Transcript
Hi, everyone. I'm Saptarshi Mukherjee. I'm a product leader at Snowflake. I lead our streaming data integration, zero-copy interoperability product areas based out of Seattle. Thrilled to be here. Subin. Good afternoon, everyone. Subin Abraham. I'm the director for data analytics and AI at Daikin Comfort Technologies, where I lead our enterprise data and AI strategy for SAP transformation. Hi, everyone. Good afternoon. I'm Farooq Munshi, partner in EY, based in New Jersey. I lead our overall data and AI business for advanced manufacturing. I'm really excited and thrilled to be here today. Wonderful. Thank you all. So we're going to get started with Saptarshi first. So our partnership, SAP and Snowflake, just reached a brand new milestone. We are officially generally available with the two joint offerings. There's SAP Snowflake, which is a SAP certified solution extension for BDC, and then SAP BDC Connect for Snowflake. What I'd like to start with is, can you talk about the significance of the partnership? And then, what does this mean for customers? Absolutely. Happy to share. So when Snowflake started working with SAP, what was top of everyone's mind was AI is generationally shifting industries, whether it's core manufacturing, whether it's retail, whether it's consumer packaged goods. But AI needs data and a very strong data foundation across different applications. And the more important thing is AI also needs semantics. When you are building agents that are reasoning around data, that agent needs to understand what is product, what is material, who is a customer, who is a supplier, not just technical bits and bytes. So the Snowflake and the SAP relationship is grounded on two truths. One, we will make sure that AI solutions and agents get governed access to SAP data at scale. You don't have to worry about ETL, the things of broken pipelines anymore. And also, you're able to serve SAP's full business context, reach semantic data into your Snowflake agents so that those agents can operate with business reasoning. So that's what we have been focusing on. And as Ryan, you mentioned, we are generally available. That means you can start building AI agents. And this is why we have Subin, who's going to talk about their journey with Snowflake and SAP together, how they are thinking about AI and business analytics. Great. Thanks for that, Sap. So now on to Subin. So I'd love for you to share, like Sap mentioned, the journey you've been on, not just SAP transformation, but also a transformation for your data foundation. So can you share a little bit about those journeys where Snowflake fits in today and how you're thinking about the next phase, and especially for the next phase, how you're thinking about scaling your foundation to be AI-ready, leveraging this new zero-copy integration? Absolutely. So for Daikin, the AI journey or data journey started with a business problem. The business problem was fragmented data, multiple ERPs, isolated SQL environments in our landscape, multiple products for analysis and analytics, and also multiple AI journeys throughout our organization. And what we focused on was not about building out models, but how do we create the right platform as we start out on scaling up our AI or advanced analytics journey. But how do you scale this platform? With our phase one of our implementation, our CFO and our CI also talked in a previous session here, was we kept data front and center of any journey. And as we think about data, we ensured that we create a platform. And Snowflake was our choice of platform, which gave us the scalability and the agility so that we could bring SAP and non-SAP data together. The reason why we wanted to bring in the non-SAP data is we are not on a year-long journey, but we are on a five-year journey. So we have to live in transition. So we had to bring all this data together. And Snowflake gave us the choice of platform in terms of agility that we wanted. As we built out our models, our semantic layers, all those existed within the Snowflake environment. And SAP Datasphere, which is a solution from SAP, was mainly used as a pass-through channel for us to push data into Snowflake. But all our AI, ML workloads were running on Snowflake. But as we have become more and more mature, and we are getting into a manufacturing rollout in the next few months, we took a step back and started looking at our data platform and the strategy that we already had. And we started realizing that, hey, we have to contextualize our data more than what's needed, especially for a manufacturing organization like us. It makes sense when we look at forecasting, or we are looking at inventory, or even looking at credit management in our business context. It's absolutely necessary to add the business context around your data. And how do we do that? So we started looking at Datasphere again. And we have started building our semantic models, our data LAD, I would say, in a LAD architecture, raw, silver, and gold, medallion architecture, and the gold LAD becoming our semantic model that could be consumed by any other downstream systems. Now, what this has helped us do is keep the business context around our data. And now what role is Snowflake playing is playing the larger enterprise data platform role of not only running our AI, ML workloads, but we have use cases which we are running on Snowflake, especially on forecasting, credit application. We have these outcomes that are being generated out of Snowflake being fed back into Datasphere and being leveraged in solutions like SAP Analytics Cloud for our insights. Now, what this has helped us do is not only scale at the pace that we want with Snowflake, but also keep the business context around the data from an SAP standpoint, and also bring SAP and non-SAP data together in one environment. But where does the journey take us from here? And that's where BDC came in place. The journey started last year, and we started the discussions not only with SAP, but we started talking to Snowflake at the same time. Hey, where do we go from there? And it was about mid last year, we heard about Zerocopy. And what Zerocopy is going to bring to us is it's going to streamline our