Unlock the Power of Agentic Data — Tableau Conference 2026 Keynote Replay

[00:00:00] Please welcome Executive Vice President and Chief Product Officer Tableau, Southerd Jones. [00:00:07] Hello, TC! [00:00:12] Welcome, welcome to the beautiful, sunny city of San Diego and welcome to the world's largest [00:00:21] and greatest data and analytics event. [00:00:25] What makes it the greatest? [00:00:27] All of you. [00:00:28] This community, the DataFam! [00:00:33] I want to start by thanking you, thanking everyone here in person for taking time out of your [00:00:41] busy schedule, traveling long distances. [00:00:44] We've had people from Australia, Japan, Europe, Africa, South America, everybody has come here [00:00:49] for this community. [00:00:50] And thank you for those online. [00:00:51] I think we have like 20,000 people in Salesforce Plus. [00:00:54] Thank you for joining us virtually. [00:00:56] And I want to take a moment to thank a few special groups. [00:00:59] A few groups that make this community really what it is. [00:01:02] So when I call out your group name, please stand. [00:01:06] I want to thank our visionaries first. [00:01:08] If you're a visionary, please stand up. [00:01:09] Be recognized! [00:01:10] Thank you for leading us forward with your innovative thoughts and driving our product forward and [00:01:24] driving this community forward. [00:01:25] There's a couple of you I want to recognize specifically because you've been a visionary [00:01:29] for five years and you've done some other amazing things for us and that makes you a hall of famer. [00:01:34] So Tristan Gillivon, yes, Tristan, thank you. [00:01:42] Zach Bowders, Zach, thank you. [00:01:45] And Jim Daynor, thank you. [00:01:47] Next, if you're an ambassador, I'd like you to stand. [00:01:56] Please stand up, ambassadors. [00:01:57] Visionaries, stay standing. [00:01:58] Ambassadors, please stand. [00:02:02] This is so cool to see. [00:02:06] Your evangelism, your ability to bring people together and share your passion has made this [00:02:12] event what it is. [00:02:13] Many people are in this room because of you. [00:02:16] So thank you. [00:02:17] Thank you. [00:02:17] Great appreciation and gratitude. [00:02:19] Stay standing. [00:02:20] Next, we want to thank Tableau user group leaders. [00:02:23] If you have organized a user group, if you presented at a user group, or if you've just been somebody [00:02:27] who's taught others, please stand up and be recognized. [00:02:32] Thank you. [00:02:34] Thank you. [00:02:37] It's a passion for sharing and teaching that makes this community what it is. [00:02:40] Without you, people wouldn't be building data skills, which are some of our greatest life skills. [00:02:46] So a huge thank you as well. [00:02:48] You've all taken us on this amazing journey. [00:02:50] You can stay seated now, but thank you. [00:02:51] You know, it brought us on this great journey. [00:02:54] We've been on this journey for quite some time. [00:02:59] And the journey started for many of us as a first timer at an event. [00:03:06] So if you're a first timer, I'd like you to stand up now. [00:03:10] Who here is the Tableau Conference first time event? [00:03:12] First time. [00:03:13] Stand up. [00:03:14] Wow. [00:03:16] Really? [00:03:17] This is awesome. [00:03:19] Look at this. [00:03:21] Look at that. [00:03:21] Wow. [00:03:24] Wait, wait, stay standing. [00:03:26] Turn to your right or turn to your left. [00:03:27] Give a fist pump or shake hands. [00:03:32] Right here. [00:03:34] This moment is what this event is about. [00:03:37] It's about sharing your passion. [00:03:39] It's about meeting new people. [00:03:40] It's about making acquaintances, making friends. [00:03:43] Some of you might even meet your future spouse. [00:03:46] That's happened. [00:03:48] It has happened. [00:03:49] So thank you for coming out and stepping outside, outside your comfort zone and making yourself [00:03:53] available to meet new people. [00:03:54] That's what this event is all about. [00:03:55] That's what this community is all about. [00:03:57] You can take a seat. [00:03:58] Thank you. [00:03:59] So this journey started in Seattle. [00:04:00] It started with 187 people at our first event. [00:04:03] Imagine 187. [00:04:05] There's like 6,000 people here. [00:04:07] There's 20,000 online. [00:04:08] 187 people to where we are now. [00:04:10] And that is that journey went across the globe. [00:04:14] And because of you, we've reached so many different people. [00:04:17] Because of you, we've been able to reach underdeveloped countries. [00:04:21] We've been able to reach new and unknown places that had never even worked with data before. [00:04:26] Because of you, we are in 48 countries today. [00:04:30] Over millions of people. [00:04:31] 210 user groups. [00:04:33] 57,000 user group members. [00:04:36] 70,000 events on a monthly basis. [00:04:38] And almost every single one of them is because somebody here online is organizing it and setting [00:04:43] up and sharing their passion for what we all want to do is help people see, understand, and act on data. [00:04:50] Such a massive thank you. [00:04:51] And that's the power of this community. [00:04:52] That's what this event's about. [00:04:54] And that's what we're here to talk about over the next couple days. [00:04:57] Now, it's not just about the community. [00:04:58] It is a little bit about this product called Tableau. [00:05:01] And you guys have made that Tableau product great. [00:05:04] You started by asking us for Tableau public, so we made Tableau public. [00:05:07] Thank you for all those who pushed for it. [00:05:09] And thank you, Jack McKinley, for this really cool viz on Tableau public. [00:05:12] It shows all the features we've been releasing over time, and it's a beautiful viz. [00:05:16] That's just an example of what Tableau public is. [00:05:18] It is often the place where all of us have brought light to the causes that we have passion in our hearts [00:05:25] about. [00:05:25] I still think it's one of the best properties online there is today. [00:05:28] It wasn't just public. [00:05:29] You asked for something else. [00:05:30] You said, "Hey, can you make visual analytics better? [00:05:32] Can you bring us Tableau-free desktop?" [00:05:35] Yes. [00:05:36] That's what we did. [00:05:37] That's what we did. [00:05:39] And get this. [00:05:40] A hundred thousand people have downloaded Tableau-free desktop since March. [00:05:44] A hundred thousand people who previously had not been exposed to Tableau, [00:05:49] and now are being exposed to Tableau. [00:05:50] And that's because of us listening to you and together leading forward this evolution. [00:05:55] It wasn't just that. [00:05:56] You asked for a bunch of features in our product. [00:05:58] You asked for rounded corners, custom color palettes. [00:06:02] Yes, rounded corners. [00:06:04] Road time areas in Tableau maps, my new favorite feature. [00:06:08] All of them because of you and us together leading that charge. [00:06:12] But today, today, you can't go anywhere without someone talking about AI. [00:06:18] And you know what? [00:06:19] You know who's leading that? [00:06:21] We are. [00:06:22] You are. [00:06:23] I see you out there. [00:06:24] I see you on LinkedIn. [00:06:25] Vibe coding, connecting to Tableau MCP, building custom apps, leveraging Tableau AI. [00:06:30] I see you leading the future. [00:06:32] And yet, there's a lot of uncertainty because there's so much hype. [00:06:35] But that's something I want everyone here to take away. [00:06:37] In the next three days, you're going to learn a lot about how AI can accelerate [00:06:40] our mission to help people see and understand data. [00:06:43] But there's one thing AI cannot do that all of you have already done. [00:06:48] Any time someone logs into Tableau and looks at a viz or a dashboard or a metric, do you know why they trust it? [00:06:56] Do you know why? [00:06:58] Because someone here has gone to great lengths to make sure it's correct. [00:07:03] Someone here has said, I'm going to make sure the join is correct. [00:07:06] I'm going to make sure the level of decalculation is accurate. [00:07:09] I'm going to make sure the visual representation is right. [00:07:12] I am personally going to make sure that someone can trust this because they make key decisions on that. [00:07:18] AI can't do that. [00:07:19] Only you can. [00:07:20] So, the next couple days, you're going to learn a lot about AI. [00:07:22] You're going to learn about Tableau features. [00:07:24] You're going to learn about how together we're going to accelerate our mission with AI. [00:07:29] Now, right now, I want to introduce somebody who's going to help you and take you on that little journey. [00:07:34] But first, he's somebody who has spent his life in Tableau making key business decisions [00:07:39] based on information and knowledge that came from Tableau. [00:07:41] So, with that, I want to welcome our new GM for Tableau, Mark Recker. [00:07:49] Thank