Ron Gabrisko of Databricks and Magesh Bagavathi of PepsiCo join host John Furrier for a live interview at the Databricks Data+AI Summit 2026 on theCUBE Research livestream. The discussion examines how large enterprises operationalize data and artificial intelligence, abbreviated AI, with emphasis on lakehouse strategy, governance and agentic automation.
Bagavathi explains PepsiCo’s migration to a unified lakehouse and states the company aims to standardize 85–95% of its data and achieve 99% data quality through Unity Catalog and integrated governance. They describe the roles of ontology and semantic layers and the rollout of PepGPT and PepsiConnect to deliver contextual enterprise AI and accelerate decision making across frontline teams and supply chain execution.
Gabrisko describes how Databricks’ Genie embeds contextual insights into everyday workflows to empower frontline decision making and to enable agentic automation across processes such as accounts payable and autonomous replenishment. They emphasize responsible AI practices and end-to-end observability as essential to operational resilience and trust.
Key takeaways include the strategic move to a single pluggable lakehouse architecture, the importance of integrated data governance for data quality, the practical application of contextual AI to frontline operations, and agentic automation use cases that improve supply chain execution and operational efficiency. The conversation also highlights considerations for implementation, including ontology design, semantic layers, governance frameworks and observability for responsible deployments.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
Databricks Data + AI Summit 2026. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open the link to automatically sign into the site.
Register for Databricks Data + AI Summit 2026
Please fill out the information below. You will receive an email with a verification link confirming your registration. Click the link to automatically sign into the site.
You’re almost there!
We just sent you a verification email. Please click the verification button in the email. Once your email address is verified, you will have full access to all event content for Databricks Data + AI Summit 2026.
I want my badge and interests to be visible to all attendees.
Checking this box will display your presense on the attendees list, view your profile and allow other attendees to contact you via 1-1 chat. Read the Privacy Policy. At any time, you can choose to disable this preference.
Select your Interests!
add
Upload your photo
Uploading..
OR
Connect via Twitter
Connect via Linkedin
EDIT PASSWORD
Share
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
Databricks Data + AI Summit 2026. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open the link to automatically sign into the site.
Sign in to gain access to Databricks Data + AI Summit 2026
Please sign in with LinkedIn to continue to Databricks Data + AI Summit 2026. Signing in with LinkedIn ensures a professional environment.
Are you sure you want to remove access rights for this user?
Details
Manage Access
email address
Community Invitation
Ron Gabrisko, Databricks & Magesh Bagavathi, PepsiCo
Ron Gabrisko of Databricks and Magesh Bagavathi of PepsiCo join host John Furrier for a live interview at the Databricks Data+AI Summit 2026 on theCUBE Research livestream. The discussion examines how large enterprises operationalize data and artificial intelligence, abbreviated AI, with emphasis on lakehouse strategy, governance and agentic automation.
Bagavathi explains PepsiCo’s migration to a unified lakehouse and states the company aims to standardize 85–95% of its data and achieve 99% data quality through Unity Catalog and integrated governance. They describe the roles of ontology and semantic layers and the rollout of PepGPT and PepsiConnect to deliver contextual enterprise AI and accelerate decision making across frontline teams and supply chain execution.
Gabrisko describes how Databricks’ Genie embeds contextual insights into everyday workflows to empower frontline decision making and to enable agentic automation across processes such as accounts payable and autonomous replenishment. They emphasize responsible AI practices and end-to-end observability as essential to operational resilience and trust.
Key takeaways include the strategic move to a single pluggable lakehouse architecture, the importance of integrated data governance for data quality, the practical application of contextual AI to frontline operations, and agentic automation use cases that improve supply chain execution and operational efficiency. The conversation also highlights considerations for implementation, including ontology design, semantic layers, governance frameworks and observability for responsible deployments.
Ron Gabrisko, Databricks & Magesh Bagavathi, PepsiCo
Ron Gabrisko
CRODatabricks
Magesh Bagavathi
Global Chief Data & AI OfficerPepsiCo
search
John Furrier
>> Welcome back everyone to theCUBE's livestream here in San Francisco at Moscone Center for Databricks. Data and AI 2026. This the theCUBE's . I'm John Furrier, your host. Been covering every single year of the Databricks event. Started out really small there, first one, and then it got bigger. 31,000 people plus busting out in the seams. They're going to need a bigger boat for the next year, probably Vegas. We got two great guests here. We're going to break it down. We got the CRO. Ron's here from Databricks. Good to see you, Ron.
Ron Gabrisko
>> Good to see you, John. Thanks for having me.
John Furrier
>> . I can go back and do a little metric of what happened between those years. And Magesh, global chief data officer, AI officer at PepsiCo. Thanks for coming on and sharing, taking the time.
Magesh Bagavathi
>> Well, thanks for having me, John. Thanks for having me, Ron, as well.
John Furrier
>> Ron, first of all, you got a big customer here in Pepsi.
Ron Gabrisko
>> Yeah.
John Furrier
>> They must have a lot of data.
Ron Gabrisko
>> A lot of data.
John Furrier
>> Talk about the relationship with Pepsi.
Ron Gabrisko
>> I mean, listen, we've been working with Pepsi now, partnering with Pepsi for, I think, seven plus years. Been amazing. We love partners that innovate on Databricks, push us to do lots of things, a massive scale of data. They also have a vision on what they want to try to accomplish. And so they've kind of pushed forward in a number of different areas. We were working, how can we give data and insights and predictions even to Pepsi distributors and sales reps and make sure they can maximize what they're putting on shelves and things of that nature. And so they've been a great partner just in general on how do we maximize the value of data and AI for Pepsi and giving us feedback on the product. So it's been amazing.
