of the companies. They survey and they survey more than a thousand company, 1 billion US of revenue or more across the world, across industry. And they said they need four pieces of the foundation. They need an ontology now. Mm-hmm. ontology slash taxonomy slash semantic model slash data slash data fabric. Let’s say I stop here. It’s the same, the same area, right It describe the concept, what data is being manipulated. Ontology is one thing. The other piece you need is, it’s all good if you have all data, but if the data is not of, sufficient quality, and quality is not just accuracy, could be coverage. Do you cover just the us but Canada Mm-hmm. has view on it. I saw companies who had good accuracy, but geographic accuracy.

Richard Langlois: They didn’t have coverage. I saw companies aggregated data. So there is many ways to define, I have seven criteria. What is quality It’s not just about accuracy. Mm-hmm. . So governance was the number, I don’t want to say number one, number two, but it, there was ontology that was one. So slash architecture slash D d m, the conceptual enterprise of the model semantic layer. I know it’s not exactly the same thing, but it’s in the same area. Let’s see, governance. Mm-hmm. , what mm-hmm. To make sure proper process, to make sure data is of the, of the quality required to reach the objective. I’m not saying 100% quality all the time. I’m saying the quality required to reach the objective. I don’t want to have 100% of everything perfect. It’s not business savvy to do it. I need to do it where it makes sense.

Richard Langlois: This is where the roadmap I was talking earlier comes to play. Mm-hmm. , understand the business data. Mm-hmm. , as you said, the finops thinking, M d M also mm-hmm. , because you’ll have multiple platform that you’re going to use. There is issues. So M D M is, could be seen as a set of technology that helps you to improve quality. So it goes along with governance, but for them, it was a third, element of the foundation, and then a flexible data analytics platform. Those were the four key ingredient. And then when I talk about the data and analytics platform, you recall when I was talking real time lakehouse So for me, this is an example of a flexible platform. I use, what’s called a vendor architecture, where I have a speed layer, I have a batch layer, and I have a serving layer.

Richard Langlois: I played a bit also on that definition. essentially my, my, my batch layer is quite fast now because , I, I, everything goes through, if I’m talking, Azure, everything goes through, obs or iot obs, and then it goes both in the speed layer to execute some ai, and it goes into the, data lake zone to create the raw data, the, integrated zone and the, con, the dimensional slash consumption zone. And then whatever I put, for speed, it’s in the serving layer. So it’s a subset of the data lake. There are other finance reason for it. I can put a lot of data on sheet storage, lab storage of the lake. Mm-hmm. , what I need to put on the layer is what requires, speed and so on. So again, this is without going into what should be a good lakehouse architecture, and you can do the same thing with Amazon and other, other cloud.

Richard Langlois: But the point is this is, this is how I, I I put the data, foundation, which is based on, you can say, but I was doing that before, you can see the survey. I was just happy to realize that of the successful company, these are the four pieces identified. The only thing I add to it, I like to have a roadmap, fundamental roadmap, which is a plan. You, you said it. Mm-hmm. . Yes. ’cause I plan. So a roadmap is a planning, and it’s the, an actionable planning. Many times people do architecture, people don’t know what the hell they are these guys are doing. But a plan, everybody understand a plan. So, so that’s also, one very important thing. The, in the US government, they did a survey. Architects spent too much time defining what exists. because, so they, they define too much time.

Richard Langlois: They take way too much time on defining everything. Mm-hmm. , but not enough time in doing a strategy where I’m going, what’s the roadmap Because this is the real, output. So one little tip for architects on, on this, call is that guys, if you want to be efficient, you need to line up your roadmap at just, it needs to be finished at the time. There is the budget, the budgeting process, right mm-hmm. , you know, multi months effort where everybody say, I need this. I need this. Well be ready when that happened, because, you will. So my, my, my business case is also useful there, because when you have tons of projects and then there is an executive group here going through, let’s say these project all above this amount, I want to know why, who’s the sponsor for which reason, and what are, what’s the business case

