In this episode, Dana Gardner, Principal Analyst at Interarbor Solutions is joined by Sridhar Ramaswamy, CEO of Snowflake. They discuss the impactful integration of AI with enterprise data, Snowflake's efforts to demystify AI for its customers, and future prospects for agentic AI and its potential to transform business processes.
In this episode, Dana Gardner, Principal Analyst at Interarbor Solutions is joined by Sridhar Ramaswamy, CEO of Snowflake. They discuss the impactful integration of AI with enterprise data, Snowflake's efforts to demystify AI for its customers, and future prospects for agentic AI and its potential to transform business processes.
[00:00:00] Producer: Hello, and welcome to the Data Cloud Podcast. Today's episode features an interview with Sridhar Ramaswamy, CEO of Snowflake. Hosted by Dana Gardner, Principal Analyst at Interarbor Solutions. They discuss the impactful integration of AI with enterprise data, Snowflake's efforts to demystify AI for its customers, and future prospects for agentic AI and its potential to transform business processes.
[00:00:31] Producer: So please enjoy this interview between Sridhar Ramaswamy and your host, Dana Gardner.
[00:00:36] Dana Gardner: Hi Sridhar and welcome back to the Data Cloud Podcast.
[00:00:40] Sridhar Ramaswamy: Hey Dana, excited to be here. Thank you for having me.
[00:00:43] Dana Gardner: You bet. You know, it's been just over a year since you joined Snowflake as CEO, and so congratulations on your anniversary.
[00:00:50] Sridhar Ramaswamy: Thank you. It's been an amazing year. It's a great company. Couldn't be more excited.
[00:00:55] Dana Gardner: Yes, it's been a very successful year for Snowflake, and we'll get more into that. But first, let's unpack the year in AI and data management trends in general. From your vantage point, uh, what have been some of the most impactful technological developments that highlight AI's true value and promise?
[00:01:15] Sridhar Ramaswamy: Yeah, the world of AI, first of all, is changing incredibly rapidly, absolutely making progress along many, many different fronts, but it's quite hard for us to make long-term predictions. I. For example, last year we thought that open source models would catch up, be as good as the best. Now we are not quite sure, and then we changed our minds again.
[00:01:36] Sridhar Ramaswamy: But what is becoming clear to everybody is that AI models, especially when combined with enterprise data, is going to be a game changer for lots of companies. Everything from how quickly you and I can get access to information, but much more importantly, what we can do with it. We are beginning to see the first.
[00:01:59] Sridhar Ramaswamy: I know it's an over hyped word agent. We are beginning to see the first agentic applications come out. I think this idea that you can stitch together different pieces of functionality that you have built with AI models and enterprise data into meaningful new components, I think that's going to be a game changer and is likely the place where we are going to see more and more impact this year.
[00:02:24] Dana Gardner: And then conversely, for your enterprise customers, those who are implementing and consuming the latest AI and data services, what's impressed you about how the market has adopting this? We know what the technology can do, but you can lead a horse to water. You can't always make them drink. How are the people actually consuming this that's encouraging for you?
[00:02:44] Sridhar Ramaswamy: We paid a lot of attention to this last year. We didn't want to sell hype based software. As you know, Snowflake is a consumption company, so to a certain extent, we are tied to creating utility for the customer, meaning we make money on products only if they use it. With that in perspective, we deliberately went about demystifying AI to make it easy for people to adopt.
[00:03:10] Sridhar Ramaswamy: Our motto with AI and Snowflake is easy, efficient, and trusted. And that last part is something that I'll come to again and again. Which is that unlike consumer AI tools, which can be wrong some of the time, and we sort of accepted as the cost of using these tools, especially free ones, the bar for enterprise tools is a lot higher.
[00:03:31] Sridhar Ramaswamy: And so when we went about creating components like Cortex search, which is for unstructured data, our Cortex analyst, which is for structured data. Our customers actually adopted them pretty wholeheartedly, and we would go to them and talk about utility, talk about reliability, and also talk to them about how they didn't have to invest huge amounts of money in order to build prototypes, in order to launch them.
