AI Revolution: Unlock Your Enterprise's Full Potential!

>> Please welcome Snowflake Chief Executive Officer Shredar Ramaswami. Officer Shredar Ramaswami. [music] Hello and welcome to our biggest and best snowflake summit ever. Over 20,000 of you here with us. over 500 sessions, 500 sessions, almost 700 speakers from 350 organizations. Thank you all being here with us. I may be a little biased, but I think we have truly the best community in the world right here. [applause] And we're here because of all of you. this incredible group of customers and partners and snowflakes who motivate us to go further faster. I want to take a moment to thank the thousands of customers here with us. Thank you. You're shaping the future of data and AI and we're super proud to be your partner. And we have hundreds upon hundreds of partners right here today. These are the organizations that are supporting Summit. They are the ones that are working together with our customers to show what is possible with AI and data. And finally And finally to our snowflakes. to our snowflakes. This week comes to life because of you. Every demo, every launch, every unforgettable moment reflects the incredible work that you put in every day. Thank you. [applause]

This week is going to be opening up the realm of possibility. Last year, we could not have imagined some of the things that we are doing right today. We have customers like Canva, one of the largest design platforms on the planet with an eyepopping 265 million people showing up every month to create. And last year, they stood on this on this stage and shared how they're going to use Cortex Analysts to get more value from data faster. Fast forward a year from now and they're using Snowflake AI to transform how they make product decisions, turning weeks of analysts on user behavior and feature impact into near real-time insight. This enables their team to move faster, invest in new areas with confidence and deliver incredible experiences at the crazy scale that they operate in. And then we have folks like Nestle. It's a global company with over 2,000 brands across 185 countries. We have Nestle fans. Let's give them a hand. [applause] And as part of their enterprise digital transformation, Nestle has embarked on an end toend reimagination enabled by data and AI. They've built enterprise products that are now deployed to over 50,000 users and have 150 global capabilities. But what does it do? It's driving greater interoperability, greater reuse of data, and they're using AI to help anticipate supply chain disruptions, giving planners the insight to respond to shifts in supply and demand and production across their global operations. This means that they're creating an operation where data isn't reporting on the news. It is driving proactive decisions in real time. And overall, I have to tell you, and I'm sure there are many people here that feel like this, this is one of the most magical periods of technological change that I have personally ever seen. AI isn't a promise anymore. It's creating real outcomes, real tangible opportunity. And our goal this week to make AI real for your business.

The very nature of work is changing. Increasingly day-to-day work is about partnering with and guiding intelligent agents. And soon they will be operating across your business continuously autonomously. autonomously. This truly is going to be the era of the agentic enterprise. That's a vision though. You have to make it a reality. How do we go about doing that? There are four essential components to becoming an agentic enterprise. First, you need to have your enterprise data and context. your customers, your financials, your product data, everything that makes your business your business. And then you need AI models. Cloud Opus, Armyos, Gemini, GPT, these are the amazing models that interpret, reason, and act on your enterprise data and context to deliver meaningful insight. There's more. all the software and applications that you folks use that we all use every day. Gmail, Outlook, SAP, Salesforce, Zoom. These are the applications that we use to keep our businesses running. But there is a fourth component that brings this all together. And to be successful as an agentic enterprise, you need an agentic control plane. This is a mission control center to coordinate across your data models and applications to ensure that every decision and action is governed, trusted, and grounded in your business context. And that is what Snowflake is delivering

with Snowflake intelligence and Cortex code. For knowledge workers, Snowflake Intelligence is the personal work agent that helps you work smarter. With Snowflake Intelligence, you can access all the data you need, but take action across the applications that you use every day, but do it in natural language. And for builders, Cortex Code or Coco as we call it affectionately is the AI powered coding agent that helps builders tap into any data anywhere and build faster with confidence. You're able to turn ideas into working pipelines, applications, or agents. Again, all using natural language. These are the building blocks of a system of decision. These are your agentic control planes and they're already powering incredible transformation across thousands of leading enterprises. With agentic enterprise, things just look different. And here is a view of what this future looks like that we at Snowflake are living through. >> Data shouldn't slow down your business. Hey, we're suddenly losing customers. What's happening? What's happening? >> I can set up a dashboard, but it's going to take me a while. When questions move faster than answers, you need AI that breaks silos and adds context. The largest driver is onboarding delays. What can we do about it? Allowing you to get answers quickly to make decisions and take action. Now, here's what we automate next. Early churn signals, triggered interventions, proactive outreach. >> I've already built a skill with Coco so we can selfs serve across the team. >> In the Aentic Enterprise, everyone becomes a strategic force shaping what the business does [music] next.