architectures tremendously, not moving data, copying data from one environment to another, but it's going to have one centralized source of data, which is having the business context around it, but can be not only used by Snowflake, which runs our AI ML workloads and advanced analytic workloads, but also any other downstream systems which wants to consume this information. And the Zerocopy is not only going to serve the purpose, but also streamline our end-to-end value stream as we think about our data journey. It's amazing. Thank you for walking through that. And I think that's the exact kind of real world context that makes this real, and makes this conversation tangible, and sets up the next part of our conversation really well. So Farouk, from your perspective, leading manufacturing for data and AI at EY for the Americas, what do you think are the most important things that manufacturers need to get right in order to turn this new SAP and Snowflake partnership and foundation into the higher value outcomes from AI? Absolutely, Ryan. And we saw a lot of the teams coming out already in some of the journey that Subban and team have been driving at Daikin. And as we have been working with our manufacturing clients, there are some exciting themes that we see emerging. And I'll kind of walk through some of them in terms of where we see some of the biggest opportunities. So if we start with this aspect of AI and manufacturing, that requires very trusted operational data, and not just models. So across the enterprise, what we are seeing with several manufacturers, they are really focused on investing in large scale data AI analytics initiatives. But without having that trusted, reliable data foundation, the timely nature of your real time data, and also not understanding what the business context is, you are not really getting the true value out of your use cases, whether it's in your shop floor, whether it's in your supply chain value stream. So that's a key area that we see really front and center. The second piece is with our clients, really starting with SAP as the core. That's a key opportunity, but that's only part of the problem, part of the overall story that they're looking at. It's very important. And I think so many have touched upon it, is how do we look at not just the SAP, but non-SAP information that is coming from your shop floor telemetry, or your supply chain signals, planning your customer demand? How do you bring those data together to really unlock some of the end-to-end use cases? And this is where we see kind of zero copy being a force multiplier. Really having the ability to get context from your source system and having the ability to run it in the cloud, where you are able to not just look at an individual use case, but across a portfolio of opportunities, can be significantly game-changing. The last piece I would add on is, as we think about data in the manufacturing ecosystem, semantics and governance is so critical. It is very important to think about context, especially as you are dealing with complex use cases like quality monitoring, on-time delivery, inventory optimization. How do you bring all of these things together? Having the right governance and context is important. Otherwise, as you look at agents and automated decision making, there is significant amount of risk. And that's where kind of bringing that end-to-end integration opportunity, but also bringing in the right context from your source system, where you have some of the SAP nuances being exposed, really gives you that governance that our clients are looking for. For us, we look at this journey through the lens of really advising our clients to think about what's the highest value opportunities. When you look at manufacturing, again, as I go back to yield optimization, or you're looking at downtime reduction opportunities, or quality monitoring, those are some of the outcomes that you want to really start with. Use that as a building block as you start setting up some of your reusable data products, and really scale that across your enterprise. And I'm really excited to see the journey that Daikin has started with, and the partnership with Swaben, and setting that data-first mindset and the foundation early on that really helps you build those enterprise-grade solutions and capabilities that supports a lot of scalability or AI in the future. So really excited about the journey we are in. All right, thanks, Farooq. So SAP, I'm going to bring this back to you. So focusing on customer opportunity. So SAP data, business context, they're now available, accessible in Snowflake, as well as bidirectionally back in SAP. What does that exactly mean for customers? What can they do now that was not possible before, or just very challenging to do? I think that's a really, really good question. I get to talk to many CIOs and CDOs across the globe. And the number one question that I hear from them, where do I get started? Should I build the agents in BDC? Should I start thinking about building AI agents in Snowflake? And the way we need to think about this area is across three layers. The first layer is the data layer. And we spent some time talking about, when you are thinking about an AI solution, use case matters most. The use case is going to drive the need for data. There is no strategy where let's start building agents. We need to figure out, OK, let's start doing financial planning better. Let's do supply chain optimization better. And for that, what are the data sets that are required? Where are they sitting? If you have zero-copy access, that makes life easier, firstly. The second layer is, do I understand semantics in a consistent manner? So what we did, we created something called Snowflake Semantic Views. What SAP has in Business Data Cloud called Core Schema Notation, CSM, to express semantics. So our AI system can take SAP semantics and translate that directly into Snowflake Semantic Views. And the third thing we did, we invested in an agent in Snowflake called Cortex Coding Agent. In natural language, it's very similar to Cloud, but much more contextualized for data-intensive problems. You can go and start building your, what I call, data insights agents. That