Richard Langlois: You better be ready. And then you better understand the sequence between things for double counting and so on. So that’s how I put, what’s, foundation is the ontology, the governance, the M d n, the platform and the roadmap. Now on the roadmap, I did, I, you heard me before talking about use cases. I have more use cases that I can do now because I have the list of processes, or the architectural processes and the data. so I do like to compile a list. I don’t, I cannot apply AI on everything, by the way. So it depends. Some, there are much better use cases on processes than than other. So I do compile to a list of ai. And in Canada, while I’m here, we have, innovation almost equal, artificial intelligence. There are almost 50 programs for helping company to, to grow by using ai, but it could also be blockchain, by the way, digital twins and so on. But there is a lot about ai. So when you compile those lists, it’s important because you’ll reuse the business case, not only to sell within the corporation. You will reuse that sell to get subsidies from the government to help you invest so that you know, you can grow the company

Dave Mariani: And also, and also invest in, in, in your partners too. I see a lot of, a lot of, our customers. It’s not just analytics and AI internally, but sharing that with their partners and their supply chain to make, you know, their partners more efficient. and, and they, and obviously everybody benefits from that. So, Richard, like that, this is super good advice for our data leaders out there. Everybody who is a data leader should listen to richer because you’ve been through the trenches and you’ve added that you’re a risk taker, first of all. but you’re a risk taker. But when you talk about the roadmap, that’s something that’s much more of a planner. And so you gotta, you can take risks, but I like your approach of, if you’re gonna take the take a risk you got, doesn’t mean you can’t plan. calculate

Richard Langlois: Risk.

Dave Mariani: There you go. There you go. So, so Richard, so, let’s wrap up. But, I always love to ask, any guest on the podcast, you know, what excites you about the future Like, and that future could be, you know, within the next five years, let’s say, what do you think, what do you predict might happen What, how things might change What gets you excited about what the future holds for data and analytics

Richard Langlois: , it’s, it’s not an easy question. because if I knew, if I knew I will be, one of the richest person in the world, right ,

Richard Langlois: It’s one thing to innovate. It’s something else to, to know what’s going to happen, which part is going to work, which timeframe’s going to work sometime, many times I’m too early. And then to realize I should have been more patient and, you know, or the vice versa. I’m using the numbers. I look at price earning ratio, and I never invest, for example, I didn’t invest in, in, Nvidia, and I was using their, their, their, their, their G P U for many, many years before people knew about G P U, and now I’m looking at their price. I’m like, why did that Because times, the PE is not, you know, I was burned in 2000, like others. So I guess that’s experience. So my point is very tough to predict. What I want or what I hope is that we’re going to use data and analytics to solve very, very difficult, use case.

Richard Langlois: I’m talking here about air pollution ’cause of wildfire. Mm-hmm. , I’m talking about helping people with healthcare, really helping people. I know people focus on extending life that will be for the 0.0 rule, 1% in the next, whatever number of, of years, decade. But it’ll be nice if we could use data and analytics to improve, make the, the world better, honestly, better for, for all. So that way it’ll be, a global, lift might help in reducing war conflicts and so on. I, I really want, that it goes there. I know I talked, fi finops and I talked business case and so on, but when you ask me for the future, I really hope it’s going to be used for the, for the right thing and helping, overall the, the humanity versus what’s being done right now.

Richard Langlois: But, we need to start somewhere. And as people learn and it becomes, and people do it at scale, I hope this, that’s where it’s going to go. We need news, right We need some innovation that will help everybody. That that’s really, what I’m hoping, what makes me, read these are these articles go to conference and come up with the idea. I hope, I hope I, I can contribute to that. You know, my, they are the data scientists and engineers, by the way. So, so becomes knew somehow, maybe because I influenced them, maybe, I guess as a, as a parent, that’s what I’m hoping

Dave Mariani: For. Hey. Amen. Amen, brother. So, and you’re the guy to, you’re the guy to do it. So, you know, you’re, it takes an innovator to really sort of think of how you can use data and, and a good foundation to really change the world for good. So, with that, Richard, it’s been, it’s been great chatting with you and, to everybody else out there who’s listening, be like, Richard, be an innovator and stay data-driven. So thanks for, thanks for listening. Thanks, Richard.

Richard Langlois: Thank you. Thanks Lee. Thanks all. Yeah.

Be Data-Driven At Scale