you. [00:07:51] Thank you, Southern. [00:07:53] Wow. [00:07:54] I am thrilled to be here at TC. [00:07:57] This is my 55th day in this GM position. [00:08:03] And I want to start with a thank you. [00:08:06] A thank you to our customers. [00:08:08] A thank you to our partners. [00:08:09] A thank you to the incredible Tableau team. [00:08:11] But most importantly, if you're part of the DataFam, can you stand up? [00:08:16] Can we give them a really, really loud thank you? [00:08:27] TC is about you. [00:08:30] It's about the DataFam. [00:08:32] And listen, I've been looking forward to this day for a very long time. [00:08:36] It's actually the first thing I told my team. [00:08:38] I said, we have 55 days until TC. [00:08:40] You can ask them. [00:08:41] They're right over there. [00:08:42] But I wanted to be here in rainy San Diego as we shape the future of Tableau together. [00:08:49] Now, I've had the opportunity to meet many of you in the last 55 days. [00:08:53] But not all of you, because we know data. [00:08:54] There's thousands of people here. [00:08:56] That would be statistically improbable. [00:08:58] Just yesterday, I had coffee with some folks. [00:09:00] I ran into folks and saw the power of this community and this conference. [00:09:05] But you all don't know me yet. [00:09:06] So what's the first thing you should know about me? [00:09:10] Well, I have been data obsessed since before I can remember. [00:09:15] I was the kid scouring through box scores, trying to understand patterns in signals, [00:09:22] to understand why my Minnesota twins were always losing, which was mostly bad pitching and bad [00:09:28] hitting. [00:09:28] It's a bad combination. [00:09:30] But I love data. [00:09:32] I love it. [00:09:32] I love the visuals, the stories it tells, how it sparks curiosity. [00:09:39] And the core belief that I have that has shaped my career is that when you can see and understand [00:09:45] data, you make better decisions. [00:09:49] And so for the better part of the past decade, my day has started in the same place. [00:09:55] You want to guess? [00:09:56] Tableau. [00:09:58] A cup of Phil's coffee and my Tableau command center. [00:10:03] Now, the Phil's part is an unhealthy and expensive habit, so I don't recommend it. [00:10:08] But Tableau, Tableau would tell me what changed, what needed actions, what decisions were ahead in my day. [00:10:17] So I ran my business on Tableau. [00:10:20] I grew my career on Tableau. [00:10:23] So I think I can safely say that I have been the DataFam's number one DataFam. [00:10:33] Now, today, we have more opportunity with data in analytics than we have ever had before. [00:10:42] And that is an incredible, incredible opportunity that we get to channel for the next three days here. [00:10:49] And that's why we come here to San Diego every single year to innovate, to be curious, [00:10:55] to push the boundaries of what we can do together. [00:10:59] Now, last year, we showed you this slide. [00:11:03] We introduced Agentic Analytics, building on the 20-year journey of the DataFam and Tableau. [00:11:12] Look at the eras that we've conquered before: self-service analytics, visual, augmented. [00:11:19] And in each of those eras, it was you, the DataFam, that took data and turned it into insights [00:11:27] in the form of viz's, dashboards, alerts, notifications, triggers, so that a human could take them [00:11:36] and make decisions and take actions. [00:11:39] But guess what? [00:11:41] The audience for our work is expanding. [00:11:44] In the future, it's not just going to be humans. [00:11:47] It will. [00:11:47] But it's going to be humans and agents. [00:11:50] Agents you create. [00:11:52] Tableau agent. [00:11:53] Third-party agents. [00:11:55] And that requires more from us because we know this. [00:12:00] Data alone is not enough for an agent. [00:12:03] In every era of analytics, there has been a human on the other side of those insights [00:12:10] who had the intuition to look and say, "Hmm, that's not accurate." [00:12:15] Because they have context. [00:12:17] Raise your hand. [00:12:19] I would expect every hand to go up. [00:12:21] If you've ever looked at a dataset or a dashboard and said, [00:12:24] "God, that is not accurate. [00:12:26] I'm going to have to spend more time on that." [00:12:28] Yeah. [00:12:29] We've all felt it. [00:12:30] Guess what? [00:12:31] An agent doesn't know that because they don't have context. [00:12:35] And that is exactly why 89% of leaders say they've experienced inaccurate or misleading AI outputs. [00:12:43] Show me the other 11%. [00:12:46] So, what does that mean? [00:12:49] What do agents actually need? [00:12:52] They need more. [00:12:53] They need knowledge. [00:12:56] They need metrics. [00:12:57] They need definitions. [00:12:59] They need semantics. [00:13:01] They need published data sources. [00:13:04] Is this sounding familiar? [00:13:07] It should. [00:13:08] Because for the last 20 years, it has been the data fan that has been architecting knowledge inside of Tableau. [00:13:18] In the form of, and this statistic shocks me every time, 33 million semantic data models that you built in Tableau. [00:13:29] So, while this era requires and inspires new ways of thinking, new ways we can orchestrate data in analytics, [00:13:39] it requires one thing above everything else, and that's knowledge. [00:13:44] The very knowledge you've built in Tableau. [00:13:49] And so, that's why I'm excited to unveil to you where we're headed with Tableau. [00:13:57] Based on what you've already built inside of it. [00:14:02] So, today, Tableau is your agentic analytics platform. [00:14:10] Now, every layer of this platform is powered by AI. [00:14:16] And show is way better than tell. [00:14:19] So, we're going to show you for about 40 minutes. [00:14:21] We have some fantastic demos. [00:14:23] And we're going to show you that it's in every product experience. [00:14:27] Not just Tableau Next. [00:14:29] I will repeat that. [00:14:31] Not just Tableau Next. [00:14:35] I figured that would be popular. [00:14:38] It will be in cloud, in server, in desktop. [00:14:43] But before we show you, yeah, you can do a pause. [00:14:46] All right. [00:14:49] Before we show you, you have to listen to me. [00:14:53] Because there's four things I want to highlight that you're going to see. [00:14:56] The first is our knowledge engine. [00:14:58] The ability for you to take structured and now unstructured data. [00:15:02] And in natural language, turn that into knowledge. [00:15:06] And semantic data models. [00:15:08] Number two is a knowledge graph. [00:15:12] So you can take and understand your entire enterprise's knowledge base. [00:15:18] Because we know there are not insights or actions without knowledge. [00:15:24] And this is point number two. [00:15:26] Our decisions engine. [00:15:28] This era is no longer about insights decoupled from actions. [00:15:34] It's about insights and actions in real time by humans and agents alike. [00:15:44] And you will see that. [00:15:45] You will see conversational agentic analytics in cloud, in server, in Next. [00:15:51] You will see Tableau pulse. [00:15:53] You will see third-party agents driving decisions. [00:15:57] Because that's what we do with knowledge. [00:15:59] And the third point, and this is very, very exciting. [00:16:04] And I think it's aligned to Tableau's founding principles of democratizing data and trust. [00:16:11] Which is you can take your trusted knowledge anywhere. [00:16:16] Wherever your organization is working through headless analytics and MCP. [00:16:21] Will Sutton is going to show you. [00:16:23] You can take it to Claude. [00:16:25] You can take it to ChatGPT. [00:16:27] To Slack. [00:16:27] You can even take it to Microsoft Teams. [00:16:33] Salesforce. [00:16:34] And Tableau. [00:16:35] So that gives you the power as a data fan. [00:16:39] To tell your organization. [00:16:41] What they need to know. [00:16:42] Before they even understand they need to know it. [00:16:46] And this fourth point. [00:16:49] Which I think in the AI era is more important than ever. [00:16:52] Is our platform is secure. [00:16:54] Governed. [00:16:55] Composable. [00:16:57] And extensible. [00:16:59] So you can safely take your trusted knowledge. [00:17:02] Wherever your organization is working. [00:17:04] And guess what. [00:17:05] Any assets or knowledge you're building elsewhere. [00:17:08] You can bring that back in. [00:17:10] To your trusted agentic analytics and knowledge platform. [00:17:14] Okay. [00:17:15] So what does this mean for us? [00:17:18] I had a favorite line in the video. [00:17:23] It was never about the data or the dashboards. [00:17:27] It was about you. [00:17:29] The data fan. [00:17:31] And in this era of AI with all of this noise that rings truer to me than ever before. [00:17:39] Now we showed you this slide last year. [00:17:42] But frankly we were a bit nebulous about it. [00:17:46] And that's okay. [00:17:47] Because a year later it's getting clearer and clearer. [00:17:51] We