Data + AI Summit 2025 | Shanku Niyogi, Databricks & Marc Geall, SAP

Clips
News
More from Data + AI Summit 2025

Shanku Niyogi

VP, Product Management

Databricks

Marc Geall

SVP, Head of Product Technology & Experience, Data & Analytics

SAP

Inside Databricks’ AI revolution: 10 key takeaways from the Data + AI Summit

Artificial intelligence is becoming more intuitive, interactive and embedded, not just in systems, but in the way companies build, think and create. From live coding and autonomous agents to unified governance and automation, Databricks Inc. is moving AI closer to the application layer and aligning it with developer intent.TheCUBE Research’s John Furrier and George Gilbert break down the keynote analysis during Databricks’ Data + AI Summit.“You sort of tell the AI engineer, the software developer, an agent, what you want to do,” said George Gilbert, principal analyst for theCUBE Research. “In the past, all it did was generate code

play_circle_outline Challenges with data pipelines requiring engineering and specialized knowledge in enterprises.

play_circle_outline Integration of Databricks into SAP products enhances user accessibility and functionality.

play_circle_outline Serverless model simplifies Databricks deployment within SAP environments for rapid user access.

Info
Transcript

Shanku Niyogi, Databricks & Marc Geall, SAP

Shanku Niyogi

VP, Product Management Databricks

Marc Geall

SVP, Head of Product Technology & Experience, Data & Analytics SAP

Shanku Niyogi, vice president and product manager at Databricks Inc., and Marc Geall, senior vice president and head of product technology and experience – data and analytics, at SAP SE, join theCUBE’s John Furrier at the Databricks Data + AI Summit 2025. The conversation highlights the strategic collaboration between Databricks and SAP to accelerate enterprise AI and streamline data infrastructure.

Niyogi and Geall discuss how tight platform integration, standardized data governance and unified engineering pipelines unlock business value. They share... Read more

explore Keep Exploring

What challenges do customers face when integrating SAP data into Databricks, and how do partnerships help address these issues? add

What is the approach for integrating AI capabilities with data in the context of Databricks and the business data cloud? add

What changes have been made to Databricks in relation to serverless infrastructure and its deployment process? add

bolt Powered by CUBE AI

Shanku Niyogi, Databricks & Marc Geall, SAP

search

>> Welcome back everyone, theCUBE's live coverage here in San Francisco. I'm John Furrier, host of theCUBE. theCUBE all-day coverage at Databrick's Data Plus AI Summit. I got two great guests here talking Databricks, SAP Shanku is back. Good to see you. Vice president, product management at Databricks Mark SVP, head of technology and experience data Analyst at SAP. Great to see you again.

Marc Geall

>> Good to see you, John.

>> It's been a while. Guys, the world's changed. Talk about the partnership, SAP, and Databricks. What is it about?

Marc Geall

>> Where to start? Right? I mean, I think bringing together two great platforms, really helping customers bring AI together with business content to drive value and we can do that around AI. We can do that around agentics. So I think a great opportunity there.

>> As systems of record become experiences, which you're seeing that's the high-velocity takeaway here is that it's fast. Get that data into generative process as fast as possible. Take me through that. What's the playbook? What are you seeing as a template? A best practice, a blueprint.

Marc Geall

>> I mean from our standpoint it's about data products and governance, but also making product standards. So you think about what we did with SAP products, it was about standardizing processes. We're now standardizing data so that you can enable agentic in a governed way.

>> Shanku, talk about the pipelining of... You're seeing this is a big topic. Yeah, we got workflows they're well-defined. Enterprises know their workflows, they got the process down, but the data doesn't always match that. Is this-

>> Yeah, I think from our perspective when we talk to our customers, they always ask whenever we launch a new product, what about SAP, right? Because all that super valuable data is there, but to get that into Databricks, you have to do these engineering pipelines. They're brittle. You have to have specialized knowledge because you've got to reverse engineer all the business logic. So you've usually got to hire an analyst and then you're stitching all that stuff together. There's drift, there's failure. And in the meantime, we're in a world where you need insights from that data. You need to put agents on that data quickly and so the number one thing we hear from customers is it's killing speed to be... And I think the really beautiful thing about our partnership is let us do the data engineering together between our platforms and shape the products in a way, as Mark says, data products that are curated that you can just take.