>> And what do we want to talk about and help you all do? Bring that vision into a reality for you, all the people in this room that are going to be driving this transformation in your companies. Now, I'm sure you're thinking, "Uh, that sounds easier said than done." But we at Snowflake are here to help you get there. We work across all four of these components to make this coordinated action possible. Let's break down the journey. This agentic enterprise. Of course, all of this starts with your enterprise data and context. the foundation. We hear it constantly from all of you. Our data is a mess. There's more of it than ever before. Structured, it's unstructured. It's not working for our business. It's not yet ready for AI. That's a heck of an expensive gap. Why? Because your data is your most defensible mode. It's what makes your company truly your company. It's the foundational context that make AI outputs relevant, differentiated, and strategically valuable for you. If it's sitting fragmented and unused, so is your competitive edge. So, why isn't your data ready? We ask you. And if you ask anyone that's tried to build a data pipeline, they'll all tell you the same thing. data siloed across applications across a dozen systems, multiple cloud providers, legacy platforms, each with its own format, each with its own source sliced version of the truth. And bringing all of this together has been this painful, difficult problem. It's a never quite finished engineering problem. Snowflake is here to help you unify your data and break down those silos. We make it easy to get data into Snowflake through seamless AI powered migrations, Postgress support and openflow for continuous data movement. We make it easy for you to share your data across teams even with partners using solutions like data clean rooms and we support you in your full data journey in one platform. for migrations to lakehouse and analytics to AI powered BI all in a single place. But bringing data together is only part of this challenge. You also need to make that data enterprise ready. That means

you need to have governance. You need to have security. You need to have auditability and compliance built from day one, not bolted on later. Because we've been built for the enterprise from the very beginning, organizations can rely on the resiliency and business continuity controls that they depend on. Knowing that only the right people can access the right data, we offer you the flexibility to operate across cloud providers and model providers. We are also open and interoperable so you can work with the systems, tools and technologies of your choice. And with governance, security and compliance embedded into the platform from the start, your data isn't just unified. It is trusted and ready for AI. And that's what it takes to unlock performance from your data at massive scale. And now with Koko, the work of getting there just got dramatically faster. That never quite finished engineering problem, it's now automated with Koko. Migrations can be done in days. Pipelines are built autonomously. And now all of you can spend your time on higher level strategic work. Setting a strong enterprise context foundation moving towards autonomous workflows. That's step number one in becoming an agentic enterprise. And few have moved faster or helped more companies do the same than Accenture. They've not only advanced their own data