have many fantastic roles inside of the data fan. [00:17:55] And in the future they're likely going to converge a little bit. [00:17:59] And in an empowering way turning you into a more technology forward function. [00:18:04] Now you have always been a trust enabler, an outcome whisperer, a decisions orchestrator. [00:18:11] But in the future you're an agentic architect. [00:18:15] You're a decisions architect. [00:18:18] You're a knowledge architect. [00:18:21] And that is empowering you to drive more impact than you ever have before. [00:18:28] You are the composer driving your company's entire analytics and data strategy. [00:18:35] And that is a multiplier to the massive impact you already provide to your organizations. [00:18:42] So what does all of this mean for us, for Tableau, and for the data fan? [00:18:49] It means the same thing it did 20 years ago. [00:18:53] It means opportunity. [00:18:55] And so for the next three days, we'll start building that future together. [00:19:00] As we always have through every era of analytics. [00:19:05] So, guess what? [00:19:09] The tell part is over. [00:19:11] We're going to show you, across every product experience, how we architect knowledge, [00:19:16] how we power decisions, and how we agentify actions. [00:19:20] But before we do, I would like to end the same way I started. [00:19:24] The last 55 days have been incredibly inspiring to me. [00:19:28] I'm wildly excited about our future. [00:19:31] But I'm most excited to build it with you. [00:19:35] So thank you to the data fan. [00:19:37] Thank you to our customers. [00:19:38] Thank you to our partners. [00:19:39] Thank you to the incredible Tableau team. [00:19:42] And with that, Southern Jones. [00:19:51] All right. [00:19:52] Thank you, Mark. [00:19:53] And just as Mark said, we are going to show you product now. [00:19:56] That's the fun part. [00:19:58] We're going to do something a little bit different this year. [00:20:00] Instead of a Tableau product or engineering people showing you product. [00:20:04] You get to see that with devs on stage coming up soon, tomorrow. [00:20:08] Today, yes. [00:20:10] I'm excited for that, too. [00:20:12] Today, we're actually going to have people from this audience. [00:20:15] Customers of Tableau and community members like you show the product. [00:20:19] So, I'm going to start by bringing on stage someone who's been with [00:20:23] Tableau for quite some time. [00:20:25] Please join me in welcoming the VP of Global Retail Analytics, Chad Stroup. [00:20:28] Chad, come on up. [00:20:35] Chad, good to see you. [00:20:36] Thanks so much for being here. [00:20:37] Really appreciate everything you've done. [00:20:39] Disney's been an amazing leader and at the forefront of data analytics for quite some time. [00:20:43] So, really appreciate you coming here to represent. [00:20:45] Yeah. [00:20:46] Thank you. [00:20:46] Thank you. [00:20:46] I appreciate it. [00:20:48] If we're going to do this journey, we've got to do it right. [00:20:50] And so, what I brought was Stitch. [00:20:52] I heard that might be your favorite character. [00:20:54] Is that supposed to be me? [00:20:55] This is you, right? [00:20:56] Chaotic, mischievous, problem-causer. [00:20:58] It's in trouble, but a disrupter, right, of the market. [00:21:02] And in the end, everybody loves Stitch, right? [00:21:05] So, here we go. [00:21:06] Oh, boy. [00:21:08] You might need to do that. [00:21:09] You can hold it. [00:21:11] I'm going to be Groot, like baby Groot, deeply rooted in data, [00:21:15] but continuously growing. [00:21:21] We've got to do it right. [00:21:23] This is awkward. [00:21:26] And a special thanks to my account manager, Eddie, here. [00:21:29] He's the force. [00:21:29] So, he's got a baby Grogu. [00:21:33] Yes, let's do this. [00:21:35] Thank you. [00:21:36] Now that I'm embarrassed and red-faced. [00:21:39] I'm supposed to ask you some questions. [00:21:41] No, in all seriousness, you know, you've always pushed us in terms of where you want to take Tableau [00:21:47] and how your users are using this. [00:21:48] Maybe you can share a little bit about what we could do better with Tableau today [00:21:52] to help your users be even more powerful. [00:21:53] Sure. [00:21:54] Thanks. [00:21:55] Well, great, Mark, for sharing the early vision here. [00:21:59] It's exactly what we're looking for. [00:22:01] So, at Disney, what we're thinking about is how do we get away from the reporting layer to a decision [00:22:07] layer? [00:22:07] Exactly what you kind of mentioned there. [00:22:10] So, for us, it's really important to be connected to the systems that are making the decisions, [00:22:15] right? [00:22:16] Today, there's a bit of a gap. [00:22:18] Our users are in the tool. [00:22:19] They're looking at dashboards. [00:22:21] Then our location planners, for example, might be going down Main Street, early in the morning, [00:22:26] walking through the Emporium. [00:22:28] They saw a dashboard, said, okay, here's our top sellers. [00:22:32] Here's how to stock. [00:22:33] But then they've got to go back to the office. [00:22:36] And then they've got to make the decision. [00:22:38] What do I do? [00:22:39] How do I, you know, make that guest experience better? [00:22:43] And so, what we're asking is how do we get that closer and connected to the systems where [00:22:48] I'm actually making change? [00:22:50] So, that's for us is if we can do that continuous loop, bring us closer to that real time, [00:22:56] that would make a big difference for us. [00:22:58] Well, that is good to hear. [00:23:00] And in about three and a half minutes, hopefully, we'll show that to you. [00:23:03] So, I'm excited. [00:23:04] Thank you. [00:23:05] So, the other conversation that you can't escape these days is AI. [00:23:09] How do you see AI change in the way your users interact with data? [00:23:14] Yeah, good question. [00:23:15] I'm, you know, we have some of the best data, I think, around, right? [00:23:19] We understand our customer very well. [00:23:22] But we're excited. [00:23:23] I mean, really excited about agentic AI, conversational AI, and really using that with Tableau. [00:23:31] So, today, our dashboards are great at answering the question they were designed for. [00:23:37] But what it doesn't do is answer the next question, right? [00:23:41] And there's always a next question and a next question. [00:23:44] And so, for example, we have great dashboards that are monitoring our real-time sales in stores, [00:23:50] looking at conversion. [00:23:52] Conversion might be working well for an item. [00:23:54] It might not be working well for this item. [00:23:56] Why is that? [00:23:57] You know, I want to be making, you know, ask that second question. [00:24:02] Is it, are we out of stock? [00:24:04] Maybe this item sold great, and we're in stock, but now it's not selling. [00:24:09] Well, guess what? [00:24:09] It might not be on the floor. [00:24:11] So, how can I use agentic AI to, you know, help me answer those second questions? [00:24:18] And really, the way I think about it is Tableau and AI together become my personal data scientists. [00:24:25] Right? [00:24:25] They're by my side. [00:24:26] I want to give it a name. [00:24:27] I want to call it something. [00:24:29] They'll be my best friend, right? [00:24:31] And help me really answer those questions real-time. [00:24:33] Well, hopefully, yes. [00:24:35] You agree with that, yes. [00:24:36] Hopefully, we can show you that today, and I think that's what we all want. [00:24:40] A, we want to create that trusted advisor, but B, under the covers, [00:24:45] we want to know that we can trust what that agent is telling us, which is the key thing there. [00:24:49] All right. [00:24:49] So, last thing is, as I said, you've always been wanting to help us bring us towards the future. [00:24:54] Maybe you can do a little future telling three years from now. [00:24:57] Imagine coming back here in three years. [00:25:00] How would you see, how would you like describe the way that Disney is using analytics three years from now? [00:25:05] Yeah, well, great question. [00:25:08] In my mind, we're not talking dashboards like we talked today, right? [00:25:12] I want to be naturally conversing with Tableau. [00:25:15] I want to be, you know, really connected to the systems like I talked about. [00:25:20] You know, I want Tableau and AI to be continuously monitoring my business, [00:25:25] looking at trends, looking at park traffic, looking at out of stocks. [00:25:30] And I want it to really help me answer that question that I haven't thought about yet. [00:25:35] That's right. [00:25:36] Letting you know what you didn't even know to ask or something like that. [00:25:40] Exactly. [00:25:40] Well, we look forward to doing that with you. [00:25:43] So, thank you again for your partnership. [00:25:44] Thanks for embarrassing me with Stitch. [00:25:47] Really great to see you here. [00:25:48] Thanks so much. [00:25:49] Appreciate it. [00:25:55] All right. [00:25:57] Next up, to show