means you can give it a prompt. It can create an agent. It will use the data and the semantic already in Snowflake. It is the fastest-growing product Snowflake has launched in the last one year or so. There's massive adoption where enterprises are able to build governed agents in Snowflake using Cortex Coding. You build these agents, but these agents cannot make SAP supply chain decision. It cannot place a purchase order. It cannot go ahead and make a pricing adjustment. In SAP, you will build what I call them action agents, agents that can actually understand the business process fully, can get governed by SAP's governance policies. So when an insight agent in Snowflake generates insight, that insight becomes available via zero copy in BDC. So when you build insights app, jewel agents in SAP, now that agent is able to work on business insight that is coming from enterprise data across the board, right? So think about these three layers, data, the semantics, the insights agent, and the action agents, and how they're going to interoperate. Pick a use case, and then let's start operationalizing that use case, right? So that's what's going to be my recommendation. That's fantastic. Thanks for that. And just a few minutes left, I do want to close it out by, want to get one final perspective from each of you about, you know, something practical the audience can take away as they think about their own SAP data and AI journey. And Farooq, let's start with you. So from your perspective, from manufacturing, what are the most, what are you most excited about as customers start to bring these capabilities into real operations? Absolutely, Ryan. And I would say what we are seeing in the industry overall, but even some of the opportunities is really that transition from that hype to real tangible value. That's really something that excites me. And as I look at that kind of, that inflection point, I would say the two pieces are driving that overall journey. One, like if I think from a technology standpoint, we have two great companies, SAP and Snowflake coming together, really helping simplify the overall technology ecosystem and making sure that we have that interoperability of the data, which is allowing our clients to really get the best of both worlds and really driving a lot of unlocking some of the value. The second piece I would say is really the pace of execution that we are seeing with our clients. It's no longer the weeks and months of getting to a use case, identifying the solution architecture and delivering value. Here, as we are working with clients over a period of 24 hours, we are able to take a concept, put it together, prove out the value. And then within a week or two, we are able to deploy it in production. And a big part of that consideration is what these technology ecosystem is bringing out of the box and really simplifying your execution landscape. So that would be one area I would encourage everybody as they are kind of looking at these new innovations. How do you quickly, not just kind of put it in a sandbox, but also how do you take it to production and show the value to your enterprise? All right, thank you for that. Suman, let's move on to you. So there's a lot of pressure for enterprises to move AI from experimentation to global deployment and higher value AI outcomes. So based on your journey at Daikin, does this feel like you're headed in the right direction to get those higher value outcomes that your leadership is really starting to expect now? Yeah, absolutely. I think we are on the right path now as we think about the way we have set up our platform. If you think about any AI journey as Daikin, we first focused on setting up our foundation layer. And not only with Snowflake, but also with SAP and BDC coming our way, I think we are thinking on the right path of how do we leverage the capabilities each of those platforms could offer. Like Saab mentioned, you have your insights layer, you have a semantics layer, you have a foundational layer, but how do you interrelate these value chains together, tie them together to have your agentic workforce make the right decision. So I think we are on the right path and building the right foundation. A lot of AI initiatives fail when they go against wrong data, fragmented data, and I think we are not going that path. We have made sure that we are making a consensuous decision of what to do before we unleash AI agents in Daikin. So I'm really excited what BDC is going to offer with what Snowflake is going to bring to the table and how do we connect these different data sets together to make things happen for Daikin. That's great. All right, Saab, final question. So from the Snowflake side, now that the partnership is generally available, both joint offerings, what excites you the most about what comes next? I mean, there is no more excuse why not to build an AI agent? Why not to start operationalizing, picking one use case, start working on building an agent? If you haven't built an agent yet, please strongly consider starting that journey this week. We are investing in tools that's going to bring the bar of building AI agents that are governed, secure, performant, down to a level where business user departments can start working on building their own insights agents. So start getting acquainted with the product, the technology. If you are a Snowflake customer already, the connector is generally available. Let's start a POC. I mean, it's a few clicks. We have ENY, we have partners who can help you. You can go ahead and sign up and start using it. If you are not a Snowflake customer, but you're thinking, hey, SAP is my broad application landscape. Christian talked in the morning about ERP is the brain. Snowflake is part of the solution extension of the business data cloud. You can buy Snowflake through your existing SAP contracts. I will say, let's get started with the journey. We are here to help you. And next time when we meet, hopefully we'll have many production agents live and helping your business. And with that, check out snowflake.com slash SAP for more information, including an ebook we just dropped. So thank you all. Have a great rest of the show. Thank you. Thank you.