John Furrier
>> Magesh, talk about your data estate. Scope the order of magnitude. Just share, you have to give the specifics. Just high level, what are we looking at here?
Magesh Bagavathi
>> So PepsiCo is a $95 billion company operating in 200 countries. We've got the largest private fleet in North America. And we've got to service 1.4 billion occasions a day. That's a pretty sizable scale of operations. When you have that kind of sizeable scale of operations, you also need that sizable scale of data that's able to operate. Now challenges with data is fragmented data, non-harmonized data. We've talked about garbage in, garbage out, and just the dysfunctional nature of data where it can actually be completely non-cataloged or non-harmonious. So we actually had a vision of trying to start bringing everything together, which means that also rationalizing our ERPs and really standardizing on our ERPs, which also means standardizing on our master data management. You standardize master data management, you got to write, which is a very, very hard thing to do, specifically when you talk about a company the size and scale of PepsiCo. So standardized platform. Six, seven years back we were 10, 15% cloud. Today was 75% cloud. The entire data backbone is on cloud. So what we decided, this is a from-to for us and this is our strategy, is that we wanted to get to on data lake, one lakehouse that is flexible, Lego block, which is pluggable architecture, that helps us get to cataloging, security, harmonized, cleansed data through a medallion architecture. So the problem always is when I want to actually go from data to insights to action, oh, it's a four-week effort. It's a six-month effort.
John Furrier
>> Got to have meetings.
Magesh Bagavathi
>> Yes.
John Furrier
>> I can't get a meeting for four weeks. We were just talking about that earlier.
Ron Gabrisko
>> Yeah. Right.
Magesh Bagavathi
>> And then you got to put a big technology team behind it, functional partners, I'm like, "No, no, no."
Ron Gabrisko
>> Then once you get it, it's not right.
Magesh Bagavathi
>> Yes, exactly.
Ron Gabrisko
>> They got to redo it.
Magesh Bagavathi
>> So we love talking about one-stop shop. We love talking about click of a button. We can never get there. And then by the time you go from dev, QA prod, it's already a pain. The whole point for us was really how do you get to one data lake that works for us where now it's really truly a click of a button. That was a goal for us.
John Furrier
>> So where are you now? Because the journey, you had to go through all that assessment, you got legacy environments, preexisting conditions, okay, sometimes brittle, but you want to take advantage of that. They're not going away. There's a system of record. We love that. So how did this all come together and where are you right now?
Magesh Bagavathi
>> So the good thing is that you cannot get here without strong tops down mandate. There's just no way. Right? Strong board level support, CEO level support. And I reported to the chief strategy and transformation officer, Athina. That level of support is important. And then really going and pitching to the board that we want to standardize our backbones. Why? Because we can now drive PepsiCo in a very significant fashion for our North Star, which is a 2030 chapter. And we want to fundamentally make this a really future-fit company. So tops down, key piece. Second piece is in terms of where are we with regards to that architecture. By end of this year, 85% of all PepsiCo universe will be in this on standardized, pluggable, flexible, agile lakehouse architecture, which will be 95%-
John Furrier
>> By end of this year?
Magesh Bagavathi
>> End of this year. 95% catalog and with 99% data quality. All measured, certified within this platform as well. With the right security lineage, everything is under Unity Catalog. So we're almost there.
John Furrier
>> Yeah.
Ron Gabrisko
>> They did the hard work first.
John Furrier
>> And governance. Got to get the governance security. Unity is perfect for that. You got open source options, choice. All right. So you're on a transformation journey because now this could be a competitive edge, obviously. The hard work is done. That's the key. No shortcuts to success here in this business. Okay, next question. When you saw what was presented on stage, you got to go, "That GD was pretty cool." So you start to see that. I'm sure you went, "Ah, I got a kid. Okay. We got to get there." Now when you want to go to the next level, as a leader, two questions, what's been the C-suite dynamics? Because it's not just you. It's not just the CSO and the CIO. It's operational leadership. In some cases, CFO and HR is involved because workers are involved, digital workers. So have you seen a dynamic at the C-suite? And two, what's your plans to move the needle on that next journey? Start with this transformation edge with the C-suite first.
Magesh Bagavathi
>> You cannot move forward in this journey without bringing the C-suite along. So the way we've structured our organization, I reported to Athina Kanioura, who's the chief strategy and transformation officer, is technology leaders, business transformational leaders, strategy leaders all reporting to her. So it's a one-stop shop for business transformation. You don't have this thing of IT now saying, "Oh, I've talked to the business, business saying..." It's a one-stop shop from a business transformation. And fundamentally it's coming tops down, but the groundswell is also bottomed up. 320,000 employees are yearning for this. But the good thing is that it's not like this is the start of our journey. We've been in this for the last seven years, as Ron mentioned. So everybody's attuned to the fact that data is important, AI is important. I'll give you an example. Your field sales rep, they have their applications. Their app is the only way in which you can actually go and get product ordered in an organized trade store. Or we've rolled out a B2B app, which is called PepsiConnect in the markets across almost two million customers. That is the only way by which these customers in traditional trade markets like Mexico or Turkey or India order the products. So there is no other mechanisms. So once you've started moving this direction, you start eliminating all of the mechanisms. This is the only way in which you do business.
John Furrier
>> You're enabling value creation to everybody.
Magesh Bagavathi
>> Yes.
John Furrier
>> Specifically when you get into the line of business, share some of your thoughts on how you see that playing out because I'm seeing evidence of once you get the foundation, stuff just starts happening. Can you share your vision on what you see unfolding because you have all kinds of data, distribution data, sales data, productivity data, maybe new apps that might come out. What's your thoughts on that?