[00:03:56] Sridhar Ramaswamy: And so we have over a thousand use cases in production and over 4,000 customers using our AI and ML products on a weekly basis. And that's a testament to the low key. But effective way in which we have been bringing these to market, and even internally, the amount of acceptance of these tools. It's pretty stunning.
[00:04:18] Sridhar Ramaswamy: So for example, we have a sales assistant for going through all of our enablement material, and let's face it, that's a whole lot better than searching on a website that gives you eight links that you then have to go click into and figure out. Instead, you get a written crisp answer with citations. Mind you, from the sales assistant that we have built, it is that mentality of actually delivering utility one bit at a time that's resonating most with our customers.
[00:04:46] Dana Gardner: And given the impact that we're seeing and the adoption that we're seeing, how have these trends and patterns buttressed and validated Snowflake's 2025 strategies? How have you seen the alignment between where you're going and where the market is?
[00:05:03] Sridhar Ramaswamy: There is a lot of customer feedback. One of the things that we knew we had to do in fiscal 25 last year was we knew we had to iterate.
[00:05:13] Sridhar Ramaswamy: Quickly and effectively with our customers, because AI was a brand new field and Snowflake was not known for doing ai, and our customers were hesitant about what AI could do. So we got a lot of feedback that validated many of our principles. As I said earlier, we are all about how do we make things simple?
[00:05:32] Sridhar Ramaswamy: How do we make AI simple? How do we make it efficient so they don't have to spend a huge amount of money? And, but most of all, again, how do we make sure we can put this in the hands of business users and not have them think about hallucinations, think about whether the tool that they're using is going to get stuff wrong.
[00:05:49] Sridhar Ramaswamy: It is on top of that feedback that we are now building. The next generation of frameworks like Snowflake Intelligence, which offers the chance for our customers and for Snowflake to mix and match different data sources now to now be able to take follow on actions to now be able to actually go update systems if we want them to be updated.
[00:06:10] Sridhar Ramaswamy: And that is what gives us a lot of hope that we are on a winning strategy because we know that our customers have actually agreed with and implemented on top of the things that we have already released.
[00:06:24] Dana Gardner: Right. Well, there's not many software companies growing at 30% scaled at scale these days, and so your recent corporate financial reporting certainly supports, what you've been talking about.
[00:06:34] Dana Gardner: How do you see the market maturing around AI on that utility basis on making sure that they're getting those business outcomes rather than perhaps falling for the vision?
[00:06:46] Sridhar Ramaswamy: I'm glad you touched on that. Indeed, there are very, very few companies that are making billions of dollars and are still growing at 30%, and last year was a litmus test for us as a company.
[00:07:00] Sridhar Ramaswamy: At some point in the middle of the year, there were serious worries about whether we were slowing down. You know what? I'm really proud of how the entire team at Snowflake stepped up the product team to deliver more products, especially in critical areas like ai, but also areas like open data, which is all the rage shockingly.
[00:07:22] Sridhar Ramaswamy: But our sales team, which learned a whole bunch of new skills around how to sell AI, how to sell new data engineering products, how to sell analytics on top of iceberg data that doesn't sit inside Snowflake. So that team came together and really turned the corner on Snowflake as a company, which resulted again in not only the impressive performance numbers for the year, you talked about revenue, but we also started showing expanding operating margin.
[00:07:54] Sridhar Ramaswamy: For fiscal 26, not only did we guide strong, we also said we're going to expand operating margin and lower stock-based compensation. It's very hard to do all of these things at the same time, but this is what gives us hope that we are on the right path. Our focus on utility, our focus on consumption. I routinely tell all of our sales teams that a customer over consuming.
[00:08:18] Sridhar Ramaswamy: In a way that is not commensurate with the value that they're getting with Snowflake is bad news for them is bad news for us. And so we embrace much more of this mentality of do right by the customer and good things will happen. And this is one, I think that'll stand us in very good stead for many years to come to be able to sustain this period of rapid growth, but always by delivering more and more value for our customers.