you the product, someone who I think is a perfect person to show you the product, [00:26:04] because he himself has been in Tableau probably longer than most people in this room. [00:26:08] He was an early member and a huge fan of data and analytics, specifically with Tableau itself, [00:26:12] and he can tell you his story. [00:26:13] And today, he is the chief data officer at Salesforce. [00:26:16] Please welcome Michael Andrew. [00:26:19] All right. [00:26:21] Boom. [00:26:23] Everybody. [00:26:25] I am so excited to be here today to share what we're doing at Tableau. [00:26:30] And let me start with this. [00:26:32] I've been now using Tableau for almost 20 years. [00:26:35] Back in 2007, yes, I'm a little old. [00:26:38] I discovered, oh, my God, I can see my data. [00:26:42] I started my career, like so many of you, as a data analyst, looking at data, making charts, [00:26:48] finding those insights that could move the business forward. [00:26:51] And then I built a business on Tableau. [00:26:54] I was a customer. [00:26:55] I paid for Tableau Server. [00:26:57] I was providing analytics to some of the best brands in the world, like Nike and Google and Verizon. [00:27:03] And I made sure I had Tableau Server. [00:27:05] I made sure every single analyst in my business had what they needed, had Tableau to do this analysis. [00:27:12] And then when I joined Salesforce in 2019, there was something peculiar. [00:27:16] They didn't have Tableau. [00:27:17] I was like, what's up? [00:27:19] Come on, guys. [00:27:21] And believe it or not, I took about 40 emails, a bunch of meetings, escalation, all the way to CIO, [00:27:26] so my data science team could get Tableau. [00:27:28] I was successful. [00:27:30] We got Tableau. [00:27:31] That's right. [00:27:31] We started using it to do some real insights. [00:27:34] And then a little later that year, Salesforce kind of wised up and said, well, maybe we should buy the company. [00:27:40] I don't know if I had any influence on that, but I'm going to give myself a little bit of credit, maybe. [00:27:45] And what I'm really proud to say is now, more than seven years later, as a chief data officer at Salesforce, [00:27:52] we run our business on Tableau. [00:27:54] Every single day, tens of thousands of employees are logging into Tableau. [00:27:59] They're viewing the dashboards. [00:28:00] They're making decisions, whether they're in product, whether they're in sales, HR operations, [00:28:06] everybody. [00:28:07] And I would love to show you all of our real internal data, but that might be a little irresponsible [00:28:14] and not fitting our trust value as a chief data officer of a public company. [00:28:19] So today, we're going to show you this story through Bolt Bikes, a really awesome imaginary company [00:28:27] that sells awesome e-bikes to everybody so that we can show you what Tableau of the future looks like. [00:28:34] Now, to do this, please give a major shout out to John Denby, our awesome demo driver. [00:28:42] All right. [00:28:45] So we're looking at Bolt Bikes, and sales are going pretty good. [00:28:48] So here's my e-commerce sales, having a lot of great acceleration, but that only tells one story. [00:28:54] I kind of want to know not just e-commerce, the whole thing. [00:28:57] Now, in the past, when you want to join this data, let's say data in one database, [00:29:02] data in another database, you have to blend the data. [00:29:05] What we are now announcing is something called composable data sources, where you can take [00:29:11] different data, different published data sources, and make them as one. [00:29:15] Actually do the join. Materialize it. Yeah. Yeah. [00:29:21] I know. Excited. I'm excited, too. So look at this. [00:29:26] So now we're joining our -- of course, for some reason, they had two databases. [00:29:30] That never happens in the enterprise, right? You never have data in different places, do you? [00:29:34] And so we're taking, now, the web data with the retail data, all in one view, all explorable. [00:29:41] That's pretty cool. Composable data source to everybody. My team is super excited. [00:29:45] And so what can you do with it? Well, now you can make super awesome Tableau dashboards like this. [00:29:50] Beautiful. You can see the whole world. So I'm looking at it. I'm seeing my sales. [00:29:54] I'm seeing things going around. But maybe I have some questions. [00:29:57] Questions about, well, how are we doing? How are these bikes being sold in San Diego? [00:30:02] Now, if you had to do this today, maybe you'd create a filter. Maybe you'd create a tab. [00:30:07] But now we are offering to all of you agentic analytics right in Tableau cloud, right in server. [00:30:13] Now, you can ask the agent this question, well, what was shipped in San Diego last month? [00:30:19] And what did we get? It's thinking. It's thinking. It's coming. It's computing. Boom. [00:30:25] We can see that San Diego is the most popular for e-bikes. Yeah. Shredding some of the trails around. [00:30:31] And all of this is grounded with more than 95% accuracy. Because I don't know about you, [00:30:38] we've had some vibe coded dashboards showing up around Salesforce, and they're basically totally wrong. [00:30:43] Beautiful, but completely inaccurate. So I really care about making it accurate. And you can see all [00:30:48] of this is sourced on that data. But how are we getting to that answer? How are we enabling these [00:30:54] agentic analytics? Behind the scenes, we are offering to all of you an analytical knowledge graph. [00:31:02] This connects to all your data. It understands it, the semantics, the maps. Don't worry. You don't [00:31:08] have to build this. We're building this for you automatically. Now, you can modify it. You can tune it. [00:31:16] It will keep self-learning, and you can adjust it. You can feed the intelligence. And it's this grounding [00:31:22] in the knowledge graph that gives the context so you can trust the agentic analytics. Amazing. I can't [00:31:29] wait to see all of you use this. Now, why don't we switch over to Tableau next, and let's talk about a [00:31:36] semantic model. Semantic models, while amazing, maybe sometimes are a little tedious to build, so we're [00:31:42] offering you new tools to speed up your semantic models. Call it AI to build your AI. Where it can [00:31:48] suggest fields. It will check if you have conflicts. We have our little Einstein magic kind of wizard. [00:31:55] And here we go. Optimize the model. So all of this, again, is to help you more quickly get to [00:32:02] putting your intelligence to describe your data, to model your data, so that you can ultimately power [00:32:08] trusted analytics. And why do you want to do this? So you can put agentic analytics in the flow of work. [00:32:15] So why don't we switch over, and here we are in Slack. And here's Jennifer, our awesome CEO of Bolt [00:32:21] Bikes, and she wants to know some questions, right? Well, what's really happening? What are the top bike [00:32:27] models doing? And right there in Slack, she can ask the question. It's going to think. It's going to, again, [00:32:34] use that same trusted knowledge graph, the trusted data, to be able to pull it up, get an answer. [00:32:41] And there you go. So electric bikes are crushing it. Mountain bikes, doing okay, I think. But look at [00:32:47] that. You also have a recommendation. And what's unique about this is it brings the insights and [00:32:53] conversations your teams are having with the intelligence. But you don't just want the one answer. [00:32:59] You want to be able to take that intelligence and turn it into action, turn it into a recommendation, [00:33:04] begin to make a decision. And then you can produce this. Your report automatically right there. Again, [00:33:11] grounded in the trusted data, having the key insights, and having strategic recommendations [00:33:17] right in the flow of work. All of this. So, again, you pick your data, you connect it, the graph builds it, [00:33:25] and here we go. So, what did we just show you? We just showed you a lot. So, we're going to go to [00:33:32] the slides now, and let's recap. We showed you a few things. We showed you the composable data sources, [00:33:41] which enable you to now bridge all of your data into one data model you can work with. We showed you the [00:33:47] conversation analytics in cloud, in server, coming to desktop. We showed you the automatic analytics [00:33:53] knowledge graph that understands your data, maps, builds it. And we showed you how you could bring [00:33:59] all of that into Slack. So, all of this is to ensure that all of you can keep having your data [00:34:07] grounded in the trust and knowledge you do every day, but empower that AI-driven agentic analytics [00:34:14] with 95% accuracy. And I'll just say a personal story. This is literally what we're worried about [00:34:19] Salesforce. We have those dashboards. We have a few maybe sales leaders kind of excited, like, [00:34:25] "Oh, my God, if I coded