30 Hours to 90 Seconds: Blue Yonder’s Semantic Layer for Trusted Enterprise AI

Speaker A: Foreign. Hi, everyone, and welcome to another episode of our Data Driven podcast. And today we have, um, some real practitioners who are doing real groundbreaking work in the area of semantic layers. So I want to welcome Brad Lindsey, who's the head of Enterprise data for Blue Yonder. So welcome to the podcast, Brad.

Speaker B: Thanks, Dave. Happy to be here.

Speaker A: And Jeremy Arendt, who is a senior director of analytics engineering at, uh, Blue Yonder. So, Jeremy, good to see you.

Speaker C: Yeah, thanks, Dave. Happy to be here.

Speaker A: All right, so before we get into the gory details of what these two gentlemen are doing, uh, and the work they're doing, uh, in creating semantics for AI, let's, uh, first, first talk a little bit about, uh, each of them. So, Brad, I want you kick it off, tell us about yourself and how you sort of got to be the head of enterprise data at Blue Yonder.

Speaker B: Yeah, thanks. All right, thanks again, Dave, and thanks for having us. Again, thanks for having us on the podcast today. We're living and breathing Semantic layers for the past 18 plus months.

Speaker D: So it's nice to kind of step away and be able to have a conversation and talk through this.

Speaker B: So, um, as you said, I'm the head of enterprise data at Blue Yonder. My team focuses on the Data foundation that supports analytics, governance, and AI across the company. Um, Blue Yonder is a supply chain

Speaker D: software company, and we work with retailers, manufacturers, logistics companies to help them plan

Speaker B: and run their supply chains more effectively. Um, I've been with blue yonder since 2023. I've been working in data since 20 2002, um, which I've seen a ridiculous amount of changes.

Speaker D: Like, we're talking about semantic layers today.

Speaker B: Um, you know, when I started off, we were working in SQL and Microsoft Access, and then it's like, oh, uh, you know, microstrategies comes out and look, we can build dashboards that are interactive and, uh, so definitely seen an insane amount of change.

Speaker D: And, uh, you know, almost 25 years

Speaker B: now that I've been doing this, um, I've worked across multiple industries. So started in financial services, um, worked for, um, marketing company as well on their analytics side, and then moved over into healthcare and to property management and

Speaker D: now in supply chain.

Speaker B: So worked across multiple industries and like

Speaker D: I said, just seeing a ridiculous amount of change. But I've never been more excited than I am really right now, today with just what's possible.

Speaker A: Yeah, it's really, uh, I share your enthusiasm. I am so excited about what we can do now in analytics with AI. Um, and the semantic layer is really a critical piece of infrastructure to make all that happen. Um, so Jeremy, how about you? So how'd you get to where you are today?

Speaker C: Yeah, so I've been at Blue yonder now about 18 months or so. Um, been doing data and analytics for, gosh, about 10 years now. Um, like many people in data, I had a varied path before landing in data work, which included stints in education and playing music for a living and all sorts of things. And um, you know, prior to working at Blue Yonder, I worked with JLL for, ah, about seven years and really focused on the large industrial and tech clients there. Um, at Blue Yonder I found focused primarily on engineering and infrastructure. So, you know, I started my career really doing dashboards and analysis and working very closely in that area. Um, and it was always kind of this, you know, we'd get to a certain point and everything would start to break down when we wanted to scale, when we wanted reusability, when we wanted single source of truth, when we started wanting to get into different ways of interacting with data. And so all these kind of frustrations year over year kept pushing me more and more into the infrastructure, more and more into the engineering side, more and more into how we build and serve and think about the foundations of all of this work. Which, um, got my first exposure to atscale, uh, in a demo, uh, at JLL and poc we did there. And uh, you know, I've been working on it, you know, pretty much every day for the past 18 months, uh, over here at Blue Yonder.

Speaker A: That's awesome. Awesome. Well, uh, so let's just start from the beginning. So, um, uh, from what I understand about, uh, Blue Yonder, you made a real tough choice about, I guess, 18 months ago to say, you know what, let's change our team's charter from building dashboards for the business to building data infrastructure. And I think that's really interesting. I think a lot of people and a lot of companies, uh, out there are sort of contemplating the same kind of movement. So Brad, tell us a little bit about how you made that choice. Um, and, and what drove you to make that transition.

Speaker B: Yeah, absolutely. And you know, I don't think there was really one dramatic moment that led us to the choice. It was really more of like the

Speaker D: accumulation of a lot of signals that started to tell us that the old model was breaking down.