[00:08:44] Dana Gardner: Well, you seem to be executing very well and that enables you to help your customers execute even better. And we hear these days from a lot of enterprises is they really wanna accelerate the pace of change. They're facing a lot of complexity and there's dynamic elements to so much of business now they wanna compress the time to AI utility.
[00:09:06] Dana Gardner: They literally can't wait to get better business outcomes. So how would you look at actionable data and putting it into more people's hands? How specifically have automation, the power of natural querying and the use of agents, as you mentioned, how are they enabling more people to execute in their environments, getting better data and using AI resources?
[00:09:29] Sridhar Ramaswamy: Everything that you talked about just now, which is wanting to get utility faster.
[00:09:35] Sridhar Ramaswamy: Wanting to deliver and show promise to our customers, internal stakeholders. These are all things that we pay a lot of attention to. In fact, whenever I talk to customers, again, I tell them I personally do not tolerate or accept long proposals for migrations and or changing systems. Five years is a very long time to wait to get anything done, especially in this age of ai.
[00:10:01] Sridhar Ramaswamy: This is why our incremental approach, you can build a chat bot very quickly, five, 10 minutes, and soon we'll make it possible for our customers to build chat bots by directly connecting them to a data source like a SharePoint or a Google Drive, so that we automatically sync the data and keep the chat bot updated.
[00:10:22] Sridhar Ramaswamy: It is these kinds of efforts that really resonate with customers. Similarly, as a team, we leaned in a very, very big way to supporting open table formats, to supporting data in this format called Iceberg, where it does not have to be ingested into Snowflake. It can live in data lakes, it can live in cloud storage outside of Snowflake.
[00:10:46] Sridhar Ramaswamy: And now we talk to them about how they can bring the full power of Snowflake on top of the data and get value from that efficiently. It's these steps. Make it easy to make incremental progress, have a plan, have a strategy, but accompany that with execution in the small step-by-step, month after month where we are making progress.
[00:11:08] Sridhar Ramaswamy: That's the thing that customers are really happy about when it comes to working with Snowflake, and we are preaching what we practice, meaning use tools extensively. We are learning a lot about what does it take to roll out new tools in a responsible way to an organization that is nearly 8,000 people.
[00:11:28] Sridhar Ramaswamy: And we are also learning about how to show utility within our companies. And again, these kinds of projects become very helpful for us to then talk to our customers about in terms of their change management and how they can manage that across their large organizations.
[00:11:47] Dana Gardner: And of course, we can't just focus on that greater speed to value and democratizing the data services without being concerned.
[00:11:54] Dana Gardner: That there could be risks. And so how are Snowflake's approaches to that data security and governance helping to foster speed and trust at the same time?
[00:12:05] Sridhar Ramaswamy: This is a great question. One of the things that Snowflake has been known for from the very beginning is its excellent governance framework. Of course this starts with things like role-based access control, but it extends into things like supporting MFA multifactor authentication.
[00:12:22] Sridhar Ramaswamy: We have done that now for 10 plus years. We have supported network policies where a company can decide which IP addresses can access their Snowflake systems again for a decade plus. We go a lot farther than that. We make it very easy for data governance administrators within a company to find data, for example, that might have sensitive PII information in it.
[00:12:48] Sridhar Ramaswamy: If that is unprotected, that can then cause problems. The beautiful thing about Snowflake and its AI products is all of the governance work that people have done on top of Snowflake carries right through the difference between creating a chat bot on Snowflake. And doing it by extracting some data and sticking it into a new tool is you have to figure out how to export your governance framework over to that new tool, right?
[00:13:14] Sridhar Ramaswamy: When you do it on top of Snowflake, all of the role-based access control is obeyed out of the box. All of the data masking policies out obeyed out of the box. That's the magic of how we make things fit together seamlessly so that our customers don't have to do additional work to realize value from ai.
[00:13:32] Sridhar Ramaswamy: Of course, AI is also bringing up, you have to worry if you put out a chat bot that some person with malicious intent is going to type in crazy stuff into it and then get the chat bot. To say pretty bad things. And so we offer things like Cortex Guard, which is a shield for making sure that AI products can reject bad input so that they cannot be manipulated.