this thing." And we look at it, and we go, "Yeah, but every metric is wrong." [00:34:29] Right? And that's the danger. If you just kind of use the AI itself without this grounding, [00:34:36] you're going to get hallucination, you're going to get security, you're going to get gaps. We want to [00:34:39] make sure that all of you and all the work you and the data fan put on your data can make sure that [00:34:44] your company runs and has trusted decisions. So, we just showed you how you can become the architect, [00:34:50] the knowledge of your company, and Rayka's going to come up and show you power decisions. But first, [00:34:56] we want to cut to a film from Engine, one of our great customers, of how they're using Tableau to [00:35:01] succeed. So, let's take a look. [00:35:06] The role of analytics is like, what's the role of food and water? It's incredibly important. [00:35:14] Over 30,000 businesses, over a million travelers are on the platform now. [00:35:20] Engine is a modern travel and spend management platform. It's a very big space ripe for disruption. [00:35:25] Things come together not in weeks, months, years. Things come together in minutes, hours, days. [00:35:31] We were all doing the best we could with the tools and resources we had available to us at the time. [00:35:35] There was only so far you could go with that. Salesforce is the operating system that makes [00:35:41] it seamless. It's the platform that the world runs on. [00:35:46] Agentec Analytics and Tableau change the way that we think about data. [00:35:50] It's not just for my team, it's for the full company. [00:35:52] Without it, they would be different silos. We wouldn't be unified. [00:35:55] AI is only as powerful as the systems that you connect it to. [00:36:00] With natural language, I'm able to tell it how I want to join some tables and all of a sudden it [00:36:04] builds a semantic model for me that I can then deploy. Then have another agent go and analyze [00:36:08] data without me having to do anything other than type some sentences. [00:36:12] It's the ability to interact with data in ways that were otherwise not possible. [00:36:17] It changed everything. [00:36:19] CSAT goes up. Costs to deliver go down. Customers are happier. [00:36:23] We're getting the mantras faster. What's the trade-off? There's no trade-off. [00:36:27] And now Slack is essentially an extension of them. [00:36:31] Slack to us is more like an operating system. I'm able to talk to Slack bot and get all of [00:36:36] these details ready for me so that I have them in front of me. [00:36:39] Having that information available to all in Slack right then and there at your fingertips gives us [00:36:44] the ability to iterate that much faster. I can interact with our analytics agent right from my phone [00:36:49] and be like, "Hey, what's going on with this? This looks funky. Like, tell me what the deal is here." [00:36:52] They can dive into the data and give me some insights. [00:36:54] Have data to inform the direction you build and then have data after you build to inform your iterations, [00:36:59] to inform if you made the right bets. It's going to be revolutionary. It's going to be game-changing. [00:37:04] We're going from the dumb software era to the smart software era. That's literally what's happening in real time. [00:37:09] It's an incredible opportunity for all these businesses. If you run on Salesforce and Tableau, you're nuts if you don't. [00:37:15] Thank you. That was an incredible story. So many companies, so many organizations like Engine have [00:37:28] been using Tableau to power their day-to-day. We spent the last 10 to 15 minutes talking about [00:37:34] architecting knowledge. That is critical. That is the foundation for AI. Great. Now, how do we take [00:37:41] that knowledge? How do you take your data? How do you take your semantics, your context, and actually power decisions? [00:37:48] And that's what I'm going to show you in this next chapter. But, hello, data fam. This is Rekha. [00:37:55] And in my role, I get to be in a lot of meetings where big decisions happen. You know, what's the catch? [00:38:03] Unfortunately, with access to all the AI tools, all the reports, all the dashboards, [00:38:09] the whole decision-making process itself is slower, it's harder, and it's more disconnected than it [00:38:17] needs to be. Hmm. I'm sure we all know a better way to do this, right? To do that, let's zoom out [00:38:24] for a second. Like I said, all of us have access to all your tools, your Slack, your Teams, OpenAI, [00:38:31] ChatGBD, you name it, you have it. You all have a question. What do you do? Go to that, ask a question. [00:38:37] The problem? Every time you ask the same question, you get a wrong answer or a different answer from any [00:38:45] one of these tools. 99.9% of the time. Same question, different answer across the tools. [00:38:53] That's the problem. The problem right now is not about just getting access to your data. It's also not [00:38:59] about getting the answers. It's about getting the right answers at every single time. Now, [00:39:05] is it all of us, are we going to go back and check every single tool to find the right answer? [00:39:11] Do you have the time to do it? I doubt it. So, there must be a better way, and that's why we have [00:39:16] Tableau. What Tableau does is we are able to build models based on your data. It's your trusted knowledge, [00:39:25] your data. Now, composability and extensibility have always been in our DNA from the day we started [00:39:33] Tableau. And if you were to break down those jargon words, what it actually means is you can use Tableau [00:39:40] to power every surface, every agent, and every workflow based on your trusted knowledge. That is our knowledge [00:39:51] engine in action. And that is an absolute game changer and one that gives up our productivity in extreme [00:39:58] ways. And that also reimagines our decision making, right? Now, think about it. You're working wherever [00:40:04] you are and you're able to pull up your insights and you're able to take action right there. No toggling, [00:40:09] no going back and forth, and all of those answers are correct and accurate because it's based on our data, [00:40:16] your data. So, here's the fun part. We took all of this powerful knowledge and put it in the hands of [00:40:25] all of you. What do you all do? You create magic. You go beyond creating dashboards and you create [00:40:32] magic with this. You reinvent new ways of working on how we do, how we operate. The best part of Tableau, [00:40:40] and I've said this every time on every stage of every one-on-one meeting I've had, is DataFam. You're so [00:40:46] central, you're so critical to our strategy, and that's how we do it. That's our secret sauce, and [00:40:53] that's how we operate. So, I'm really excited to show you our new DataFam AI and analytics showcase. [00:41:00] This is where your every project, your every breakthrough, your every big bold idea that you've all created [00:41:07] is on stage. So, go ahead, scan, get inspired by that QR code and that page that we've built because, [00:41:15] you never know, one of your big projects can be on this very stage at the keynote next year. That's [00:41:22] our promise to you. So, speaking about incredible AI showcase, I have the distinct pleasure of welcoming [00:41:30] an incredible visionary from the community who has done some amazing things. To do that, please welcome [00:41:37] Will Sutton on stage. [00:41:47] Hi, Will. So excited to have you here. Thank you for joining us. Thank you for having me. Okay. A little [00:41:52] birdie told me you were a past IonWiz champion. So, what's changed since then? I kind of feel that [00:41:59] everything's changing, but also nothing is changing at the same time. I started using Tableau in 2013, [00:42:06] and I didn't really know what I was doing, but it was thanks to the folks here. The Tableau community [00:42:11] gave me the skills and knowledge to go and take on that contest. I stepped away after that contest. [00:42:17] I thought, well, this is it. My career is set. You know, I've got a long, happy life building dashboards. [00:42:24] Yeah. How wrong was I? The next year, ChatGVT was out. Huge buzz in the market about that, but it [00:42:31] hasn't stopped. And I feel even now, like every week, every day, there's something new coming out. And [00:42:37] while I see a lot of change happening there, I still come back to this group here. They're the people [00:42:42] that give me the grounding to say, what skills do I need? What things are actually going to shift the [00:42:47] dials? What my clients need to know about this stuff? These are the people that helped me stay [00:42:51] on top of this. Oh, wow. So a lot has changed. Also, not a lot has changed. That's cool. You also [00:42:57] built the Tableau MCP on Lantern. You were the first one to do that. What drove you to do that? Did you do [00:43:03] that? Yeah. So as part of this project, I think it was a lot of curiosity