Magesh Bagavathi
>> So two ways, right? First is play to your advantage, which is scale, but prioritize effectively. Two is get edge speed and drive outcomes in the field, right? Two priorities. So now when we're talking about scaling and prioritization, there's 40 plus programs that are AI programs, agentic programs, AI programs, but they're also programs that's going to be driving the maximum opportunities for us in execution. So the company is betting on these 40 programs and that's how we're going forward. But at the same time, when we talk about edge speed and frontline execution, it's now bringing the 320,000 employees and really making sure they bought into this, which is where the big movement for us is. So we've rolled out capabilities such as PepGPT, which is the equivalent of GPT for all of our employees. We've rolled out a lot of chat capabilities for all of our employees. Now we're rolling out all of these capabilities. So if I'm in a platform, for example, Savvy is our B2B platform. I want to make sure every bit of knowledge is available for me in the Savvy as a platform. So that's how we're thinking about two speed execution, scale and prioritization, edge and speed.
John Furrier
>> Ron was showing me a demo. I want to go to Ron because... I don't mean to put you on the spot, but I will. Before you came up, he was showing me a demo of Genie and we're both riffing like, "Man, this AI is really helping us out as workers, individuals." He showed me a demo of what he's pulling up. This would have been like weeks of work.
Ron Gabrisko
>> Oh, yeah.
John Furrier
>> This is like a step function change. This is like order of magnitude.
Ron Gabrisko
>> Incredible.
John Furrier
>> Talk about quickly the demo and what that relates to Pepsi and how your customers are doing this because this is not just one customer. If they do the foundation right, the evidence is there.
Ron Gabrisko
>> I mean, think about it. Now we're putting data insights, recommendations, predictions in the hands of literally every employee. So the game changing thing about Genie for enterprise AI is it has the context of your data. So a lot of our bigger customers like Pepsi have done the hard work of, "Okay, how do I learn data in my ontology?"Genie actually does Ontology, right, to understand what's important to my company, which tables, who has what access. But now that I put that power in the hands of every single employee, I can start asking questions like, "What products should I be selling to this particular grocery chain? How can I maximize sales? What promotion should I be running?" You put that power into every single employee's hands, you're going to increase productivity, you're going to increase revenue, versus before you had all that kind of centralized and you'd have to go ask, "Hey, build me a report?" Or, "Here's our dashboard." And then as the business kind of changed, you'd have to update those things, right? Now, you can update those things in real time, put it in the hands of every single employee versus just centralized decision making. It's a huge game changer.
John Furrier
>> Magesh, I want to get your thoughts on agentic because the puzzles are putting together, Databricks put the puzzle together. One of the final pieces comes to life, which is the transactional piece. Analytics, it's a view into the business. It's kind of passive, historical, but transactional is executing the business and you have everything else kind of wrapped up all in one unified piece. That brings in the agent conversation. Genie is here, you're starting to see the ontology work. Agents will do work. What is your vision for agents as you're knowing what you know? Just how do you see it in your mind's eye playing out? Is it supply chain, every business? What use cases? I'm curious because you're there now.
Magesh Bagavathi
>> What's Nirvana for us? Nirvana for us is eventually how do you go from agents where you're actually holding hands or human in the loop, to eventually autonomous where there are components that the agents can actually do autonomously. What does that mean? It means that data to insights to actions with the responsible AI guardrails and security guardrails so that agents can execute. It could be something as easy as, how do I do accounts payable and three-way matching. Easy problem. All the way to how do I ensure that if I actually have excess inventory or I have basically stockouts, how do I ensure that I'm activating those agents are now acting autonomously to ensure the replenishment life cycles. So this is where I think we love Genie because we think it has... And specifically we are now let's say 85, 90% on the Databricks ecosystem. For us now, it's about activating Genie with all the capabilities, including eventually our ontology as well, this semantic graph layer and ensuring the disparate points that we could not connect before we're now able to connect, which means that if I want to drive fulfillment from here with inventory sitting here, I should be able to fulfill it with the lowest amount of cost and the highest amount of velocity for our customers. That's the big unlock for us. So now when we talk about six million retail outlets, we want to ensure that velocity, cost-effectiveness, and really sustainability and outcomes for our sales rep matters. That's why I think that we are so excited with not only the new launches today, but also how we've actually built the platform because we know now going to the future is not miles ahead.
John Furrier
>> You feel good right now.
Magesh Bagavathi
>> We feel good. We feel good about it.
John Furrier
>> What are you most excited about if you have to pick one thing? Leaving San Francisco, what are you most excited about?
Magesh Bagavathi
>> In terms of the products or-
John Furrier
>> Yeah, products and your position, what you're going to take away from this event and go back to the ranch and roll out and woo?
Ron Gabrisko
>> I like that.
Magesh Bagavathi
>> We actually have a chat, a channel going on with me and my teams. There's almost 40 of us are from PepsiCo. We've got a lot of partners as well here. And super excited, right? Ontology is something that we've been playing around for a while. We've been evaluating so many different options. And this connected knowledge semantic layer is something we've been super excited about, but it's always been very narrow to that specific domain or business context. Start thinking about, you can now start connecting-
John Furrier
>> It's not horizontally scaling.
Magesh Bagavathi
>> Yeah, horizontal. Supply chain to commercial, to consumer, to sustainability, to HR, to finance, to people. All of a sudden now you're talking a completely different game than ever before. So ontology is one that we were super excited about. To your point, now starting to bring OLTP with OLAP. And zero ops.
John Furrier
>> Yeah, zero-copy.
Magesh Bagavathi
>> Zero-copy. Oh my God. Those are big unlocks because we run 300,000 pipelines. That's a pretty massive scale.