and AI strategy, they've helped thousands of customers do the same. They're a leader showing up how you can make AI real for your enterprise. And let's hear from someone who sees AI's impact play out across every industry, every market, and every kind of enterprise. Let's hear from Julie, chair and CEO of Axenture. Thank you, Shredar. And hello, everyone. You've just heard Shredar describe how the agentic enterprise will define the next phase of work. It is the daily conversation for me and my leadership team. Every client is asking the same question. How do we move from AI ambition to actual business outcomes? And here is what we tell them. AI is a powerful tool. It creates real value when humans are in the lead. That means leaders knowing how it's used, where it's trusted, and which outcomes matter most. When leaders take personal responsibility, AI opens up entirely new growth. And that hands-on responsibility starts with leaders changing how they work themselves. If as a leader you don't understand AI, you can't explain it. And if you can't explain it, you can't reinvent how work gets done with your teams. Scaling AI takes leader-led adoption, a decision to reinvent that inspires as much as it informs. And reinvention only lasts if it shows up in the P&L. Every pilot looks promising on a slide. What matters is whether AI is embedded in how the enterprise actually runs with clear accountability and ownership. That is when impact gets real and durable. As Shreidar said, the Agentic Enterprise is built on data, strong, trusted, governed data. That is why our partnership matters. Together, we help clients put the data foundations, controls, and governance in place to move past the insights and into outcomes at scale. It has never been more important to get this right. You will hear more now from Anise Chararma who leads our reinvention services on how we're bringing this to life together. Thank you. >> Thank you Julie. [applause] >> Please welcome Manish Sharma, Chief Strategy and Services Officer of Accenture to the stage. Manish, >> thank you for being with us. This is quite the crowd. quite the crowd. >> Incredible. [laughter] >> Incredible. [laughter] Well, Well, Manish, you are uh at the forefront of transforming Accenture, which is a sizable organization, into becoming an agentic enterprise. How are you turning that vision for where you want to go into the reality of what needs to get done? I think first of all let me kind of just say incredible event like just even before it is started just meeting the folks around the buzz it is incredible so thanks for organizing thanks for having us here >> thank you for being an amazing partner >> and we love this partnership uh and I want to start by saying congratulations on the strong momentum on the snowflake intelligence as well as the cortex code it has been incredible what we are seeing in the market and with our client clients and with ourselves. >> That's right. That's right. >> So I think the in terms of what we are doing uh in Accenture, I think two months back virtually we changed every single job in Accenture of a 70 billion company 750,000 folks across the world. We kind of organized >> every single job 750,000 people >> 50,000 people right you know and we created it across seven key seuite personas like whether it is CFO CHRO supply chain and manufacturing song which is the customer part but each one of them required a strong data foundation so we have organized Shri on what we have been always been discussing but we reorganized just to ensure that we are able to facilitate and reinvent ourselves every single process, every single business outcome that we need to deliver based on a strong data foundation working with you. >> That's right. That's right. That's right. Uh in fact, I remember talking to Manish. I think it was early January when I showed him what was possible with uh Snowflake intelligence and he gave me a glimpse of what they had already built on their data foundation on top of Snowflake and we went to work making that data come alive for Julie and Manish and everyone else and practicing that told us a lot which brings me to my next question. How do you how has AI changed how you operate daytoday and how do you use that to inform what your customers should do? So I think right now we are doing at scale across the globe with thousands of clients but every time Shri and the team comes to us with something brilliant and he's always got to show me something on his mobile you know and he's fascinating whenever we talk to him the first thing we want to do is apply it on Accenture because if you can do it in Accenture with our complexity with 57 countries 750,000 people thousands of clients. We have 9,000 clients across the world. You can do it anywhere. >> That's right. >> So the first one is always customer zero. So we built and most of our clients AI is not the challenge. The challenge is the data and the process debt that people have. >> 85% of our clients have a data problem. And that is where we come and partner with you. And in our case, we built our the full data foundation is built on your stack. >> Thank you. >> Thank you. >> You know, everything we have tried before, we have gone to every single client. I remember there was a one very large uh you know uh wealth advisory uh financial services company and we were going there and it's the messiest the most legacy stuff and you know you will get frightened to kind of just see the different data structures and the data stuff that they have fragmented we went with you sorted it and now we are getting insights so quickly that I can't even imagine three or four years back that we could have done this but that is again to the strong partnership that we have with snowflake >> and that's the magic of the moment what are you most excited about what we can unlock for our customers together say in the next two quarters >> I would say relentless focus on business outcomes relentless focus on linking every project you do to a line item if your P&L it's impacting your revenue venue is it is impacting your COGS or it is impact impacting your SGNA got to have clear linkage create value >> and I'll give two examples one of them is one of my favorites uh a utility company again a legacy European uh utility company we went there together and billions of data elements what is the outcome an analyst wanted to query From taking weeks to 2 seconds that is exact stat from scope of a project to actual execution of the project 12 weeks from months compute time down by 85%. That is kind of the uh one piece. The other one which again is a client which I'm personally involved manufacturing client US I thought I can never get in every meeting you start there they will first argue for first half an hour about whether the data is correct or not correct and whose data is right and creating a single source of truth among huge you know number of uh fragmented systems and data records. Now we have one single source of truth. Ability to have decisions faster, much quicker. Incredible. And I think if I were to kind of just tell the folks here, that is what we should be looking at. Relentless focus on business outcomes linked to your line item. Don't do a project if you are not linking it to business outcomes. There's a a simplicity and beauty to all of this which is we succeed when we help our customers make more money or spend less money. either I mean use COGS or SDNA there is nothing more than that right >> that is that is that is amazing it's also such crystal clear just clarity on how to work I want to thank you again for and Julie for the partnership our teams work extraordinarily well together and we are transforming ourselves and I could not be more thrilled to have a partner like you thank you >> and we will create more and more value for our customers and clients every single day thank you thank you for everything >> thank >> thank Thank you for being here. [applause] >> What Accenture shows us is that organizations winning with AI are doing it by connecting AI to what they actually know, their customers, their operations, their data, but also a clear hardeyed view of what value creation means. So that was the first big step in this journey. So you have a strong enterprise context foundation. How then do you create meaningful insight for your business? And this is where