of like, well, what would [00:43:07] happen if you took the Tableau environment and just connected it up to AI? And I want to give a shout [00:43:13] out to Joe Constantino on the Tableau side. Joe put together an initiative to bring together Tableau [00:43:23] developers and community developers working on this project. It's really exciting. A lot of like, [00:43:28] testing things out. And we landed on Chat with a Data Source, which is just, again, it feeds into [00:43:33] curiosity. We always want to know a bit more of a data source. And so having another way to do [00:43:37] that with our chat points is really helpful. And now this has become a staple of the Tableau MCP [00:43:42] server. It has. Thank you for doing that for us. And it's so powerful to watch that in action, too. [00:43:49] We are all really excited about AI, agent tech analytics and all things Tableau. What are you most [00:43:55] excited about? Yeah, I've got a lot of excitement for the future. But one thing that's been burned [00:43:59] away for me is vibe coding with Tableau. Yes. Now, I love Korean charts and Tableau. I find a lot of joy in [00:44:06] that. But one thing I don't love is recreating charts in Tableau. Fair. If you've ever had to [00:44:15] migrate from a different tool to Tableau, there's no fun in that. There's no creativity. There's no joy. [00:44:21] So I feel that's a really good opportunity for an AI to come and help accelerate that process. [00:44:26] Yes. Speaking of that, can I show you a few things? Oh, my gosh. Yes. Do we want to see that? [00:44:32] Yes. Please take it away, Will. [00:44:38] Hi, folks. I'm Will. I work as a consultant at the Information Lab. I lead on AI implementations [00:44:43] and enablements for our clients. Recently, I stepped into a new role. And it wasn't what I was expecting it [00:44:50] to be. It's been long hours. I've had people screaming at me day and night. And you should see [00:44:56] some of the crap I have to deal with. That's right. I became a parent. [00:45:07] Oh, boy. Oh, boy. I find myself forever on the go nowadays. There's never enough time in my day. [00:45:15] And so anything that makes me a bit more productive is a big win for me. I've come on at Bolt Bikes. I'm here [00:45:21] as an analyst helping build out their retail inventory dashboard in Tableau desktop. [00:45:30] There we go. There's my viz. It's coming together. I'm really pleased about this. I got this together. [00:45:35] And I've had one more request come in from the CEO, no less. They came over to me. They are on the desk. [00:45:42] And they were really excited. And they started drawing out this viz that they wanted me to build. [00:45:46] I was like, kind of look. Yeah, it was looking like this. I think, oh, that's a double chord chart [00:45:53] right there. That's going to be some calcs. The trigonometry alone is making my head hurt. [00:45:59] And I think, well, I love their enthusiasm. I don't want that to go away. But, you know, [00:46:04] I don't think I have enough time for this. But I know someone who does. My mate, Claude. [00:46:11] So here I am in Claude. I'm going to go and pass over this sketch over to Claude. And I'm going to [00:46:16] ask it now, can you go and make me make this viz come to life? So what will happen here is we'll [00:46:22] start building out a Tableau viz extension with vibe coding. What this will do is a Tableau viz extension [00:46:30] really good for these more complicated vizes that you just want to see how it looks. So this is really [00:46:36] fun way of actually just getting an idea of what that prototype is going to look like without having [00:46:40] to do the heavy lifting. And then yeah, I come back later and here we go. We've got a viz. I've [00:46:46] been dragging and dropping this in Tableau. So the AI here has done a lot of the heavy lifting for me, [00:46:51] and I've now got that viz ready for us. Okay. Because I've saved so much time building this viz out, [00:46:58] let's go and get this published. You know, let's go and see what the stakeholders say about this viz. [00:47:03] So to do that, I'm going to jump back into Claude. I'm actually going to ask it, hey, [00:47:07] can you go and get this embedded on our Tableau portal for our retail partners? [00:47:13] So Claude is going to work away. It's not just -- it's going to go and write the embed code for us, [00:47:18] and also go and get this published on our Tableau site for me as well. [00:47:23] So this is a real nice competitive advantage for our retail partners. Now they have the data [00:47:29] right where they need it, right on the shop floor, and they can go and interact with this data, [00:47:33] answering the questions where they use their data -- where they have their analytics. [00:47:39] Cool. Now we're not just at bulk bikes. We're not just shipping charts here. [00:47:44] We're not just shipping bikes either. We're also shipping the Tableau MCP server too. [00:47:51] So I have a follow-up question from a distributor, and they can use the Tableau MCP server to answer [00:47:58] that question in natural language for them. So here I am. I've connected up to Claude again. [00:48:04] What's going to happen here is we're going to connect up to Tableau MCP. It's going to find the relevant [00:48:09] content for this, and then it's going to start querying that data source to answer that user's query. [00:48:16] This is really powerful for me and the clients I work with. We've seen so much of this being a major [00:48:21] win-win in these cases because this is saving me a lot of time. This would be a request that comes [00:48:27] into my inbox, can now be handled by this. It gives me a lot of time and value back in my day. [00:48:32] But also for our stakeholders, they get immediate answers to their questions. So here we can see a [00:48:38] full product breakdown here. Okay. So this has been conversational analytics, but [00:48:44] the really nice thing about MCP is it's here in Claude. It can be there in Tableau with you as well, [00:48:49] but it doesn't just have to be tethered to a desk, that's all. [00:49:06] Hello? [00:49:08] Hey there, mate. I'm your BI analyst here at Bolt Bikes, ready to help you shred through some data. [00:49:14] What can I dig into for you today? [00:49:17] What am I top selling e-bikes at the moment? [00:49:26] Let me pull that for you properly. Your top seller right now is commuter e-bikes, [00:49:31] with about 19,800 units shipped in the last 30 days, followed by lifestyle at roughly 10,500 units. [00:49:39] Catch you on the next wave! [00:49:44] Thanks, dude. Could you break that down by region for me? [00:49:51] West is absolutely crushing it with over 10,500 commuter units, while east and central are pulling in [00:49:58] solid numbers too. Commuter bikes are dominating every region, but west's also got strong lifestyle [00:50:04] sales at nearly 6,000 units. Catch you on the next wave! [00:50:10] Okay, there you go. The data you want are hands-free on the go, so whether I'm in the office, on the road, [00:50:29] or back at home with the kids, I can have the access, the insights and data I need. This has been really [00:50:35] powerful what we've seen here today. I feel the real thing you've got here is choice. Previously, [00:50:40] you saw a lot of decisions that you had to go and make manually. Now you have an automated alternative [00:50:47] that you can run at scale here. We showed you embedding analytics. That's where we took basically [00:50:52] a sketch. I have a notepad full of ideas that I now can go and pass over to an AI to see what it's [00:50:57] going to look like. Get that into production within minutes. Next, we went back to Tableau and CP, [00:51:04] where you can have that conversational analytics with your AI assistant or on the go. And I'm really [00:51:10] pleased to say that all of the stuff I've shown you is available now and wherever you are on server, [00:51:15] cloud or next for Tableau. Now, if you've enjoyed some of this, I want to give a big shout out to [00:51:29] devs on stage. We've got our hosts here, Sophia and Lauren here. I have had a little preview of what's [00:51:39] coming up, and honestly, it's really amazing, cool stuff. I'm super excited for that, so do not miss [00:51:44] that session. So this has been Powering Decisions. What we're going to do next is go and move over to [00:51:52] MK to show you our identifying actions. See you all, DataFam. [00:52:01] Thank you, Will. Wasn't that amazing? [00:52:05] All right. I'm MK. I'm the president and CTO for engineering and run all engineering and salesforce. [00:52:11] Great to be here to meet you all. Now, one thing that both Michael Andrew and Will was talking about [00:52:19] is how long they've been in the analytics sort of business. And we know the analytics business [00:52:25] itself has been changing from the spreadsheet era, the data warehouse era, the dashboard era, now to the [00:52:33] agentic era. But one thing has been constant through all of this. That has been Tableau. [00:52:41] We've