[00:13:56] Sridhar Ramaswamy: So we are working on this at a number of levels, but the thing that customers love the most about Snowflake is that products work as expected. They've already done a bunch of work in governance. It just carries over to how the AI products in Snowflake work as well.
[00:14:12] Dana Gardner: All right. Well, nobody exists as an island these days.
[00:14:15] Dana Gardner: That includes, of course, those that are consuming and implementing, but also those that are supplying. So let's step back and examine how Snowflake's partnerships and alliances are coming together to energize and extend the value. Uh, how are you and your collective ecosystem providing greater value across these collaboration chains, if you will?
[00:14:34] Sridhar Ramaswamy: The world of data is large. You've probably seen scapes, but it, there are hundreds of companies at Snowflake. We've been very proud of how we have partnered with a ton of companies. So we have companies like Observe, which are platform right on top of Snowflake. We support people like RelationalAI and Kumo AI and Elementum that build products.
[00:14:58] Sridhar Ramaswamy: Again, on top, they're available as native applications. We have a long history of working with other partners like Fivetran, like dbt, like Matillion for data engineering, and so our partnership tradition goes back a long way and we are very proud of what we are able to do for the ecosystem together.
[00:15:20] Sridhar Ramaswamy: The world of AI was a different challenge. These AI models are not quite software, and we had to figure out how we would work with companies so that their models could be available natively within Snowflake. This is where we are really proud of our partnership with both Anthropic and with Microsoft's open AI products.
[00:15:42] Sridhar Ramaswamy: And what these partnerships let us do is author the world's best models from right within the Snowflake security perimeter. One of the things that people sometimes do not realize about Snowflake is that it is a completely managed service. It is a global managed service, meaning that we control every aspect of that service, and our engineers are able to make very strong security guarantees to our customers about what is going to happen with their data.
[00:16:12] Sridhar Ramaswamy: This is why now saying Anthropic is a part of that security perimeter. Our open AI's models are available within our security perimeter, is a game changer. It completely lets our customers not worry about what is happening with their data. We've also updated our AI data governance policies. We have said, for example, that input into these models cannot be used, cannot be shared, cannot be used to create new models.
[00:16:42] Sridhar Ramaswamy: For example, our customer's data remains their data, and we take extraordinary care to protect it. It is a combination of all of these things. Bringing together these amazing partners, having them be part of our ecosystem that lets us create amplified value for all of our customers, making AI safer, easier to use while it's early.
[00:17:04] Sridhar Ramaswamy: There's a lot to come in this space, and we are very excited about things like Snowflake Cortex, along with Anthropic to create some amazing agentic applications. Same with open AI models as well.
[00:17:19] Dana Gardner: Sure, and playing well with others is a great attribute when the focus is on a solution, not on a platform or a technology.
[00:17:27] Dana Gardner: And so let's look at the global systems integrators who are focused on those solutions. How are your relationships with them and how are they an important go-to market for you, as well as helping to create that solution level value?
[00:17:42] Sridhar Ramaswamy: What the GSIs bring to the table. And there are a number of them, household names as it were, is amazing knowledge of customers, their companies, their systems, and what is needed to create value.
[00:17:59] Sridhar Ramaswamy: And where Snowflake is a big ally for them is in bringing a data platform that truly lets them focus on how they go about creating that customer value. We have worked very, very closely with Accenture, for example, to be part of their AI foundry. We have similar agreements that we are working on with other people, whether it's a Deloitte or an Ernst and Young, or Infosys or Cognizant.
[00:18:26] Sridhar Ramaswamy: We think of these folks as massive force multipliers for us. It's funny, but I go to the CEOs of all of these companies and basically ask them, how can I help you make a billion dollars? And, you know, as you can imagine, that's a message that goes over pretty well with them. I feel like we are at an inflection points when it comes to GSIs and our relationships with them, and you're going to hear a lot more about how much value these partnerships deliver in addition to what they're already doing for us.