Speaker B: And the thing was, when I joined

Speaker D: Bull Yonder, we had a successful program. Like every team was hungry for data.

Speaker B: We had dashboards everywhere throughout the business and they're being used thousands of times each month.

Speaker D: So really more usage when it came to dashboards, I think, than any company I'd been a part of beforehand.

Speaker B: But underneath it, it was really like

Speaker D: every dashboard was becoming its own version of the truth and every request was

Speaker B: turning into a rebuild. And teams were spending almost more time

Speaker D: debating definitions than team discussing the business

Speaker B: problem itself that we were trying to solve. And the project though, that really I think crystallized this for us was a,

Speaker D: ah, company wide help scorecard initiative. So it's pretty straightforward concept, um, relying on key metrics, targets, trends, rag statuses,

Speaker B: all the things that companies typically put into executive scorecards. And it was interesting because everyone was bought into it, nobody pushed back on the importance of it or the need for it. But the friction came when we tried

Speaker D: to standardize definitions because we'd ask what sounds like a simple question, how do

Speaker B: you define this metric? And we'd get a clear answer. But then we'd look somewhere else in

Speaker D: the company at a dashboard or a slide deck or another report, and we'd

Speaker B: realize that the metric was being calculated

Speaker D: completely differently in other places.

Speaker B: And that was when, at least for me, it became clear that the problem wasn't reporting. It was more of like the absence

Speaker D: of a shared data foundation.

Speaker B: The shift then wasn't just from dashboards to infrastructure.

Speaker D: It was from producing outputs to building a system the company could actually operate on consistently.

Speaker B: And to be clear, dashboards weren't the problem. And even today they still matter.

Speaker D: People still love dashboards like people still love Excel. We still use them every day.

Speaker B: But now where we are is they're

Speaker D: starting, these dashboards are starting to sit on top of governed, reusable semantic models instead of this isolated, isolated logic that's

Speaker B: built independently over and over and over again.

Speaker A: Yeah, I love that. And you know what, we always talk a lot about the technology. I mean, look, we're a technology vendor. Right. But, uh, but when I was talking to Jeremy, and I'll go to you, Jeremy was like, you talked about the people process to be able to drive those, those, the definitions of the metrics.

Speaker C: Yeah.

Speaker A: Um, and to get feedback and maintain them. Can you talk a little bit about just how you did that? Because that's really a people in process, um, not necessarily a technology problem. Right.

Speaker C: One of the big challenges that drove a lot of this, every time we get a lot of requests for dashboards, get a lot of requests for analysis and insight, we only have so many people on the team, so we're always a bottleneck. And so, all right, every organization I've worked with, including Blue Yonder. The next step is like, all right, why don't we open up self service? Like then everybody can help themselves. Well, it's like, all right, well someone in this part of the business wants to use the data. All uh, right, you got to sit down with us for 45 minutes while we walk through all the contacts. Here's this case statement, here's this, here's what you need to know. This rule doesn't apply here. Self service has always played out like that, right. And so as we've started to move towards a semantic foundation, we can have those conversations, but we have them once, right? And we start to encode everything we need, every calculation, all the context around that calculation in one place. Make that available. It enables a type of self service that we've aspired to for years and years and years that wasn't possible. Right. And how do we get those definitions right? When it is a people process. And I also think it's an engineering problem as well. Right. So there's a lot of information that's available in dashboards, Excel workbooks and teams messages. And so we've tried to really start by saying, hey, what do we have available? What's out there? What tools can we use to extrapolate that? Whether that's uh, AI tools or um, you know, the things. And let's kind of build our best guess and then let's bring that to the business and start talking to those owners and say, hey, this is how we see this as defined in these areas primarily. Here's what we think works, it makes sense, here's something to think about. Um, and then we start getting those definitions in sign up. I mean this is something we've really focused on along this journey is we never show up with nothing to any call ever. Like we always do our homework, try to figure out what we have, build what we can with the information we have and go from there. But it's, it's uh, balancing those engineering problems with all those conversations is kind of where a lot of our work is going, um, every day.

Speaker A: And Jeremy, I think I saw actually that you actually assign a business owner to, to a particular metric, is that right?

Speaker C: Mhm, yep. That's the goal. Every key metric needs to be defined and owned by someone within the business.

Speaker A: Yeah, I think I, I know that's, that's may sound obvious, but that's not obvious at all in terms of practice out in the industry.