AI models enter the conversation. Just the pace of innovation with AI models has been nothing short of stunning. There is a butt to that. The model is not your unique advantage. Why? Because your competitor has that model too. It is when you combine that model with what is uniquely yours, your data, which is your advantage, that things begin to shine. your customer data, your operational and performance data, your partner data, your application data, all of the proprietary context. That's what makes AI useful for your business. So then the challenge becomes enabling models to reason over the full context of your enterprise. And Snowflake has always been about choice, and that extends to models and to providers. One workload may benefit from a frontier model that's built for broad reasoning and exploration, while another might be better served by an open source model that's optimized for speed, customization, or cost efficiency. Through native partnerships with amazing model providers like Entropic and Open AI, Snowflake gives you flexibility to choose the right model for each use case. Instead of moving data to the model, you bring AI directly to your data with the freedom to switch, combine, and evolve models as your needs change. And we help connect the intelligence to the applications and systems that your business runs on. Snowflake has always been about zero copy integrations with applications and systems that you use like SAP, Salesforce, Service Now, Workday, IBM, Aviva, many more. So you can tap into all of your data without moving or copying it. And traditionally integrating applications connecting them has

required complex brittle integrations. MCP which is a protocol pioneered by anthropic changes that it's an open standard. It gives AI systems a simpler way to connect with the applications and platforms that all of you use every day. But you know what? In the enterprise, connectivity alone isn't enough. It obviously has to be secure. It has to be trusted. It has to be governed. And that's why And that's why we have announced our intent to acquire NATO. With NATO, Snowflake is bringing MCP powered connectivity natively into the AI data cloud. All of the applications that you folks use whether it is Google Drive or Gmail, Zoom, Jira, Slack, GitHub, Microsoft 365, these all are now visible to AI models and every interaction with these applications will be governed by Snowflake security and access controls, giving organizations visibility, oversight, and safeguards such as human in the loop approval. And this expands Snowflake government's perimeter from data to AI action and workflows across the enterprise because the future of AI isn't just getting an answer. It's about turning fragmented systems into actionable insight. But of course, the next question that you will have is how do you turn these insights into actions that you can take? How can you start executing complex agentic workflows? How can you operate at the speed of AI? And this is where there is a new