[00:18:56] Dana Gardner: And I suppose another facet of playing well with others is embracing openness. And so open market, shared technologies, open source, bringing the most cost effective way to go to market, avoiding getting locked in, that sort of thing. And that openness has worked very well when it comes to accelerating access to more and more types of data.
[00:19:17] Dana Gardner: But now we're seeing that openness extend up into the higher AI adoption values and patterns. So how is openness extending from embracing a vast variety of different types of data into perhaps more types of many models and attributes of AI itself?
[00:19:37] Sridhar Ramaswamy: Yeah, this is a subtle and really important question.
[00:19:42] Sridhar Ramaswamy: As you know, Snowflake is a commercial entity. We are a company. We are not a nonprofit. We are not a foundation. We exist to make a profit for we exist to make a profit for our shareholders, for our employees. How we embrace openness while we also serve our fiduciary duty is an interesting challenge, and the way we set about creating a win-win is by again, focusing on what our customers wanted.
[00:20:11] Sridhar Ramaswamy: One of the clear pieces of feedback that we got from our customers was they didn't want data to get trapped. Many of them are coming off of very painful migrations from systems like Teradata, and it's very hard, very risky. And one of the things that's very clear to us from our customers is they want their data to be interoperable.
[00:20:34] Sridhar Ramaswamy: We listened, we agreed. And we have implemented our support for open data formats like Iceberg. It's the most popular format in the world. It is not controlled by a single company. It is controlled by a group of open source developers that truly, truly care about open form.
[00:20:55] Sridhar Ramaswamy: And then further up the stack, what we decided was that we would make our components for search, our components for structured data called Cortex Analyst and Data Agents. Also callable via APIs application programming interface so that they could be embedded into solutions that our customers were building. I go and tell our customers this.
[00:21:16] Sridhar Ramaswamy: I want Snowflake's components, AI components, all components to be embeddable in things that you're building. Do I have our own framework for building AI agents? Absolutely. We do want to make it easy for our customers to create it right within Snowflake, but if they choose to have components from Snowflake be part of other systems that they're building, absolutely we support that.
[00:21:40] Sridhar Ramaswamy: This is why we are partnering with Microsoft, for example. To have Snowflake intelligent components they call data agents, be embedable into their copilot, be embeddable into things like Power bi. It is this interoperability that we think will unlock more value. So one of these strange cases where cooperation sometimes even with a competitor.
[00:22:04] Sridhar Ramaswamy: Actually leads to much better outcomes for the ecosystem. As you know, we are in the midst of a massive migration from on-prem systems to cloud system, and the more the key players in the data ecosystem collaborate with each other to create great outcomes for our customers, the more business we are going to create for each other and we very much practice what we preach.
[00:22:25] Dana Gardner: I am going to guess that while the governance and safety was essential for being able to move fast and democratize data, that that same value on governance trust is extended to openness, that you can embrace more types of models, if you will, knowing that you've got that layer, that capability of trust and governance already in place.
[00:22:46] Sridhar Ramaswamy: That's right. I think having those just makes it easier for us to embrace these. It also makes things like our components embedable in other products, and let's face it, the best models on the planet are made by the likes of Anthropic and OpenAI. Being open to partnerships there brings those models within Snowflake.
[00:23:08] Sridhar Ramaswamy: It is a trite phrase, but win-win is a very real thing when it comes to partnerships at so many different levels.
[00:23:15] Dana Gardner: Alright, let's just shift gears a bit, if you will. We've been telling people a lot about what can be done, but it's important to show. So I'm wondering if you have some examples that we can look to that help illustrate some of the points we've been making, maybe share some customer success stories that exemplify where Snowflake's value is best demonstrated.
[00:23:35] Sridhar Ramaswamy: Yeah, there are a ton of them. One of my favorite ones is from Siemens. Their CEO Roland is visiting us today, so it's top of mind for me. We created a chat bot off all of the PDF manuals for their products. They make over 150,000 products and anyone that's tried to figure out how to repair a washing machine knows what a pain it is to find that magical PDF manual where you have to go to page 35 to figure out what to do with it.