John Furrier
>> That's huge.
Magesh Bagavathi
>> And so now we have got a huge ops team. It's an opportunity for the ops team to now start having all of these pieces really start coming together in a single pane of glass and automated troubleshooting.
John Furrier
>> I mean, I got to think your staff and the teams are pretty pumped because the manual labor and the time savings, just that alone, psychological boost. Nevermind the productivity revenue opportunities. And that's why I brought up the C-suite, Ron, because we're seeing deep tech, which we're talking about here, kind of start sistering up against C-suite dynamics because if I'm the CFO, I'm leaning into operations saying, "Oh, I will see a little shadow AI. Give me some of that."
Ron Gabrisko
>> AI is the top for CEOs, right? They've realized like, "Now this is transformative not just for my business, but for the whole industry, it's going to be how I can compete." And so top down CEO, CFO, CIOs, CDOs, chief AI officers, that's the heart of how they're going to change and modify their strategy to be able to compete. It's all about the data. How do you attach AI to the context of the data?
John Furrier
>> Magesh, we're going to have the holy grail North Star conversation. As you look at it, one of the things I liked in the messaging was this idea of decision making because it pivots off of... Not pivot. It's piggybacks off of last year's big theme of evaluation, which Jonathan Franklin and I talked at depth last year and this year was talking about what's going on behind the scenes. There's a lot of stuff going on and the app's hitting. But decision making is now coming to the table. So that has to have a lot of things in place. You mentioned a few of them. Security for sure. But governance. I mean, I've been doing theCUBE. It's my 17th year doing this. In the past year since we last chatted, I've said the word governance on camera more in one year than all 15 years combined.
Magesh Bagavathi
>> Yeah. Sure.
John Furrier
>> Why? It's mainstream. It's not master key, master data management stuff. I mean, it's been a category, but the data category is now everything.
Magesh Bagavathi
>> Correct.
John Furrier
>> What's that decision-making vision? Because it's going to come fast. How are you preparing for that?
Magesh Bagavathi
>> So the beautiful thing is we've doubled down on governance almost four years back, right? So as we now really embarked on this one enterprise data foundation, we have five planks. One of the cornerstone of the plank was enterprise data governance. And then along with that couple years back, two, three years back, we are responsible AI as well and AI governance. So it's now data and AI governance really in one focus and one stack. So when we talk about governance, we're talking about multiple pillars of governance. We're focused on really making sure that the data is cleansed, harmonized, catalogued, has technical metadata and business metadata. All of these pieces are cornerstones of governance. And now of course we've got partnership as well with other partners where we now look at the whole thing using AI to say, "What's the quality of our data?" And we've got various metrics that measure it. At the same time, how deep is the governance? What are things like hierarchy and how are we actually maintaining the hierarchy in that? And then most importantly, observability. As we have AI pooling data, how do we ensure that we have the right levels of observability? Are we fully there? The answer is no, but we're in the process of moving there.
John Furrier
>> I know we're running on time and I really appreciate you coming on. You guys are super busy, but I want to get this in. When I watched the keynote from the product manager giving a demo, he talked about the authoritative snippets and he was really honest about, "Hey, sometimes we don't get it right, but we're going to actually do that," which I thought was very honorable and transparent and actually legit, but that's where we're at. Okay. But the question is, if Databricks is the enterprise context layer, does that replace the traditional need of having application, metadata, and business logic? Because you start to see the tea leaves here. If I can get authoritative snippets, I can maybe do some of that metadata work, but the business logic, reasoning. How do you see that? Because humans have to be in the loop, but once you get the business logic, it might evolve in agentic. Just any thoughts on that kind of high level thing, because then you could move to the context layer handles all the business logic. That could grow and evolve over time. What's your reaction to that? Yes. Am I full of it?
Magesh Bagavathi
>> The-
John Furrier
>> Metadata and the business logic and the traditional application, it's kind of been, "Here's my application, here's the business logic and here's the metadata, next application."
Magesh Bagavathi
>> One of the things I loved about what Ali mentioned this morning was, "Yes, is AGI here?" And we were like, "No, AGI is not here." And he said, "Okay, can you do all these things?" And my answer is yes. Then to a large extent, AGI is partially here because humans-
Ron Gabrisko
>> are really smart.
Magesh Bagavathi
>> Yes.
Ron Gabrisko
>> They're smarter than most humans, right?
Magesh Bagavathi
>> Humans can never synthesize all of this context and make sense of it. Now, AI is doing that for you in order to free you to now apply your intelligence more so than just data or knowledge, but really in applying emotional intelligence adaptability intelligence, situational awareness. All of a sudden we are now being forced to become far more aware while we apply now data and business context.
John Furrier
>> All right. So if we took a CUBE time machine back in time, let's pick a year, 2016. Imagine we're there and you saw these demos, would you consider that AGI? Probably. I mean, think 2016 and you saw what's going on right now.
Ron Gabrisko
>> I mean, I was trying to figure out how to do fast dashboards back then, right? I mean, even like machine learning was like state of the art back then, right?
John Furrier
>> To Ali's point on stage, what is AGI? I think that's a holy grail people are chasing.
Ron Gabrisko
>> We're evolving... Yeah.
John Furrier
>> Well, you could categorize it. I mean, it's magic, right? I mean, in 2016 we were transported to here. If you were frozen in time, you'd be like, "Oh my God, what happened? That's magic."
Ron Gabrisko
>> Yeah, totally.
John Furrier
>> Because it is kind of magical and this is just the beginning.
Magesh Bagavathi
>> That's why they call it Genie.
Ron Gabrisko
>> That's right.
Magesh Bagavathi
>> Agentic genie.