Shanku Niyogi

You don't have to copy them. You can delta share them in and get to work on your use case or put an agent on it, get it into that Unity catalog where you can get unified governance, and go build something. That's a no-brainer.

>> What's great about the time we're living in now is that you go to theCUBE 15 years when we were last on almost nine years ago together Mark, Shanku you were on with Steve Orban and I, is that we now live in an ecosystem where cloud kind of set the table for that where companies don't have to beat each other all the time. It's always this nice swim lanes. But with data now, a whole nother operating system level's emerging in the data layer so everyone can play and open table formats and SAP. You guys run businesses. It's not like it can be brittle, the connectors, and how it gets connected, but the data runs the business. They're mission-critical workloads it's not like, "Hey, I got an app, I'm a startup. Can I work for the enterprise?" The enterprise value you guys have and what your presence. The question is, okay, I can create some value, got some Databricks, SAP, I'm a customer. I want to figure out how do I extract the value. Can you guys share your thoughts on those two variables, the creation of value with Databricks and SAP, and the extraction of that value?

Marc Geall

>> Yeah, I think in terms of the creation value, there's a time-to-value element and then there's also a cost-benefit. So by providing these curated data products, customers don't have to spend money on extraction or on recreating the semantics and doing a lot of data engineering. So they get immediacy of value, but also they get reuse, right? Because data products are available for multiple use cases, multiple business problems to solve. So both of those I think drive value from a customer standpoint.

>> What is a curated data problem? Give an example because people might not understand what that means to the level of scope it is.

Marc Geall

>> So what is my working capital within the business? So there's financial information that they need to do. They need to understand what an accounts receivable is, they need to understand what an accounts payable is. We have that all captured within our business applications and now we're enriching that with other data. So there we can leverage some AI use cases from Databricks to now help customers really leverage their working or optimize their working capital cycles.

>> One of the canonical examples I show, it's a demo. You bring in a cash flow data product from SAP, you bring in some external economic data and you play with it in a notebook, and then you use AutoML to create a derived cash flow prediction model and you publish it back into SAP where everyone can use it. And that process would take weeks or months because you had to build all that engineering. And the nice thing about that demo is the thing you don't see, which is there's no data engineering anymore. It's just you share the data in, you work in the notebook, you create a model, you publish it back. And going back to your point, it's like I kind of see those things almost less is two different things. As long as the value that you're creating is actually insights and models and agents and not a pipeline. I think you have to be a plumber to think of a pipeline as a value. And I think our goal is to get that done so that customers can move to the value that matters because-

Marc Geall

>> I think it's a glib statement, right? It's almost democratization of AI, right? Before there was a cost of AI, which was the data engineering, you remove that, you now drive immediate value.

>> Yeah. The leverage you get on this is interesting because the value creation has economies of scale. So you've got the curated data set, I'm building on top of it, I'm not re-plumbing it or doing things with it and then that's extracted. Step one, you got value extract and then you build on that, extract on that so that flywheel is emerging. This is a key point that reuse and derivative works in data become a huge concept. Thoughts on that and how you frame that idea? Because if I'm an enterprise, that's not how it was. I had to get a budget, build some stuff, show a payback-

>> Reusing things, creating derivative things like in every field, that's what unlocks real value. And I think with the concept of AI agents where you had humans reusing things and recreating value, now you've got AI agents going in and doing that. But one of the things is making those things more reusable by bringing more semantics in. But the other part of that, which I think we are partnering a lot on is the governance part. Reuse and creating derivative products, especially with agents, can be terrifying if you don't have lineage, you don't have governance. And so one of our perspectives has always been put everything, all your data products, not just data but all the derived assets into a single governance model and then be able to have not just permissions, but things like more metadata, more lineage. And so I think what's exciting for me is SAP's got all this rich stuff and they've had a very similar philosophy. We can partner together on creating a more federated governance model across the org that brings SAP and non-SAP data together for it.

Marc Geall

>> I mean, from my perspective, we understand the data, we understand the business context, and then we can use that to drive AI and value around AI. I think it is almost that simple.

>> In the world of data and you look at search and internet days, contextual behavioral are the two variables. The behavior is building and getting value. Context becomes the big thing. Domain expertise, you mentioned curated data sets. These are the key things that we see popping out of success as we see. Well, domain experts, horizontal scale. So you got to have a consumption distribution layer. So what impressed me about the SAP Databricks offering is you're distributing the product to your customers.