[00:24:05] Sridhar Ramaswamy: We created a chat bot making it easy for their employees to get access to the right manual at the right time. Simple, incredibly effective. We work with customers like Disney to power their guest experiences. When you go to a Disney Park, the recommendation for what you should do next likely comes from a Snowflake model that's been trained on all of the things that their guests do within their resorts.
[00:24:35] Sridhar Ramaswamy: We have customers like Bayer that have implemented Cortex analysts so that non-business users can get access to business data without having an analyst in the middle or without having a BI tool. These are one sum of many, many use cases. As I said, we have over a thousand in production, but what they illustrate is the simple, easy value that we let our customers create.
[00:25:04] Sridhar Ramaswamy: And no story about use cases is complete without talking about internal use cases. We launched something called knowledge assistant for the sales team. I touched on this previously. All of our enablement material now available as a chat bot, and then the next generation of products, Snowflake intelligence.
[00:25:23] Sridhar Ramaswamy: Who is bringing together not just unstructured data, but also structured data so that our reps, our account executives can ask questions about how is consumption doing within this customer? How, what are top use cases that I should recommend for them? And the magic of all of this is that all of the governance rules, they only see what they're allowed to see, and yet they have all of the information in their fingertips.
[00:25:50] Sridhar Ramaswamy: That's the magic that AI is bringing both externally to our customers, but also internally to Snowflake.
[00:25:57] Dana Gardner: Well, this is certainly a large market that's growing very rapidly. There's a lot of competition. So I want to ask you a little bit about that. How is Snowflake differentiating itself from its competition in the market?
[00:26:10] Dana Gardner: How is your breadth, openness, and safety, and your embrace of the ecosystem benefiting you vis-a-vis your competitors?
[00:26:19] Sridhar Ramaswamy: Yeah, Dana, this is an important question. In any area that is growing a lot where there are a lot of dollars, you're going to have a lot of competition. I touched on this briefly earlier, but we run the largest managed data cloud that is out there.
[00:26:35] Sridhar Ramaswamy: Our competitors don't even come close. One of them, for example, talks about a serverless offering. We have had that since day one. This is 12 years ago. In fact, we don't have a non-serverless offering, and so we have innovated from very early on in how we create technology, our amazing governance, our amazing collaboration capabilities, and of course now our AI products.
[00:27:01] Sridhar Ramaswamy: The real secret sauce behind Snowflake's success though, is we integrate everything. We go through the sweat of making sure that products work well together. We go through the work of making sure that our products are easy to use out of the box, that they are efficient in terms of solving the problem.
[00:27:22] Sridhar Ramaswamy: One of the things that I tell our customers that I tell our own salespeople is we need to be the fastest and cheapest that is out there in addition to being great. Whenever customers look at the total cost of ownership of what it takes to run a system, Snowflake always comes out ahead. Of course, this is not to say that we can rest on our laurels, and one of the things that, you know, we've pushed the team really hard on this me on the exec leadership within the team, is to get more products out faster.
[00:27:51] Sridhar Ramaswamy: This is because we are at a particular point in the tech evolution that is particularly speedy. AI, as you know, changes by the day. I think we have adapted very well. Having said that, I respect competition. I want us to be better than them. And whenever somebody has a great feature, I point that out to our team and say, we need to figure out how we can be better than that.
[00:28:16] Sridhar Ramaswamy: Having that healthy sense of paranoia about always needing to be excellent in everything is what we are able to bring. To the table day in and day out, and we have the trust of the C-Suite execs in all of our customers that know that Snowflake will be innovating for them, not just today, but also five years from now.
[00:28:36] Sridhar Ramaswamy: But even more importantly, that we will be there from me down every single day. They run into a problem. We will be there to have their back. I think it is that focus on customers doing right by them. That is our ultimate competitive differentiator.
[00:28:52] Dana Gardner: Alright, let's take a peek into your pipeline for what's exciting and valuable.