been with you in this journey as the analytical world changed to make sure we are there for you to [00:52:48] guide you through those changes. And this is why over 97% of Fortune 100 companies trust and use Tableau. [00:52:58] But it's not just the big companies. Everyone from the startup, SMBs, enterprise, everyone works [00:53:05] and uses Tableau. Now, in order to show some of this and how these agentic actions are going to come [00:53:11] live, let's actually switch to the demo. John, are you ready? All right. Now, first, this is Tableau next, [00:53:19] what you're seeing there. First, we saw Michael Andrew showcase a lot of the sales demand forecast and so on. [00:53:26] Let's do a quick inclusion of that. We're going to pull in all the sales forecast in there. [00:53:32] Let's upload the file there and you're going to quickly see the sales come up pretty fast. DC 26, [00:53:39] there you go. That's the sales demand right there. And voila, in a minute, you have a dashboard that you're [00:53:46] going to see. Now, interesting thing, what Chad also mentioned was often the sales figures are just [00:53:55] sales. What about the actual warehouse inventory? And now, with the power of Tableau, you can bring [00:54:02] together your sales and your warehouse. So it's no longer just your sales forecast that is showing up, [00:54:10] but your warehouse demand as well. So now you have one operational plane on which you can actually see [00:54:18] everything happening from your inventory to your sales forecast and everything in between. [00:54:23] That is your operational system. But that's not all. [00:54:28] Often sales is a lagging indicator. When somebody comes to your shop and doesn't find that route or [00:54:35] your toy, you're already kind of done, right? They're going to go to a competitor. [00:54:39] But most often, all these demand forecasts are stuck in unstructured documents. It could be like this [00:54:45] analyst report or other kind of things. And these were always separate from analytics. Not anymore. [00:54:52] Because today, with the power of our agent platform in Tableau, you can simply upload these documents. [00:55:00] Or even create a whole pipeline of all of these unstructured documents. And with that, [00:55:06] you'll be able to quickly see, as you can see here, you're going to upload that document, [00:55:09] and now we have one unified lens on which you have your structured data, your unstructured data, [00:55:16] and all of them integrated. And that's how you see that. In fact, that AI is recommending and [00:55:21] actually telling your root cause analysis right there on what happened. You've got a lot of demand [00:55:27] on your west coast, like we saw in San Diego. But you know what? Inventory is sitting someplace else. [00:55:32] And you can see in that fancy graph there as well. But because this is built on all of those [00:55:38] semantic and powerful models, we just don't have to see the visuals. We can actually open up the Tableau [00:55:45] agent right there and actually ask the questions. John's going to type now and actually ask the [00:55:50] question as to, you know what? Show me the forecast. What is going on? And this agent is now looking at [00:55:59] not just that sales data, not just that inventory data, but also all that unstructured data to make [00:56:04] sure this demand forecast is actually right. And so as you can see here, it gave you actually a beautiful [00:56:10] dashboard as well, right there embedded. And of course, you can do it in all the things like [00:56:14] will show it in Cloud or in like ChatGPD, wherever you want. Here, it's embedded right there for you in [00:56:21] that dashboard. But now this is interesting. Now I see that there is a problem. Tableau is no longer [00:56:28] just a passive dashboard anymore. With agents, you can now connect them to all the actions in your [00:56:35] enterprise. So I can actually say, please move that 500 units from Dallas-Fort Worth to the west coast, right there. So you're moving [00:56:43] from a passive dashboard to an active system right there in one pane of glass. And you could say, okay, you know what? I have [00:56:52] hundreds of products. Disney probably has thousands of products. You can't keep checking all which product, which inventory, what's happening all over. Can we automate it? Yes, we can. [00:57:03] Now, you can actually use the same Tableau agents with all the actions between with our agent force. [00:57:10] You can now simply say, go to auto mode. It's going to automatically make all of these actions happen. [00:57:17] And if you go, you can see that all those notifications are popping up right there where the agent is acting on [00:57:23] your behalf and doing it automatically. And soon, if you get sort of fast forward a little bit, John, if you see [00:57:30] what happens, as you can see here, now you see this beautiful network. It's all balanced, [00:57:35] all green, because the agents are working on your behalf and making this move. Remember, these agents [00:57:41] weren't just working on some raw data. We had actually created all the semantic models. You have created [00:57:46] all those semantic models and made sure the agent had the right accurate data. It's not just some white [00:57:51] coded dashboard that's just thinking some random data. This is real accurate data. And also, you see [00:57:57] this beautiful Sankey diagram below that is showing how the distribution is working. These are all new [00:58:01] charts and dashboards that we have now added to Tableau next. All right, let's switch to the slide. [00:58:08] What did you see here? You saw a bunch of things. Let's switch to the slide deck. What you saw is first, [00:58:17] we were able to bring in data, your sales data, your inventory data, exactly like what Chad said, [00:58:23] which was a big gap. Now you can bring all that together, live data. You can also bring your [00:58:27] unstructured data. So it's not just your structured data anymore. So you can actually validate all of your [00:58:32] data. But you can also bring in data to your data. And then we were able to create that unified model on [00:58:37] top of it. And with that unified model and lens, you were able to convert what was a black box logistics [00:58:43] and shipping all siloed into one unified operational platform. And more than that, we were able to then [00:58:50] run agents to go analyze automatically, create these autonomous agents that can work on your behalf. [00:58:58] And finally, we wanted to make sure it's humans and AI working together. So we had all those [00:59:03] awesome Saki diagram at the charts to make sure the AI is actually working right. This is the power [00:59:09] of agentic analytics. It's not just some dashboard on the side, not just some conversational thing, [00:59:14] but everything coming together into that unified plane. So you can move from just passive visuals to [00:59:21] active analytics on the same control plane. All right. With that, Sadat, do you think we [00:59:28] can go even bigger? There's one more thing. All right. Let's see what that is. All right. All right. [00:59:33] All right. One more thing and then we'll bring it home. So [00:59:39] I do want to show you, again, one more small thing we have been cooking up. But before I do that, [00:59:44] I want to do a quick recap. So at the beginning, you saw Michael Andrew demonstrate conversational [00:59:49] analytics on server and cloud and Next. Yes, not just Next. It's in server and cloud. And that's not just [00:59:58] an AI tool. It's actually grounded in a full knowledge graph. You also saw that. And I saw [01:00:02] many of you taking pictures being like, what's that? What's that? When do I get to play with that? [01:00:05] Very soon. Actually, it's coming in July to cloud and then server a little bit in the fall. But yes, [01:00:10] you'll be able to get access to all of that. Now, what else did you see? Not just AI features. [01:00:16] You saw a bunch of other things. This is a very short list. I said, can you put the full list on the [01:00:20] screen? They're like, it's a keynote, Southern. Come on, put all of it on there. [01:00:24] Well, we need a Tableau base for that. That was earlier in the presentation. This is just a short [01:00:28] glimpse. Everything we're doing cloud, server, Next, desktop, and even public. You see a help [01:00:35] agent went to public. I don't know if anyone's seen that yet already. But we're bringing in also [01:00:38] Tableau agent stuff to public as well. So every single one of these areas is getting invested, [01:00:42] meeting your customers, your partners, yourself, where you are. And that's our goal. [01:00:49] Now you heard Mark talk about our knowledge engine, our decision engine. I mentioned at the beginning of [01:00:53] the keynote, all of us are here to help create knowledge, turn that knowledge into an insight, [01:01:00] and turn that insight into an action via decision. That is what we're all really, really good at. [01:01:05] And