category called the agentic control plane that is becoming essential. You can spin up agents across your business, but without a control plane, these agents are effective only in isolation, but they're not really aware of each other. An agent in finance can make assumptions that contradict what the supply chain agent just decided. And a marketing agent might generate content without knowing what a support agent learned about that customer just minutes before. You need a way to coordinate across your context, models, and applications so that decisions and actions happen seamlessly across your business. And Snowflake is helping thousands of businesses do just that. With Snowflake Intelligence, every user in your business now has a personalized work agent, a single place to access data and take action. Whether it is sending a Slack, summarizing meetings, searching across email threads, the things that you do every day, all automated, all governed. And now those agents can even be equipped with specialized skills built around how your teams actually work and able to take action within these workflows. An HR professional shouldn't have to ask a generic AI assistant for help with recruiting. They need to have a skill that knows their candidate pipeline, can talk to that application, talk to other tools and their processes and can then act on it. Similarly, a sales leader that's planning for the quarter needs more than a chatbot. They need a skill that pulls their forecast, surfaces potential risk, and drafts the prep dock for the meeting without being asked to do that twice. And a marketer personalizing an email campaign shouldn't be copying and pasting between tools. Their skill already should do that because they already use it. And it should bring together these things into Slack, Salesforce, Google Workspace, Microsoft Teams, whatever products that that particular company uses. And that becomes a difference between an AI that gives you answers and AI that can turn that answer and insight into action in the applications that you're using day-to-day. And for all of you builders out there, developers, data engineers, analysts in your organization, we give you cortex code which we affectionately call Koko. With Koko, you can build anything. It lets you go from idea to working system using simple natural language. With Koko, a migration that took 6 months now is 6 days. It can assess the workload, convert the code, validate it, and deploy it all in a conversation. You can build complex pipelines, dynamic tables, quality checks, governance, data cost, cost monitoring in one afternoon. You can stand up a production application connected to live data before lunch. And Koko just doesn't write code. It understands your snowflake environment, your tables, your roles, your security boundaries. It builds, tests, deploys, and optimizes. And it learns continuously. The thing that used to require a specialist in your team now just require an idea. And it's amazing to see how quickly Snowflake Intelligence and Coco are changing the way enterprises work. Sani has been at the forefront of this agentic transformation building a strong strong data foundation and reimagining how work gets done across the company for the AI era. They were precient folks. They started this years ago and I'm really thrilled that they're here today to share their story. Please welcome Emanuel Frenahard, executive vice president and chief digital officer of Santa Fe to the stage. >> Welcome. >> Thank you. >> Thank you. Thank you for having me. What an intimate setting. >> Gosh, I uh [laughter] I love your t-shirt. >> Thank you. Yeah, I went all in. Well, I think we should start by giving folks uh an idea of the scale that you folks operate in. You operate across R&D, manufacturing, commercial, and all of it is powered by data. Take a step back and walk us through the early steps that you took to become an AI native company. So we have to roll

time back five years. Before we engaged with Snowflake, we had >> this is preGPT. >> PreGPT we had many data lakes. To be honest, they were data dumps and um and we we went all in with Snowflake because you we saw that bringing the cohesion or the continuum of data was a necessity for us in a industry where you do drug development, drug discovery for patients, you need the data continuum. And what we really liked about Snowflake where the separation you had between compute and storage that was powerful for us but let's be honest radar while we consolidated massive fragmentation on snowflake really what we got out of it were thousands of PowerBI dashboards and so the power of the data was still untapped. Well, so we will we we will come to that. But even in that early journey, you chose us as a partner which we are very grateful for. Can you talk a little bit about like how you decided to partner with us as you embark on the AI journey of how do I actually get value from all of the work that uh you did with data? >> Well, you had a little role to play in this yourself. The re I think the reality was the commitment you've made and the joint partnership that we created. >> That's right. That's right. >> And then we tapped into one of your hidden powers, your four deployed engineers. They're embedded in our teams. They live and breathe our problems and they truly are some of the best talent we've ever worked with. So I'm going to take this small stage to do a shout out to Vivvec Love Mac Tyler Ashkin Way and your teams who are just outstanding and I'm sure I'm missing some. >> Well, thank you. I remember talking to your team when I was in Toronto. I met with all of the folks in one of the data teams. It was an inspiring conversation. They were so keen to learn more and uh do more. A lot of folks talk about the transformation to becoming an agentic enterprise. You're one of the few folks that has actually done this transformation across multiple departments within Sani. Give us some examples. >> So we haven't done it alone. We've done it with yourselves and a company called Elementum. >> Yep. >> Yep. >> And um about a year ago, we wanted to change a paradigm. And let me just roll back. If you remember when we had those system of records, we had to build massive data pipelines to move data in the system of record and then move data out of system of record into the data lake and how many data engineers spend their lives doing that. And for those of us who leverage large SAS players, we realized that getting access to the data was getting more complicated and more expensive and it shouldn't be so because at the end of the day, it's our data. What was interesting with the partnership with Snowflake and Elementum was the fact that we decided we could build AR workflows straight on the data lake. No friction with a system of record just running on data lake and those are deterministic and agentic workflows that support our end to-end processes. Now don't get me wrong, we're not replicating enterprise applications on data lake. We're tailor designing new workflows that are specific to our company, our industry for things like procurement. We spend about 18 billion euros a year in procurement. Surely we can do better with Agentic for IT support, for HR and employee support, but also for sales. And I think that's combined change is where the value comes from. >> That's amazing. What is always inspiring when I talk to Emanuel is how much of a cleareyed first principles thinker he is. And I think we see this happen certainly at Snowflake at Santa Fe but at more and more places where people are disagregating what they think of as applications >> and in the places that matter to them they're building specific workflows because you operate at such a scale that you know 3% of savings from your 18 billion procurement budget is massive but in AI I have one rule certainly with all of my teams we have to show we just can't talk and uh I do demo days with a lot of my teams every Friday they show off what they have done for the week and uh I think we should do something brave we should show a live demo on live data and so Emanuel take it away what do you have for us tell us what it is doing and what you're going to do >> I'm going to give us a little bit of context for the for the sake of these next few minutes I'm going to be a sales rep we are a pharmaceutical company. I'm going to be a pharmaceutical sales rep. >> What we're going to do is we're going to hit production systems. >> And you're about to go see a doctor. >> And I'm about to go and see a doctor. >> Yep. >> So, I'm going to call concierge, which is our AI concierge, much like you have in a hotel. We call it concier because we wanted to make it feel very personal. Let's go.