Speaker D: Wow.

Speaker A: Um, so, so listen up listeners, because uh, that's, that's something that these guys have done and done fantastically well. Um, so let's go back to, you know, it sounded like you had an initiative to have this, um, uh, 360. Right, um, view that maybe drove a lot of this initial activity. But Brad, you know, what did it take to really fund or get support for doing this? Kind of a pivot? Was that the real driver or was there more than that?

Speaker B: There was more than that. And I think that one thing that

Speaker D: helped us significantly in all of this is that Blue Yonder is already a very data forward organization. I mean, from the CEO down, we're very lucky in that there's a strong executive interest in data and analytics.

Speaker B: But I think the shift that we

Speaker D: had to make organizationally was moving from seeing data primarily as an analytics capability to seeing it as foundational infrastructure for how the business operates. So basically we had to frame the problem correctly to be able to get moving with this.

Speaker B: Because executives don't generally fund infrastructure for infrastructure's sake. They fund risk reduction, scalability and operational efficiency.

Speaker D: So we made a point not to position this as a BI modernization effort. We positioned it as a control and

Speaker B: scale problem because really the reality was

Speaker D: that without standardized definitions and governed logic, which is the key to all of this, we're creating operational risk, plain and simple. That means financial reporting becomes harder to trust, auditability is harder to prove. That's when we get into making sure that we have metric owners there and we can track this. And if anything changes, we can see when it changed, who approved that change.

Speaker B: Um, but then once AI started entering the conversation, which was about the time we were kicking this off as well,

Speaker D: like more tools, more functionality became available. It really then became even more obvious that inconsistent data foundations are not going to scale if you don't have standardized governed data, which again is what we keep driving home. Standardized govern data with context, then AI,

Speaker B: you know, all you're doing is just

Speaker D: going to scale inconsistencies and you're going to get wrong answers faster with AI

Speaker B: if it's not sitting on top of this foundation. And once our leadership understood that this wasn't about prettier dashboards, but that it was about creating this trusted operational foundation for analytics and AI, at that point

Speaker D: this whole initiative stopped feeling optional.

Speaker B: Um, and there were reservations.

Speaker D: I mean, absolutely. I think anytime you're moving from decentralized logic to standardization, there's concern about flexibility or speed. But what we're already starting to see is that once this business logic becomes shared and reusable, we See, the conversation change and this is when it shifts to teams spending less time debating definitions and reconciling numbers and more time discussing the business itself. And that's the direction that we want to move the organization towards.

Speaker A: So you said something really important. You didn't position it as a BI modernization project, you positioned it as a governed data foundation project. I mean, and that's like uh, I mean look, uh, I've been doing semantic layers for 13 years now, um, at, at scale. Uh, and you know, that's BI modernization was sort of always the main sort of driver or use case. But things have really changed with AI. Um, um, and so that really was a strategic, really important distinction that you made earlier on. So Jeremy, I mean I think that you found, I think uh, you described to me in a conversation about how you had uh, some use case where somebody needed to do uh, uh, some kind of 30 hour long financial analysis, uh, and you put it into uh, an MCP question using uh, an LLM and got an answer in 90 seconds. So that really sort of is a good sort of proof point that it wasn't about BI modernization, it was about creating that data foundation which allowed you then to do things that you probably didn't even anticipate when you started this journey, which is being able to use AI to automate, uh, the kind of analysis that you mentioned.

Speaker C: Yeah, yeah, no, and that was, it was one of those moments that even caught me off guard. Right. So you know, we have a demo, we do, you know, internally, uh, when we're talking to people about the semantic layer, right. We're real big on, you know, show, not just tell, right? And we often will do a little analysis, connect it to Power bi, do a little analysis, connect to Excel, then we'll pull up an LLM and we'll have that kind of, hey look, it's all the same, same model, same everything. Um, but I had just uh, a side conversation going with the lead engineer on my team about an analysis, got pulled in for a specific client, looking at different products and different schedules and cost and a lot of different components that they were all uh, pulling together across multiple people to kind of create this analysis. Ah, they needed quickly and just the work that went into it. Um, and because we had done so many demos, I just asked, I said hey, can you send me the original ask that you got an email, um, for this sort of analysis you started on. And I mean I literally copy and pasted it into our LLM tool and it spit out this document that I pulled Everybody and said, hey, guys, let's take a look at this. Like, and everyone. I mean, it was, it was. It's just kind of speechless. I mean, they're like, how did you. It's like it just. This is no additional prompting. This is just exactly what you had sent. I just ran it through our governed semantic model. Right. That we had defined and built, connected to our LLM, and here's what we put in. Yeah, it wasn't 100% perfect, ready to ship, but it was about 85%. And we all kind of had that moment where, like, it's changing. Like, this is changing. And it, you know, it, It's. This technology is moving so fast that it continues to surprise even us, the things we're capable of doing. Um, but it's really changed our whole approach to how we. We think about it.