Ron Gabrisko
>> It's pretty incredible.
John Furrier
>> Genie in the bottle, get three wish. Unlimited wishes with Databricks. It's all you can eat. Wish .
Ron Gabrisko
>> That's right. .
John Furrier
>> Ron, we've been talking about his company. You got no airtime. So tell us what's going on with you. Business good?
Ron Gabrisko
>> Yeah. I mean, Databricks...
John Furrier
>> Customers happy?
Ron Gabrisko
>> Yeah. I always say this is our Super Bowl. It's been absolutely amazing. We had 31,000 people here. The innovations, I mean, this is an innovation factory. Like I said, I think it's an incredible company. Honestly, once in a lifetime opportunity. The things we're doing with our customers, their visions for what they want to do with AI and data, it's just absolutely incredible. I've never been more excited than now in my 10 plus years here at Databricks. And it's an incredible company, incredible event. So super excited.
John Furrier
>> Congratulations, by the way.
Ron Gabrisko
>> Thank you.
John Furrier
>> I'm just happy for you. Data people are long game players.
Ron Gabrisko
>> Yeah.
John Furrier
>> You know what I'm saying?
Ron Gabrisko
>> Totally.
John Furrier
>> Databricks played the long game. Playing the long game.
Ron Gabrisko
>> Sure. We keep playing the long game.
John Furrier
>> Thank you guys so much for-
Ron Gabrisko
>> Thank you for having us....
John Furrier
>> coming on theCUBE and ending our day here. Appreciate it.
Ron Gabrisko
>> Yeah, absolutely.
John Furrier
>> All right. I'm John Furrier, you're a host of theCUBE. We are live here in San Francisco wrapping up Databricks day one, of course, Data + AI Summit. The trend continues. 31,000 people, community's growing, the suppliers are growing, the customers are growing. AI is going to be the requirement, governance and security to make it happen. And as intelligence comes in, that's when the agents will do their superpower work. Thanks for watching. See you next time.
Ron Gabrisko, Databricks & Magesh Bagavathi, PepsiCo
search
John Furrier
>> Welcome back everyone to theCUBE's livestream here in San Francisco at Moscone Center for Databricks. Data and AI 2026. This the theCUBE's . I'm John Furrier, your host. Been covering every single year of the Databricks event. Started out really small there, first one, and then it got bigger. 31,000 people plus busting out in the seams. They're going to need a bigger boat for the next year, probably Vegas. We got two great guests here. We're going to break it down. We got the CRO. Ron's here from Databricks. Good to see you, Ron.
Ron Gabrisko
>> Good to see you, John. Thanks for having me.
John Furrier
>> . I can go back and do a little metric of what happened between those years. And Magesh, global chief data officer, AI officer at PepsiCo. Thanks for coming on and sharing, taking the time.
Magesh Bagavathi
>> Well, thanks for having me, John. Thanks for having me, Ron, as well.
John Furrier
>> Ron, first of all, you got a big customer here in Pepsi.
Ron Gabrisko
>> Yeah.
John Furrier
>> They must have a lot of data.
Ron Gabrisko
>> A lot of data.
John Furrier
>> Talk about the relationship with Pepsi.
Ron Gabrisko
>> I mean, listen, we've been working with Pepsi now, partnering with Pepsi for, I think, seven plus years. Been amazing. We love partners that innovate on Databricks, push us to do lots of things, a massive scale of data. They also have a vision on what they want to try to accomplish. And so they've kind of pushed forward in a number of different areas. We were working, how can we give data and insights and predictions even to Pepsi distributors and sales reps and make sure they can maximize what they're putting on shelves and things of that nature. And so they've been a great partner just in general on how do we maximize the value of data and AI for Pepsi and giving us feedback on the product. So it's been amazing.
John Furrier
>> Magesh, talk about your data estate. Scope the order of magnitude. Just share, you have to give the specifics. Just high level, what are we looking at here?
Magesh Bagavathi
>> So PepsiCo is a $95 billion company operating in 200 countries. We've got the largest private fleet in North America. And we've got to service 1.4 billion occasions a day. That's a pretty sizable scale of operations. When you have that kind of sizeable scale of operations, you also need that sizable scale of data that's able to operate. Now challenges with data is fragmented data, non-harmonized data. We've talked about garbage in, garbage out, and just the dysfunctional nature of data where it can actually be completely non-cataloged or non-harmonious. So we actually had a vision of trying to start bringing everything together, which means that also rationalizing our ERPs and really standardizing on our ERPs, which also means standardizing on our master data management. You standardize master data management, you got to write, which is a very, very hard thing to do, specifically when you talk about a company the size and scale of PepsiCo. So standardized platform. Six, seven years back we were 10, 15% cloud. Today was 75% cloud. The entire data backbone is on cloud. So what we decided, this is a from-to for us and this is our strategy, is that we wanted to get to on data lake, one lakehouse that is flexible, Lego block, which is pluggable architecture, that helps us get to cataloging, security, harmonized, cleansed data through a medallion architecture. So the problem always is when I want to actually go from data to insights to action, oh, it's a four-week effort. It's a six-month effort.
John Furrier
>> Got to have meetings.
Magesh Bagavathi
>> Yes.
John Furrier
>> I can't get a meeting for four weeks. We were just talking about that earlier.
Ron Gabrisko
>> Yeah. Right.
Magesh Bagavathi
>> And then you got to put a big technology team behind it, functional partners, I'm like, "No, no, no."
Ron Gabrisko
>> Then once you get it, it's not right.
Magesh Bagavathi
>> Yes, exactly.
Ron Gabrisko
>> They got to redo it.