[00:28:57] Dana Gardner: You mentioned earlier agentic AI. I'm wondering what is it in your pipeline that has got you jazzed about agentic AI and how specifically is Snowflake in a position to support the adoption safely of agentic AI?
[00:29:13] Sridhar Ramaswamy: So we should deconstruct what agentic AI is. I think of that as having access to many different sources of data to be able to answer complex questions.
[00:29:25] Sridhar Ramaswamy: That requires consulting multiple sources, perhaps looking at one, and then deciding what else to do. Perhaps if there is a technical problem, you look up a recipe for a technical problem from a set of unstructured documents, for example, in Google Drive. Use the content of that document to then devise a plan for what kind of queries you should ask, and then take an action in some system for that's the outcome that you want.
[00:29:53] Sridhar Ramaswamy: This can be partially automated. It can be human assisted, but these are the components that we're laying out there. If you think in terms of what you and I do for a living every single day. It all comes down to we look at some unstructured sources. It can be meeting notes. It can be latest consumption information, for example, that I look at for customers, and based on that, we'll be able to make some recommendation.
[00:30:19] Sridhar Ramaswamy: What agentic platforms make possible is the ability to programmatically do this. So we are looking soon to a world in which much of the work, for example, to underwrite an insurance application or a loan application, which typically involves going through 20 different documents and painfully surfacing relevant components from each of them.
[00:30:41] Sridhar Ramaswamy: Some structured, some unstructured into a new underwriting document that then goes to a council for deciding we should be able to automate things like that. To me that part's very exciting and our support for open formats, our support for connectors, which we did by acquiring a company called Data Volo, just means that more and more data is directly visible from Snowflake that can be used to power these agentic systems.
[00:31:09] Sridhar Ramaswamy: It's that end-to-end view to be there with our customers from the moment that data is born, to how it is processed, to how it is analyzed now. Then taken on to take action with these agentic frameworks is what is truly exciting, and it reinforces the different components of what Snowflake is. And to me, that's exciting in terms of what value we can get for our customers.
[00:31:34] Dana Gardner: And so let's revisit how you're implementing internally in order to attain greater execution, recognizing that these agents are so powerful. Can be embedded in processes, can be aligned with people to get the best out of human resources with what the agents can do. How are you seeing that manifest within Snowflake, that perhaps gives us a harbinger of what to expect, more generally?
[00:31:58] Sridhar Ramaswamy: These are components that are being implemented. As I said, one of the things that we are proud of at Snowflake. Is we didn't make this all or nothing. We created the components that are useful in and of themselves. So for example, if you take search, there is an employee assistant for looking over all our IT help.
[00:32:17] Sridhar Ramaswamy: There is a knowledge assistant for sales that has all of the enablement information that there is. There is a chat bot on revenue information that we created for Scarpelli, who is our CFO and me to be able to look at, so we can ask interactive questions in addition to that. Our engineering team is using products like GitHub Copilot, but also Cursor also augment to figure out how they can be more effective with how quickly code can be written.
[00:32:50] Sridhar Ramaswamy: Unit tests can be done, and so what you are seeing is a lot of independent efforts and Snowflake intelligence is busy stitching these together. One of the products that I am particularly psyched about over the past four weeks has been these deep research assistance. Now, Snowflake doesn't offer that, but Gemini does, and so does OpenAI.
[00:33:11] Sridhar Ramaswamy: I'm subscribed to both of them. We are busy now looking at how do we integrate that kind of web research into Snowflake intelligence so that that is available as a component as well. These are some of the examples of what we are doing to make more data easily available. The next thing that we are working on, it's not yet been released, is an Uber sales assistant that has quantitative as well as qualitative information in the same place.
[00:33:40] Sridhar Ramaswamy: That's when real power comes out.
[00:33:43] Dana Gardner: Interesting. And it's fascinating to me that things like Cortex agents are being used to help people better use and capitalize on the AI itself to do queries better by getting the AI to help be better at consuming AI. So it's a very, it's almost a virtuous adoption cycle.