Speaker A: Yeah, it's a real aha moment. That sounds like that was your real aha moment. Um, Brad, it's like, did that. Did Brad. Did that really help with sort of making the case of sort of building this foundation with, you know, with the management team?

Speaker B: Absolutely.

Speaker D: I mean, whenever you have something that you can, you know, tangibly show and it moves beyond words and it gets into action, and we can say, all

Speaker B: right, this is what we've been talking

Speaker D: about for the past X number of months. You've heard us say the word semantic layer over and over, but once you see that reality, I mean, it completely changes it.

Speaker B: And like, even, you know, uh, one thing that we learn pretty quickly is

Speaker D: that most people don't care about data architecture. I mean, honestly, they shouldn't have to.

Speaker B: Right? Uh, most people, present company excluded, don't

Speaker D: sit around talking about semantic layers, architectures, ontologies, govern models. Nobody wakes up asking for a semantic model.

Speaker B: But what people want is, uh, really

Speaker D: reliable answers in the tools they already use, and they want to leverage all the new tools. And that's what they started to see when, When Jeremy was showing what's possible for this. So when I think about how we started, uh, to really sell this or expand it across the organization with the type of language that we're using, when we're framing this whole initiative and this whole project, we intentionally stopped talking about technical concepts like data models and architecture, and we started talking about.

Speaker B: And then to Jeremy's point, not just

Speaker D: talking about, but really showing the outcomes and the language that I think actually resonated with people. It was really simple. It was. You get the same answers everywhere. And when Jeremy talked about this demo that we'll do now, um, we showed that it's same metric, same definition, same logic, whether you're looking at a dashboard, whether you're working in Excel, querying through AI, you know, pick your preferred tool of consumption.

Speaker B: Um, and we also talked a lot

Speaker D: about reducing rework, because like I said,

Speaker B: people across the company were spending huge

Speaker D: amounts of time recreating logic and validating

Speaker B: numbers and reconciling differences between reports. And that friction is expensive. And then to my earlier point, then

Speaker D: AI bursts through the door and accelerates this whole conversation even more.

Speaker B: Because once people started thinking about copilots and agents interacting with data, it became

Speaker D: obviously very quickly that AI only works at scale if the underlying definitions and the logic are consistent. And I think that's where the real shift came in. Because at that point people stopped seeing this as infrastructure for the data team,

Speaker B: a nice little project for us, and

Speaker D: they started seeing it as operational infrastructure for the business.

Speaker B: And it's funny because all that being said, once we were able to shift this conversation and get people on board and just thinking about language again, I can't think of a single meeting, not a single day, but a single meeting

Speaker D: that I've been in over the past six months when we have talked about the semantic layer and this, that, not just Jeremy and I talking about it, but now other people throughout the business

Speaker B: meeting with them daily. And people are starting to kind of

Speaker D: mirror back those conversations.

Speaker B: And now semantic layer is just, you know, like a, uh, known, frequently used

Speaker D: phrase in the company. And I think that's, that's pretty exciting to see where we started where we're having to define this for everybody. Now people are, you know, talking about it just like they talk about any other term.

Speaker A: Well, I can't tell you how happy that makes me feel. That's a, that's like, that's music to my ears.

Speaker D: Uh, you said at scale also.

Speaker B: They also at scale.

Speaker A: Okay, that's great. Look, all I want is semantic layer. If you're long here talking about a semantic layer, I'm fine with that. Um, but you know, what I really loved is like the same answer everywhere. That's awesome. And also about no rework. Uh, because when you think about it in the BI land, I mean, that's what it was without a semantic layer. It was everything. Every dashboard was bespoke and was net new. Um, so, you know, and you also just tying it back to sort of your killer use case. The aha moment. What I found is like, if you had that semantic layer, you could use a tool like tableau and get that. And take 30 hours to do that analysis by dragging and dropping and hunting and pecking and trying to get sort of get to your answers, but get to your answers as a human working with a ui. Like even though tableau is amazing or power bi is amazing, but when you put an LLM on top of it, but you let the LLM be creative and be the analyst, um, and generate its queries and ask its questions in rapid fire succession and then synthesize it and give you the answer, man, that's magic to me. And I don't think that a lot of people understand that when you put an LLM on top of a semantic layer with governed data, how amazing and how accelerating that is, it's not just a faster dashboard, it's not a faster drag and drop, it's completely changed.