>> Concier for field. >> Hey, concierge. I'm on my way to see Dr. Hannah Clevenger. Can you help me pull the pre-call plan, please? >> One sec. [snorts] I've got your pre-call plan for Dr. Hannah Clevenger. She's strategically important for Dupixent in gastroenterenterology, focusing on eocinophilic esophagitis. The biggest opportunity is discussing patients who are still on high steroid use, particularly one patient who has been flagged as needing attention. You may face push back if she feels. Tell me how many patients she has on Dupixen today. >> Currently, Dr. Clevenger has initiated one patient on Dupixent in January 2026. >> Oh, that's very recent. >> Which marks the pract >> Yes. It indicates recent activity and potential openness to advancing therapy. Would you like to know more about the patient demographic? >> I I need some help. I'm about to walk and meet her face to face for the first time. What would be a good icebreaker? A good icebreaker could be asking Dr. Clevenger how she's been approaching eocinophilic esophagitis treatment with her patients recently. It shows your interest in her experience and opens up the conversation naturally. Would you like more tips? >> No, I think you gave me the confidence that I needed. Can you email that to me? >> Preparing that for you now with the email I have for you on file. I've emailed the plan to you. >> Thank you. Wish me luck. >> Good luck. But with a million dollar smile like that, you won't need luck. Go get it, Tiger. >> So, I think everybody learned it's not easy to be a farmer sales rep. There was more terminology that the concers know than I do. >> That is truly truly incredible. It's really exciting to see what you're building at Santa Fe and thank you for the amazing partnership. I look forward to doing so much more together. >> Thank you so much. I think it may be an opportunity for us to wave some goodbye to some of the SAS platforms as you saw. [laughter] >> Thank you, Emanuel. Santa is just one of the many organizations organizations that are putting the power of agent AI into practice. Snowflake intelligence and Cortex code empower your entire organization to tap into the power of AI and they enable you to do so with trust. Trust that your data is governed. Trust that every insight is grounded in the full context of your business. And trust that every action your agents take operate within the guard rails that you define. That's what Snowflake delivers. And that's how you move from experimentation to production to transformation. Now, none of this would be possible