[00:34:00] Sridhar Ramaswamy: That is correct. We made it possible for people to generate documentation for tables using AI. We are going to use AI to make migrations easier and faster to do. The team showed me a prototype last week of how we can help people write unit tests faster when they're migrating from one legacy system onto Snowflake.
[00:34:20] Sridhar Ramaswamy: I think that's the magic that is unfolding in front of our eyes, which is AI components cleverly designed to make existing systems better. It's that virtuous feedback cycle, I think has a lot of potential. For just productivity gains within every enterprise, absolutely, starting with Snowflake.
[00:34:40] Dana Gardner: Hmm. Really exciting things.
[00:34:42] Dana Gardner: So how can your customers, your partners, your global systems integrator associates, if you will, how can they learn more about these values and start to unpack how to use these components together in a virtuous adoption pattern?
[00:34:57] Sridhar Ramaswamy: Well, we have our big event, Snowflake Summit, coming in just a couple of months.
[00:35:03] Sridhar Ramaswamy: It's the first week of June at the Moscone Center again. We are going to be packing incredible amounts of content into it, not just new product releases, but also how to effectively use our products. And we are going to have many, many customer examples where customers are going to tell other customers how they've been using our product.
[00:35:25] Sridhar Ramaswamy: That is one way. We also do a lot of enablement for our partners to make sure that they are up to speed on what we are doing. We routinely engage with our biggest customers to make sure that they know everything that is going on with ai. Of course, there's our site, but we work very closely with our largest customers.
[00:35:46] Sridhar Ramaswamy: But Snowflake Summit is probably the easiest and most concentrated way to get everything outta Snowflake in just two days.
[00:35:54] Dana Gardner: We're having to wrap it up now at the, coming to the end of our time. When we talk again in a year about looking back and projecting forward, what do you think will be the biggest opportunities that people will reflect on a year from now, and what are the challenges that they need to keep in mind to make sure that they can be in the best place, you know, in just 12 months?
[00:36:17] Sridhar Ramaswamy: First, I'll remind people again that there is no AI strategy without a data strategy. Hmm. If your data is not in great shape, taking steps to thoughtfully make progress towards that so you can be ready for AI is important. Absolutely. All of your employees will get value from AI models just acting on public information, but that value.
[00:36:42] Sridhar Ramaswamy: Really magnifies when you have your own data house in gear. So I would say that that will continue to be important because we know this, we are in the midst of a decade plus transition. It's not going to happen overnight, but I think we will also see many, many progressive customers launch many different kinds of agents.
[00:37:03] Sridhar Ramaswamy: Some will be interactive, some will be automated. If they're going to discover the limits of their reasoning, what they're good at, what they're not good at, that will of course create a new set of challenges and problems that we need to solve. There is one school of thought that thinks that SaaS applications as we know it, are going to get replaced by quickly built agent applications because these models are getting really good at building ui, understanding complex data.
[00:37:35] Sridhar Ramaswamy: Hopefully we'll be well on our way to realizing some of these dreams and promises by that time. But of course there will be more problems then for the AI agent. Doesn't quite work well when there are 24 sources. How do we do better disambiguation? That's the joy of being in tech early. There are still many problems to solve.
[00:37:56] Sridhar Ramaswamy: Hopefully we'll create a lot of value and then create technical challenges we all have to solve.
[00:38:02] Dana Gardner: Well, great. And I’m afraid we'll have to leave it there. Thank you so much to our latest Data Cloud podcast guest, Sridhar Ramaswamy, CEO at Snowflake. Thank you so much.
[00:38:12] Sridhar Ramaswamy: Thank you, Dana. This is an amazing chat.
[00:38:15] Producer: Witness the future of data, AI, and apps at Snowflake Summit 2025. Join pioneers and industry leaders like Snowflake’s Sridhar Ramaswamy, and Open AI's Sam Altman in San Francisco, June 2nd to 5th. Five into 500+ sessions, explore 190+ partner solutions, and experience cutting edge demos. Transform your career and organization. Register now and build the future with Snowflake at Snowflake.com/summit.