Marc Geall

>> Absolutely.

>> They're consuming SAP Databricks, you're selling that to your customers. Is that right?

Marc Geall

>> Yes, absolutely.

>> Now take me through that. So the customer experience is, hey, I got SAP and I got some Databricks. Is that it?

Marc Geall

>> No, I mean it's really about bringing best-in-class AI capabilities to the data rather than moving the data elsewhere. And the way we can do that is by tight integration between Databricks and SAP. We include Databricks within our commercial model, we do deep and rich integration so that you can now share that information within business data cloud, which is the product that we're taking to market. The customer can draw down on Databricks capacity units within that. And if they want to now also share that to their brownfield Databricks where they may have non-SAP datasets, you can now start to blend SAP and non-SAP.

>> That's the key. Talk about the dynamic there because that's the win-win, right? You got the consumption side, frictionless to your customers, you get the benefits of Databricks, but now you are now maybe in another part of the organization.

>> Yeah, we've got both ends and I think we're trying to create this best of both worlds model. I was in Sapphire in Madrid and speaking to a full room of interested SAP customers and only about 30% of them had actually used Databricks. So for them having a product that is just built into SAP and integrated with the way that they work is going to be super critical. At the same time, you've got all these Databricks customers and they have for a long time been paying that price of having to manually bring in data and for them, this is like a huge godsend because now they can just get to work on their problems. And the fact that both are built on this Delta-sharing technology we are talking about, we were just talking before this meeting about how do we share richer metadata. So we've got a bunch of things to do to make it feel like one organization. And I think reality is every large organization is going to have teams that are very SAP-centric. There are going to be teams that are increasingly, we've got 12,000 customers to date. There many of them start in the Databricks world. So making these things-

>> They don't want friction. They don't want friction.

>> Making these things work well together as one data platform is super valuable too.

Marc Geall

>> But I think there's also a buying center benefit here, right? We excel selling into the CIO and sometimes into the line of business. Databricks excels selling to the chief data officer, selling to the chief AI officer, and now I think we can bring that together and have a much broader play.

>> And Mark, to your point, the end customer, what's in it for them is they're just trying to run ML and AI and across their data.

Marc Geall

>> Yeah, exactly.

>> So, okay, sounds simple, but you guys are setting the table removing the heavy lifting. This is like-

>> Yeah. I mean this is another reason Mark mentioned serverless. Databricks, traditionally, if you get Databricks, you bring your own infrastructure and you go get your DevOps administrator to get that set up. We've been investing in the last two years in serverless to create a very SaaS-like model. And so that enabled us to partner with SAP such that you could go to SAP for me, click on that button, start the Databricks application in just a couple of minutes, have something that's fully provisioned and ready to use and those kinds of investments are super critical to us.

>> I love how the ecosystem is really evolving into co-creation on the benefits of the customer, making it easier, easier to consume, easier to deploy, and enable the AI, right? That's what the end game is. Great conversation. Love to spend more time on it. I guess the final question is, Mark, to you, what's next in the partnership and what do you think of the show here?

Marc Geall

>> The show's great. I mean, I've been super impressed. Hugely busy. For me, it's great to meet new partners, have a long-time experience with the Accentures and the Deloittes of this world, but now to meet the Databricks practice leads gives us an ability to share the story. In terms of what and that's next we've only started this journey. We need to make it successful. I think it's disrupting the industry. It's amazing the impact it's having on other platform players out there. I think some of them are running a little bit scared now. But ultimately what I want to do is see how we can leverage some of the announcements that have come out today around the agent's capabilities, the application capabilities that Databricks has announced because that really allows us to make SAP data products available to now a new ecosystem that we haven't played in before.

>> Yeah, and it's perfect for what your customers want, which is they know the environment, they know the domain that they're playing in. Now just let's get those agents and apps get built.

Marc Geall

>> Right. Exactly.

>> Vibe coding on SAP.

Marc Geall

>> There we go.

>> How's sales this month? Guys, thanks for coming on. So good to see you, Mark. Great. Don't be a stranger.

Marc Geall

>> Good seeing you again, John.

>> I'm John Furrier here in theCUBE live in San Francisco. Thanks for watching. We'll be right back after this short break.