without the stunning advancements we are seeing in AI and Antropic has been at the forefront of this progress, expanding what's possible with AI while helping define how these systems can be built and deployed responsibly. And to help us where this technology is headed and it what it means for businesses everywhere, we are joined by one of the people shaping its future. its future. She's a leader in building AI that is powerful, responsible, responsible, and trusted for enterprise use. Please join me in welcoming Daniela Ammo, president and co-founder of Anthropic to the stage. >> Thank you. Thank you. >> Thank you for being here. >> Thank you for having me. It's wonderful to meet all of you. Well, I will just wish you congratulations on uh today's amazing news. I know you can't uh talk about it. >> For those of you that don't know, you might want to check your news headlines. >> I don't even know what you're talking about. [laughter] >> Now, as co-founder of Enthropic, you talk to enterprise leaders probably every week. And last year that conversation was about experimentation seeing what was possible and what's changed in the one year and what are you seeing that excites you about what these folks can do today >> you know when I think back a year ago I mean in AI like time is this crazy construct right a year ago >> feels like 10 years ago >> it feels like 10 years ago right five years ago nobody was using generative AI or LLMs in their daily workflows in their businesses and now every major enterprise every major company says you know this is a foundational part of our workforce strategy it's a foundational part of our coding strategy financial services legal healthcare every industry around the world is using artificial intelligence and I really think a big part of the reason why that's true is just how much better the models have gotten in the course of the past year >> and I think the part that that is most mind-boggling even to us at Anthropic is what are things going to look like a year from now? >> Well, that naturally leads to the next question. We are all learning new phrases. Wipe coding came um and went to be honest with you last year. Uh but we're talking about things like uh scaling laws. >> Yes. >> And exponential progress which to your point is really hard for humans to comprehend. like something that is getting better by 30% every month. What happens to that in two years? It's like uh what how do people even think about how to plan how to think about what they should be doing and the future in an environment like that where there just seems to be so much progress. >> So I I love that you are asking about this concept of scaling laws, right? this idea that there's this kind of predictable way that if you give the models, you know, more compute, um, more data, they get they get smarter, they get better. And I think even, you know, for those of us at Enthropic where we live and breathe these models and we spend all our time, you know, charting them on on these graphs and thinking where are they going to be a few years from now, it's still really hard for us to internalize just the level of capability that these models are developing every year, every six months, every 3 months. And so the advice that

we tend to give to customers is to say don't necessarily Yes, you have to build around what the models can do today, of course. And so do we. But think big, dream, think about, you know, what is the best version of my product or my company or what I want to build for for my end customer because these models truly are developing at such a rapid clip. And of course, maybe that might change one day, but everything that we have seen indicates that the rate of progress is going to continue. And so when you're thinking about what do I really want to do with AI, don't just think about today. Also think about what is the biggest version of what I want to build and how do I start building towards that. >> Well, which naturally leads to this next question of everyone feels an urgency. >> Yes. >> To act right now. >> Yeah. >> To show results because so much is possible thanks to the amazing models that you folks are creating. But how do you balance speed with safety? What are things that could potentially go wrong that people should be mindful about? >> So, Anthropic has always aimed to be, you know, very open about what the drawbacks and kind of challenges are of AI. And we sort of get asked a form of this question sometimes, you know, how do you balance, you know, innovation and safety? And I think, you know, part of why we've chosen to primarily build for businesses, build in the enterprise, partner with Snowflake is really this concept that trust is an accelerant. Trust is something that helps you go faster. I've never had a customer meeting, maybe you have, but I doubt it, where, you know, the CEO said to me, "I would love if Claude could hallucinate more." Right? They've never said it would just be great if Claude was less predictable and better at producing, you know, bad bad outputs. And so I think this concept that sometimes it's important to to do the safety work and to build trust with your customers so that you can go faster, right? So that you can be an accelerant. I think that's so much of the foundation that both Snowflake and Anthropic are built on. And I just see a huge amount of potential for that to really be the building blocks of of the future. >> Yeah. Yeah. Yeah. Um, >> now I know I'm supposed to ask you some questions, >> please. >> please. >> Is that right? >> Absolutely. >> So, one thing I'm curious about is Anthropic and Snowflake. We've been working together, you know, for some time now. Y >> and I personally think this is a really special partnership. What is it do that you think makes this relationship so unique? So we've actually my team and I have been working with Anthropic actually even predating our joining Snowflake. We were at this little search startup called uh NEA and we are working with you even uh even back then. But you already hinted at some of this which is we are both organizations that are deeply committed to doing what is right for the customer. Um that and doing it in a responsible way. One of the foundational elements of our AI strategy for example in an early product called Cortex Analyst which is a texttosql product was we very openly and deliberately said precision getting answers right was just as important as recall how many questions you answer >> and we were very open about the fact that it is fine to trade away recall don't answer some questions you should not answer but be good with the questions that you do answer. That's the thing that compounds trust. That's the thing that lets people think, you know, this is a partner that is being open about what is working that is giving us the tools that we need for us to judge what is not working and make appropriate decisions. I think it's that mentality that's always been super aligned. There