Speaker B: It's uh, an answer to a question. Because before people had questions and our

Speaker D: answer was here's a dashboard.

Speaker B: That's not answering a question, that's giving somebody more work.

Speaker D: All right, which tab do I go to?

Speaker A: What fil.

Speaker D: Do I apply?

Speaker B: What chart? That's, that's not an answer.

Speaker D: People can now ask questions and get answers to those really instantaneously.

Speaker A: That's a great way to say it. Um, much better than I did, Brad. Um, uh, so, so back to you, Jeremy. I mean, um, uh, look, there's ah, you know, there's. Let's talk about speed for a second. I mean I've been in your shoes. I've been running analytics for a business like Yahoo and with a lot of different business users with different needs. Um, so, so what did you do? How did you sort of solve the problem of becoming a bottleneck? Um, and if you could talk about just, just what was your, your process people, process and technology that sort of helped you leverage the business and leverage your team to satisfy the needs of the business.

Speaker C: Yeah, yeah. And my, my team is probably sick and tired of hearing me talk about speed because we talk about it like every day. Um, you know, the goal really is, you know, we never stop building, right? We, if we don't have the answers we need, we go out and get them. If we have resources available to us, we use everything we have. Right. Um, and we didn't really start with this big monolithic plan of here's everything we're going to do in the exact order and here's all the things we need. Um, we knew where we were heading, right? We wanted the semantic layer, we knew we needed to connect it to our AI tools. Um, and we know we needed to start in a few key areas of the business, and so we assigned people and we got moving. Right. I think a lot of the success we've had is been based in a few areas. Right. Um, I think one. First and foremost, we have really smart, really talented people, and we're empowering them with the autonomy to run and build and make decisions and get it done. Right. Um, people who have joined are sometimes surprised at how much autonomy they have on this team to build and make decisions. Um, how much trust we have in them to follow our architectural guidelines and understand. We really focus a lot on telling the why story to our own team, on why it's important and get everybody involved in those conversations. And, you know, I'm not sitting here myself building all of these models. Right? Like, it's. It's. The entire team has to be bought into the vision, and then we empower them to go do that. Right. Um, I think the second piece that's helped us move fast and keep quality high is, like I've said earlier in the conversation, we really focus on using every single resource available to us, whether that's a bunch of dashboards that already exist, whether that's Excel spreadsheets, whether that's talking to different people who may be related or not own it. And we really try, you know, the. The people who own these decisions and models and this knowledge are often very busy, hard to pin down. So when we get them, we really honor their time and we make sure we've done our homework and we have something valuable to bring them. Right? Um, yeah. And so, you know, and we. We've, like I've said, we. We didn't wait to figure it all out. So as we. We're literally deploying functionality as we go. Um, if there's something we need, we're working together with y', all, and that's coming in the next version, or we're figuring out as we go and deploying functionality iteratively. And it's. It's allowed us to go really fast and do a lot of really amazing things. And, um, we're learning as we go, and we talk about even failure all the time. Like, we're going to try things that are not going to work, and that's good data. Like, let's use it, and let's pivot and keep going. So, um, it's definitely a much different mindset, I think, traditionally than we used to build infrastructure and traditionally than we use in bi. Like, BI is like, all right, give me a request. Let's talk about the request, let's build the request, let's move on. Uh, we're building things people don't even know they want or need. They have an idea of the kind of things they want to do, but they can't articulate. I need a semantic model with these 17 common dimensions and all this other stuff. We're anticipating that and trying to move in front of them. Right? Um, yeah. So it's a process but it sounds

Speaker A: like you're a great, a great partner, somebody who's great to partner with. Um, you know what I have seen sometimes, in some, in some situations is that before you have a semantic layer and a semantic foundation, really the business has ultimate freedom to do whatever the heck they want. Um, and they do whatever the heck they want. And so when you bring a semantic layer foundation to bear, it does put some constraints on, on the habits that the old habits that they've had. So this is either for Brad or for Jeremy. Have you seen any of that kind of resistance from the business of basically clipping their wings a little bit when it comes to sort of being able to anything goes kind of a, uh,

Speaker B: I mean I think this goes back to, you know, the earlier comments you

Speaker D: guys were making about self service. Like ultimately we want to enable true self service, true governed self service for all of our, the teams that we support.