that's why there's millions of you leveraging Tableau to do that. And our job is to give you the [01:01:09] tools to help you do your job faster, better, and with even more passion. So that's hopefully what we've [01:01:14] been able to impart on you today. But there's one thing. How do you control all those agents? [01:01:21] And we just saw a bunch of agents in the last 20 minutes. How do you control them all? [01:01:25] So we have this idea. It's called a agentic analytics command center. [01:01:30] So I'm going to show you a demo. And I'm going to be really clear. This is a vision demo. None of [01:01:34] this is baked. All right. Literally created in the last couple of weeks. It's our idea. It's our vision. [01:01:40] We want your feedback. Share it. Tell us how you think. Slack me. Slack Mark. Actually, Slack the [01:01:46] engineering team, because they're the ones that actually do all the work. So huge shout out to [01:01:49] Sarish and his team, for everybody else, for all the work they do. They should be getting all the [01:01:56] credit. All right. So let's jump into the demo. So imagine this. You have a lot of agents working on [01:02:02] your behalf. When you log in, what's your homepage? You actually get a homepage that looks like this. [01:02:08] It tells you how your agents are performing. It gives you some to do's. I'll get back to that. [01:02:12] It tells me where work's being done, both in production and development. Often we got a lot [01:02:17] of things kind of in that development stage, that incubation stage, other things that are actually [01:02:22] being utilized. And I said, my health monitor, let's open up that monitor. Let's take a look at that. [01:02:26] So what I see here is a view of all of my agents, but not just a conversational agent. Maybe they are [01:02:32] metric agents, alerting agents. Maybe they're data pipeline agents. All of them I get at. And what we're [01:02:38] giving you is a view and how to establish performance of them. Efficacy. That means, [01:02:43] is it doing the job it was intended to do? Thumbs up, thumbs down, any of those type of things. [01:02:47] Adoption and trust. Are people actually believing what they're saying? Are they having not great [01:02:52] conversations? Are they coming back after they use it? And lastly, perhaps what's important, [01:02:57] data integrity. If the data underneath is not right, then no agent can be trusted. So this is a view. [01:03:03] We're monitoring all of those agents on everything. In fact, you scroll down, you can see I'm red and [01:03:07] yellow. And I know my agents in production. I can see the sessions, how many people are logging in, [01:03:13] the health of all of them. All of that is here. It's a nice, beautiful one point view. [01:03:17] It's a lot for me to digest. Even me. Even Stitch can't digest all of that in one second. [01:03:24] But what if you could just tell me what things I should be taking action on? Because right at the [01:03:28] beginning, I have these little to do's. So this is the system looking behind the scenes and finding [01:03:33] out, hey, are there agents that people are using, some people really engage with that not enough [01:03:37] people are using? Are there agents that aren't performing well? And then it recommends actions [01:03:41] for me to take. My first one is, here at Bolt Bikes, we want to make sure that our carbon footprint [01:03:46] is as low as you can possibly make it. So we have a metric on it. A couple people are using it a lot, [01:03:50] but not everyone in the company knows about it. How do I get out the knowledge? So I click on that. [01:03:55] It gives me a suggestion. It says, hey, you should get these other groups to sustain, [01:03:58] to subscribe to this metric. I can accept the suggestion. And off it goes. It sends a notification [01:04:04] via Slack. And yeah, it'll, it'll work in teams too. Well, sorry. Yeah. But if you want Slack, [01:04:11] it's much better. And it'll send out the notification. I don't have to do anything. My work is done. [01:04:17] And I've just increased the performance of the agent because I've got more people to engage. [01:04:21] My next one is, it says, you have an agent in, in, uh, in testing. It's in my production. It's not [01:04:26] quite in production yet. It tells me my adoption trust is good in a UAT. Data integrity is good, [01:04:31] but efficacy is not great. My thumbs down rate is 15%. That's not good enough to send a production. [01:04:37] I need to make it better. So I can see it against efficacy, trust, integratives. But actually, [01:04:41] let's dive into lineage. Let's take a look at what actually makes up this agent. So what, how do you, [01:04:46] what, what's inside of an agent? Everybody talks about agent. What's inside of an agent? [01:04:49] This is a view of everything in my agent. At the top, I see where the agent's being used. Slack, [01:04:53] Salesforce, Google workspace, Claude. Then in the middle, I see the agent, and I see where it's [01:04:57] being grounded. Those are the semantic models. Those are the context. At the bottom, I actually see [01:05:01] the data that it's pulling from. This is one view of how my agent's working. And it highlights the area [01:05:08] where I have some challenges. So my customer success agent, which is allowing someone to ask questions about [01:05:14] customer success, is struggling in some areas. It's struggling. And it tells me across this knowledge [01:05:19] graph, all the information I'm doing, what do I need help with? So let's jump down and show what's red, [01:05:26] which is down below here. My customer satisfaction topics. This is a semantic model [01:05:30] on customer satisfaction. If I drill in, it gives me some suggestions. Now it's going to come up and tell me, [01:05:35] hey, people are confusing, or the agent is confusing what to use. When you ask for customer health, [01:05:40] should it use MPS or CSAT? Good question. We should tell it. Otherwise, it's going to get inaccurate [01:05:46] answers. So it actually recommends something, and I can actually tell it, hey, this is what you should be [01:05:51] using in that situation. This is a perfect example of when an agent will give different answers, but [01:05:55] you know how it should be answering. Not only can I accept that clarification, I say use CSAT MPS and [01:06:01] label them both so that they know which one am I looking at. I also say, make sure that you're using [01:06:06] that exact same thing across different regions. Don't try to mix them. So I'm going to ask it a [01:06:10] question. So not only can it give me suggestions, I can actually ask a question. And I can ask that [01:06:14] question, ask it to react to that, and it will give me the answers I want. All right, now let's x out of here, and I want to go down and show you one last thing. And that is that, [01:06:22] it's not just this agent that was confusing or getting confused between CSAT and NPS. [01:06:27] I have an issue on my pipeline. I have pulled data from my customer health data source, and it tells [01:06:34] me I've only refreshed this once a week. That's no good. I didn't know that. But it tells me that and [01:06:39] says, hey, do you want to do it every day? So yeah, in fact, I might want to come back and actually do it [01:06:43] in real time. But this is something on behind the scenes, letting me know the issues with how I make [01:06:48] my agent perform better. So that's what we call a command center. So again, this was just [01:06:52] our thoughts. We'd love to get your feedback on it, Slack me and send others at, is how all of us can [01:06:58] make these agents perform better. In the same way that we want to make our dashboards perform better, [01:07:01] our visits perform better, the same concept. So we can go back. And what I want to hope all of you do, [01:07:08] go and explore all the sessions, learn about how you can use AI to accelerate our vision to help people [01:07:14] see, understand and act on data, learn how you can build new visits, learn how you can leverage desktop [01:07:18] all the new ways. Most of all, get to know everybody here. Share your experiences, [01:07:23] make this event really what makes it so great. With that, I'm going to turn it back over to Mark. [01:07:31] All right. How was show? Pretty cool? Awesome. Okay. So I was told to go fast because we're running late. [01:07:45] Okay. We have an amazing TC plan for you. We have devs on stage tomorrow at 9:00 AM. Who's excited for that? [01:07:54] Tomorrow we're bookending the day with iron visit 4:00 PM. Who's excited for that? [01:08:01] And who could possibly be excited for data night out with Dylan Francis? [01:08:07] All right. We have an amazing, amazing week here. I will end it the same way I ended it before, [01:08:12] but a huge thank you. This is going to be a fantastic week. And I only have one more thing. We'll see you [01:08:19] all here next year in San Diego for TC 27. Thank you very much. [01:08:49] Thank you. [01:09:19] Thank you.