is also the close working partnership where we work with model previews from you um and give you feedback on what is working well and uh you know what are things that could be uh that could be improved. It's that tight partnership, tight relationship at the engineering team level but also when we go and talk to customers and a similar mentality of we know our role is to help you succeed. Having the same attitude of humility, I think goes a long way in us being successful together. And I'm genuinely super grateful for the close relationship that we have. >> Yeah, we feel the same way. You know, I have to say I'm astonished at how many people are here today. We were talking backstage. Something like 20,000 people are in attendance. >> 20,000. >> 20,000. >> More than 20,000 people. And you know it really makes me think about how quickly Snowflake and Anthropic have gone from you know ideating and thinking big to just executing together. Where are you seeing the biggest impact for customers? >> There are tons of examples. >> I'll give one of my favorites which is with uh with Block the payments company and uh we collaborate extremely well together. Block being a payments company that's always operated at scale has been has always been about oh lots of data. They're also an exceptionally strong engineering team and we are working together Snowflake powering a lot of their data platform and Anthropic with um with amazing amazing products like cloud code helping that engineering team work together to gain all kinds of new insights whether it's into real-time fraud or payment processing that they simply did not have access to before. And that's just one. There are many other examples like this that repeat itself over and over again. I think this is uh honestly the beginning of and a truly amazing partnership of creating value for uh a ton of customers. I have one final question for you. Anthropic has gone from being a little startup to a not so little startup to being a company that is operating at an unimaginable scale. What has that experience been like? >> Um, very chill. I mean, out of the gate, [laughter] very relaxing, getting a lot of sleep. Um but you you know I think in reality it is really just a microcosm of what's happening in the artificial intelligence world more more broadly. Yeah, >> things are developing so quickly and I think inropic you know we constantly feel this challenge of keeping pace and I think one of the things that we have you know worked so hard on is how do we you know grow at the rate that we need to to really meet the moment and support our customers you know as best as we can and it's really hard because you know five years ago Anthropic was about 15 people >> and today were about 3,500 people. And so if you just do the math on the growth there, there's not just scaling, you know, the models themselves, the products, the engineering, the go to market, but there's scaling the company too. I think the thing that has enabled us to be able to grow at the rate that we've grown at is this really sort of fundamental foundational emphasis on our values and our culture. So always coming back to these touchstones of why we're doing what we're doing, right? We care about developing artificial intelligence responsibly and safely and with trust at the center because that's so important for our mission, for our customers, for our values. And I think that has allowed us to sort of manage this this wild ship of just, you know, change and and growth, but it's hard. And I think it also helps build a lot of empathy for our customers who are also experiencing, you know, massive change from artificial intelligence, which by the way, we are too. We're going through the same process of integrating Claude into our tools, into our systems. But I think it it really just highlights how quickly things are changing in this industry. Well, in some ways that's a relief to hear because all of us have to manage crazy change in how much businesses have to evolve, but also how much our companies have to evolve, new job definitions, old job definitions made defunct. There's a lot of change. And uh I love what you said about dealing with all of this with uh empathy. There's no better message for all of us. Daniela, thank you for being with us. Thank you for your wisdom in how we navigate this world of AI. I'm truly grateful for the partnership that we have. Thank you. >> Thank you. >> Thank you >> Thank you >> so much. [applause]

>> I have to tell you, it's uh truly inspiring. inspiring. to have somebody that's so much at the forefront of driving so much change be so grounded, so thoughtful, and so empathetic about what this means for all of us. And her relentless focus on how we create value together for customers, it's also really inspiring to see. Well, the opportunity in front of every organization in this room is enormous. It's not incremental. It is transformational. All of you have the ability to bring together your enterprise data context, your AI models, and your applications to drive coordinated agentic action across your business. And Snowflake is the place where you can freely use AI on your enterprise data without anyone losing sleep over governance, compliance or security. Any AI model, any AI agent already governed, already yours. already yours. Welcome to the Agentic Enterprise. Thank you and have an amazing week. >> [applause]