Speaker B: But you can't enable self service truly

Speaker D: until you start to apply governance. And that means restricting access. That means you can't just let everyone have access to raw data in the data lake because all that's going to do is it's just going to propagate, you know, all of these inconsistencies.

Speaker B: And one good thing is we're now we roll up under the overall security organization within um, Blue Yonder. So we sit side by side with security, which really gives us, I think it just helps show that the trust

Speaker D: is so important in the data and we have to have these controls, we have to have the process in place

Speaker B: to let you go do what you want to do. You'll be able to do it, but

Speaker D: we have to have these controls in place.

Speaker B: And having that message coming from our security organization as well, it gives us

Speaker D: more buy in, more of that leverage.

Speaker A: Yeah, you definitely have error cover there. And especially when you start to unleash headless agents man, that, that matters. You can't just control people anymore. You got um, you got the kind of velocity that you can't control without some kind of governed layer. So um, you guys are one step ahead There, Um, so, I mean this, there's so many good nuggets in for all you listeners out there about how to do this and how to do this. Right, let's um, just finish up with sort uh, of a forward looking, forward looking question for the both of you. So starting with you, Brad, like, where do you think Blue Yonder goes from here? Uh, when it comes to this Semantic first foundation, like, what's the future look like?

Speaker B: Yeah, I mean, I think the important thing is that we don't view the

Speaker D: semantic layer as the finish line in all of this.

Speaker B: Well, like I said before, we view

Speaker D: it as foundational infrastructure. So the real opportunity for us now is embedding this again, govern consistent data directly into how the business operates, not just how it reports.

Speaker B: So what we're moving toward right now

Speaker D: is this model where governed business logic is reusable in this much more embedded way than it ever was before.

Speaker B: So when someone interacts with AI, builds a report, analyzes financial information, triggers a workflow, they're all operating from the same standard governed definitions and semantic context.

Speaker D: And I just keep saying, you know,

Speaker B: this call and also, you know, internally

Speaker D: too, just repeating standard governed definitions.

Speaker B: Um, and again, this all becomes even more important as the company increases the adoption of AI, because as we know

Speaker D: AI, it's probabilistic by nature, but businesses still need deterministic controls around metrics and governance and financial logic.

Speaker B: So that's where we really see the

Speaker D: Semantic first foundation evolving. It's into an operational control plane for all of our enterprise data.

Speaker A: I love that. Jeremy, I'll let you have the last, the last word on this. What are you excited about the future and where are you going to go with this?

Speaker C: Yeah, so I think the semantic layer solves a lot of these really sticky sort of data problems, um, that we've been working on. But it also creates some really new exciting opportunities. Right. And two big ones that we're really focused on. One is what is the user experience for an end user who needs to access the data? Right. How can we make a really delightful, easy to use experience where someone can come in and see a wide variety of assets, Whether that's a semantic model, whether that's a metric, where is that metric, how can I find it in different agents and different dashboards? Um, really we're starting to put a lot of our focus on what does that user experience look like for our average user. Because if we build all this amazing stuff and it's a total pain to navigate, doesn't really matter. It's going to be hard to get adoption. Right. And I think the second part we're really starting to look at now is we're realizing that the descriptions we have on our semantic models for our people users are different than the descriptions we need for our AI users. Right. And how can we think about things like role based context or different things that provide context based on how we're using it or how do we start to train some of that back in? Um, and that's where this starts to really take us to the next level. I know we, we got a lot of work to do to keep going and get that foundation part built but you know, we can't stop uh, thinking about what comes next too. So a lot of really exciting opportunities.

Speaker A: I love it. So uh, you guys have been great. Uh, like I said it's ah, this is very, very actionable advice. Um, you guys have been really successful and in really getting uh, to a level of maturity very very quickly. Um, so here, uh, one of my favorite customers, uh, Brad, Jeremy, favorite people, thanks uh, so much for sharing your journey. This is really, really valuable for everybody out there who is really getting their data ready for AI. Um, so thanks a lot and uh, to all the listeners out there, thanks for listening and stay data driven. Thanks everybody.

Speaker B: Thanks Dave By M. Sam.