Magesh Bagavathi
>> So we love talking about one-stop shop. We love talking about click of a button. We can never get there. And then by the time you go from dev, QA prod, it's already a pain. The whole point for us was really how do you get to one data lake that works for us where now it's really truly a click of a button. That was a goal for us.
John Furrier
>> So where are you now? Because the journey, you had to go through all that assessment, you got legacy environments, preexisting conditions, okay, sometimes brittle, but you want to take advantage of that. They're not going away. There's a system of record. We love that. So how did this all come together and where are you right now?
Magesh Bagavathi
>> So the good thing is that you cannot get here without strong tops down mandate. There's just no way. Right? Strong board level support, CEO level support. And I reported to the chief strategy and transformation officer, Athina. That level of support is important. And then really going and pitching to the board that we want to standardize our backbones. Why? Because we can now drive PepsiCo in a very significant fashion for our North Star, which is a 2030 chapter. And we want to fundamentally make this a really future-fit company. So tops down, key piece. Second piece is in terms of where are we with regards to that architecture. By end of this year, 85% of all PepsiCo universe will be in this on standardized, pluggable, flexible, agile lakehouse architecture, which will be 95%-
John Furrier
>> By end of this year?
Magesh Bagavathi
>> End of this year. 95% catalog and with 99% data quality. All measured, certified within this platform as well. With the right security lineage, everything is under Unity Catalog. So we're almost there.
John Furrier
>> Yeah.
Ron Gabrisko
>> They did the hard work first.
John Furrier
>> And governance. Got to get the governance security. Unity is perfect for that. You got open source options, choice. All right. So you're on a transformation journey because now this could be a competitive edge, obviously. The hard work is done. That's the key. No shortcuts to success here in this business. Okay, next question. When you saw what was presented on stage, you got to go, "That GD was pretty cool." So you start to see that. I'm sure you went, "Ah, I got a kid. Okay. We got to get there." Now when you want to go to the next level, as a leader, two questions, what's been the C-suite dynamics? Because it's not just you. It's not just the CSO and the CIO. It's operational leadership. In some cases, CFO and HR is involved because workers are involved, digital workers. So have you seen a dynamic at the C-suite? And two, what's your plans to move the needle on that next journey? Start with this transformation edge with the C-suite first.
Magesh Bagavathi
>> You cannot move forward in this journey without bringing the C-suite along. So the way we've structured our organization, I reported to Athina Kanioura, who's the chief strategy and transformation officer, is technology leaders, business transformational leaders, strategy leaders all reporting to her. So it's a one-stop shop for business transformation. You don't have this thing of IT now saying, "Oh, I've talked to the business, business saying..." It's a one-stop shop from a business transformation. And fundamentally it's coming tops down, but the groundswell is also bottomed up. 320,000 employees are yearning for this. But the good thing is that it's not like this is the start of our journey. We've been in this for the last seven years, as Ron mentioned. So everybody's attuned to the fact that data is important, AI is important. I'll give you an example. Your field sales rep, they have their applications. Their app is the only way in which you can actually go and get product ordered in an organized trade store. Or we've rolled out a B2B app, which is called PepsiConnect in the markets across almost two million customers. That is the only way by which these customers in traditional trade markets like Mexico or Turkey or India order the products. So there is no other mechanisms. So once you've started moving this direction, you start eliminating all of the mechanisms. This is the only way in which you do business.
John Furrier
>> You're enabling value creation to everybody.
Magesh Bagavathi
>> Yes.
John Furrier
>> Specifically when you get into the line of business, share some of your thoughts on how you see that playing out because I'm seeing evidence of once you get the foundation, stuff just starts happening. Can you share your vision on what you see unfolding because you have all kinds of data, distribution data, sales data, productivity data, maybe new apps that might come out. What's your thoughts on that?
Magesh Bagavathi
>> So two ways, right? First is play to your advantage, which is scale, but prioritize effectively. Two is get edge speed and drive outcomes in the field, right? Two priorities. So now when we're talking about scaling and prioritization, there's 40 plus programs that are AI programs, agentic programs, AI programs, but they're also programs that's going to be driving the maximum opportunities for us in execution. So the company is betting on these 40 programs and that's how we're going forward. But at the same time, when we talk about edge speed and frontline execution, it's now bringing the 320,000 employees and really making sure they bought into this, which is where the big movement for us is. So we've rolled out capabilities such as PepGPT, which is the equivalent of GPT for all of our employees. We've rolled out a lot of chat capabilities for all of our employees. Now we're rolling out all of these capabilities. So if I'm in a platform, for example, Savvy is our B2B platform. I want to make sure every bit of knowledge is available for me in the Savvy as a platform. So that's how we're thinking about two speed execution, scale and prioritization, edge and speed.
John Furrier
>> Ron was showing me a demo. I want to go to Ron because... I don't mean to put you on the spot, but I will. Before you came up, he was showing me a demo of Genie and we're both riffing like, "Man, this AI is really helping us out as workers, individuals." He showed me a demo of what he's pulling up. This would have been like weeks of work.
Ron Gabrisko
>> Oh, yeah.
John Furrier
>> This is like a step function change. This is like order of magnitude.
Ron Gabrisko
>> Incredible.
John Furrier
>> Talk about quickly the demo and what that relates to Pepsi and how your customers are doing this because this is not just one customer. If they do the foundation right, the evidence is there.
Ron Gabrisko
>> I mean, think about it. Now we're putting data insights, recommendations, predictions in the hands of literally every employee. So the game changing thing about Genie for enterprise AI is it has the context of your data. So a lot of our bigger customers like Pepsi have done the hard work of, "Okay, how do I learn data in my ontology?"Genie actually does Ontology, right, to understand what's important to my company, which tables, who has what access. But now that I put that power in the hands of every single employee, I can start asking questions like, "What products should I be selling to this particular grocery chain? How can I maximize sales? What promotion should I be running?" You put that power into every single employee's hands, you're going to increase productivity, you're going to increase revenue, versus before you had all that kind of centralized and you'd have to go ask, "Hey, build me a report?" Or, "Here's our dashboard." And then as the business kind of changed, you'd have to update those things, right? Now, you can update those things in real time, put it in the hands of every single employee versus just centralized decision making. It's a huge game changer.
John Furrier
>> Magesh, I want to get your thoughts on agentic because the puzzles are putting together, Databricks put the puzzle together. One of the final pieces comes to life, which is the transactional piece. Analytics, it's a view into the business. It's kind of passive, historical, but transactional is executing the business and you have everything else kind of wrapped up all in one unified piece. That brings in the agent conversation. Genie is here, you're starting to see the ontology work. Agents will do work. What is your vision for agents as you're knowing what you know? Just how do you see it in your mind's eye playing out? Is it supply chain, every business? What use cases? I'm curious because you're there now.
Magesh Bagavathi
>> What's Nirvana for us? Nirvana for us is eventually how do you go from agents where you're actually holding hands or human in the loop, to eventually autonomous where there are components that the agents can actually do autonomously. What does that mean? It means that data to insights to actions with the responsible AI guardrails and security guardrails so that agents can execute. It could be something as easy as, how do I do accounts payable and three-way matching. Easy problem. All the way to how do I ensure that if I actually have excess inventory or I have basically stockouts, how do I ensure that I'm activating those agents are now acting autonomously to ensure the replenishment life cycles. So this is where I think we love Genie because we think it has... And specifically we are now let's say 85, 90% on the Databricks ecosystem. For us now, it's about activating Genie with all the capabilities, including eventually our ontology as well, this semantic graph layer and ensuring the disparate points that we could not connect before we're now able to connect, which means that if I want to drive fulfillment from here with inventory sitting here, I should be able to fulfill it with the lowest amount of cost and the highest amount of velocity for our customers. That's the big unlock for us. So now when we talk about six million retail outlets, we want to ensure that velocity, cost-effectiveness, and really sustainability and outcomes for our sales rep matters. That's why I think that we are so excited with not only the new launches today, but also how we've actually built the platform because we know now going to the future is not miles ahead.
John Furrier
>> You feel good right now.
Magesh Bagavathi
>> We feel good. We feel good about it.
John Furrier
>> What are you most excited about if you have to pick one thing? Leaving San Francisco, what are you most excited about?
Magesh Bagavathi
>> In terms of the products or-
John Furrier
>> Yeah, products and your position, what you're going to take away from this event and go back to the ranch and roll out and woo?
Ron Gabrisko
>> I like that.
Magesh Bagavathi
>> We actually have a chat, a channel going on with me and my teams. There's almost 40 of us are from PepsiCo. We've got a lot of partners as well here. And super excited, right? Ontology is something that we've been playing around for a while. We've been evaluating so many different options. And this connected knowledge semantic layer is something we've been super excited about, but it's always been very narrow to that specific domain or business context. Start thinking about, you can now start connecting-
John Furrier
>> It's not horizontally scaling.
Magesh Bagavathi
>> Yeah, horizontal. Supply chain to commercial, to consumer, to sustainability, to HR, to finance, to people. All of a sudden now you're talking a completely different game than ever before. So ontology is one that we were super excited about. To your point, now starting to bring OLTP with OLAP. And zero ops.
John Furrier
>> Yeah, zero-copy.
Magesh Bagavathi
>> Zero-copy. Oh my God. Those are big unlocks because we run 300,000 pipelines. That's a pretty massive scale.
John Furrier
>> That's huge.
Magesh Bagavathi
>> And so now we have got a huge ops team. It's an opportunity for the ops team to now start having all of these pieces really start coming together in a single pane of glass and automated troubleshooting.
John Furrier
>> I mean, I got to think your staff and the teams are pretty pumped because the manual labor and the time savings, just that alone, psychological boost. Nevermind the productivity revenue opportunities. And that's why I brought up the C-suite, Ron, because we're seeing deep tech, which we're talking about here, kind of start sistering up against C-suite dynamics because if I'm the CFO, I'm leaning into operations saying, "Oh, I will see a little shadow AI. Give me some of that."
Ron Gabrisko
>> AI is the top for CEOs, right? They've realized like, "Now this is transformative not just for my business, but for the whole industry, it's going to be how I can compete." And so top down CEO, CFO, CIOs, CDOs, chief AI officers, that's the heart of how they're going to change and modify their strategy to be able to compete. It's all about the data. How do you attach AI to the context of the data?
John Furrier
>> Magesh, we're going to have the holy grail North Star conversation. As you look at it, one of the things I liked in the messaging was this idea of decision making because it pivots off of... Not pivot. It's piggybacks off of last year's big theme of evaluation, which Jonathan Franklin and I talked at depth last year and this year was talking about what's going on behind the scenes. There's a lot of stuff going on and the app's hitting. But decision making is now coming to the table. So that has to have a lot of things in place. You mentioned a few of them. Security for sure. But governance. I mean, I've been doing theCUBE. It's my 17th year doing this. In the past year since we last chatted, I've said the word governance on camera more in one year than all 15 years combined.
Magesh Bagavathi
>> Yeah. Sure.
John Furrier
>> Why? It's mainstream. It's not master key, master data management stuff. I mean, it's been a category, but the data category is now everything.
Magesh Bagavathi
>> Correct.
John Furrier
>> What's that decision-making vision? Because it's going to come fast. How are you preparing for that?
Magesh Bagavathi
>> So the beautiful thing is we've doubled down on governance almost four years back, right? So as we now really embarked on this one enterprise data foundation, we have five planks. One of the cornerstone of the plank was enterprise data governance. And then along with that couple years back, two, three years back, we are responsible AI as well and AI governance. So it's now data and AI governance really in one focus and one stack. So when we talk about governance, we're talking about multiple pillars of governance. We're focused on really making sure that the data is cleansed, harmonized, catalogued, has technical metadata and business metadata. All of these pieces are cornerstones of governance. And now of course we've got partnership as well with other partners where we now look at the whole thing using AI to say, "What's the quality of our data?" And we've got various metrics that measure it. At the same time, how deep is the governance? What are things like hierarchy and how are we actually maintaining the hierarchy in that? And then most importantly, observability. As we have AI pooling data, how do we ensure that we have the right levels of observability? Are we fully there? The answer is no, but we're in the process of moving there.
John Furrier
>> I know we're running on time and I really appreciate you coming on. You guys are super busy, but I want to get this in. When I watched the keynote from the product manager giving a demo, he talked about the authoritative snippets and he was really honest about, "Hey, sometimes we don't get it right, but we're going to actually do that," which I thought was very honorable and transparent and actually legit, but that's where we're at. Okay. But the question is, if Databricks is the enterprise context layer, does that replace the traditional need of having application, metadata, and business logic? Because you start to see the tea leaves here. If I can get authoritative snippets, I can maybe do some of that metadata work, but the business logic, reasoning. How do you see that? Because humans have to be in the loop, but once you get the business logic, it might evolve in agentic. Just any thoughts on that kind of high level thing, because then you could move to the context layer handles all the business logic. That could grow and evolve over time. What's your reaction to that? Yes. Am I full of it?
Magesh Bagavathi
>> The-
John Furrier
>> Metadata and the business logic and the traditional application, it's kind of been, "Here's my application, here's the business logic and here's the metadata, next application."
Magesh Bagavathi
>> One of the things I loved about what Ali mentioned this morning was, "Yes, is AGI here?" And we were like, "No, AGI is not here." And he said, "Okay, can you do all these things?" And my answer is yes. Then to a large extent, AGI is partially here because humans-
Ron Gabrisko
>> are really smart.
Magesh Bagavathi
>> Yes.
Ron Gabrisko
>> They're smarter than most humans, right?
Magesh Bagavathi
>> Humans can never synthesize all of this context and make sense of it. Now, AI is doing that for you in order to free you to now apply your intelligence more so than just data or knowledge, but really in applying emotional intelligence adaptability intelligence, situational awareness. All of a sudden we are now being forced to become far more aware while we apply now data and business context.
John Furrier
>> All right. So if we took a CUBE time machine back in time, let's pick a year, 2016. Imagine we're there and you saw these demos, would you consider that AGI? Probably. I mean, think 2016 and you saw what's going on right now.
Ron Gabrisko
>> I mean, I was trying to figure out how to do fast dashboards back then, right? I mean, even like machine learning was like state of the art back then, right?
John Furrier
>> To Ali's point on stage, what is AGI? I think that's a holy grail people are chasing.
Ron Gabrisko
>> We're evolving... Yeah.
John Furrier
>> Well, you could categorize it. I mean, it's magic, right? I mean, in 2016 we were transported to here. If you were frozen in time, you'd be like, "Oh my God, what happened? That's magic."
Ron Gabrisko
>> Yeah, totally.
John Furrier
>> Because it is kind of magical and this is just the beginning.
Magesh Bagavathi
>> That's why they call it Genie.
Ron Gabrisko
>> That's right.
Magesh Bagavathi
>> Agentic genie.
Ron Gabrisko
>> It's pretty incredible.
John Furrier
>> Genie in the bottle, get three wish. Unlimited wishes with Databricks. It's all you can eat. Wish .
Ron Gabrisko
>> That's right. .
John Furrier
>> Ron, we've been talking about his company. You got no airtime. So tell us what's going on with you. Business good?
Ron Gabrisko
>> Yeah. I mean, Databricks...
John Furrier
>> Customers happy?
Ron Gabrisko
>> Yeah. I always say this is our Super Bowl. It's been absolutely amazing. We had 31,000 people here. The innovations, I mean, this is an innovation factory. Like I said, I think it's an incredible company. Honestly, once in a lifetime opportunity. The things we're doing with our customers, their visions for what they want to do with AI and data, it's just absolutely incredible. I've never been more excited than now in my 10 plus years here at Databricks. And it's an incredible company, incredible event. So super excited.
John Furrier
>> Congratulations, by the way.
Ron Gabrisko
>> Thank you.
John Furrier
>> I'm just happy for you. Data people are long game players.
Ron Gabrisko
>> Yeah.
John Furrier
>> You know what I'm saying?
Ron Gabrisko
>> Totally.
John Furrier
>> Databricks played the long game. Playing the long game.
Ron Gabrisko
>> Sure. We keep playing the long game.
John Furrier
>> Thank you guys so much for-
Ron Gabrisko
>> Thank you for having us....
John Furrier
>> coming on theCUBE and ending our day here. Appreciate it.
Ron Gabrisko
>> Yeah, absolutely.
John Furrier
>> All right. I'm John Furrier, you're a host of theCUBE. We are live here in San Francisco wrapping up Databricks day one, of course, Data + AI Summit. The trend continues. 31,000 people, community's growing, the suppliers are growing, the customers are growing. AI is going to be the requirement, governance and security to make it happen. And as intelligence comes in, that's when the agents will do their superpower work. Thanks for watching. See you next time.
John Furrier
