Exploring Tableau's New Direction: Deepdive and what to expect in 2025 (Fixed)

0:00Tableau's about to change in a huge way this year.

0:03Tableau have been uh warming us up to the new vision they have for the product.

0:08But at the same time, it's also possible that you will get to the very end of this year and you'll still be using the Tableau you know and love today.

0:15In this video, I want to start what I think is a pretty important journey of understanding what the new Tableau platform is going to look like.

0:22And typically what I would do is I would make this sort of content uh behind the scenes and then process it and then I'd use it for my own benefit to make

0:33a more summarized video for everyone to consume.

0:36So what I'm going to do today is actually go through a presentation that was done towards the end of last year.

0:42It was actually a a sort of a quiet event.

0:44There was quite a lot of people on the call, but it was a quiet event

0:47where Tableau laid out the vision and what it means for analysts who are using Tableau today.

0:52In that presentation, we saw a much better glimpse of what the new workflow is going to look like in the new platform.

0:59And I think this year we've also seen some hints that more clearly set out what the platform actually is and what it means.

1:07And so what I'll try and do throughout this video is I'll give commentary

1:10I'll make observations about the product and additionally I'll also try and sort of lay some context because something you might not know is that

1:18Um, as part of the Tableau Visionary Program, we occasionally get to give product feedback, and that's not something we are typically allowed to share.

1:24Um, we can't talk about the things that we give feedback on.

1:27And it's not a process where we dictate what part of the product we see.

1:31It tends to be uh, you know, project managers or feature developers coming to us with ideas and asking for our opinion and we give direction on what we think could go better and what we think could go well.

1:41And actually, a lot of the places where that feedback has been taken on, the community has seen the benefits.

1:46What I'm going to do in this video is give you a glimpse of what that narrative looks like.

1:51But instead of sharing it with Tableau, I'll be sharing it with you.

1:54So this will be a long video.

1:56Um this is absolutely part of the I guess research and trying to understand the product.

2:00Um and and just to be sort of completely frank and open, uh the product isn't out yet.

2:04Um this is all speculation.

2:06So this is purely speculation that comes from a perspective of having known the product for a long time, having used the product for a long time

2:13and having given feedback on the product for a long time.

2:16And I think that's valuable for many reasons.

2:18And if you're into Tableau as much as I am, as much as most people who watch this channel, then hopefully you find this useful

2:25I've created timestamps so you can follow along.

2:27I think that's a really important part of this narrative.

2:30I don't expect you to watch the whole video, but I do hope that you share certain parts of it with colleagues.

2:35You maybe talk to others about different perspectives.

2:37And I'd love to know what part of the experience today you found really useful.

2:41You'll you'll be able to do that in the comments or you can just ping me on LinkedIn and we can have a chat about it.

2:46In the next week, I'll make a much more summarized video, maybe five to ten minutes, that you can share with your colleagues and is much, much more succinct and it will be based off what I learned in this video.

2:57and sort of the conversations that I have off this.

3:00This is essentially what goes on behind the scenes.

3:02So buckle up, come along for the journey, and um yeah, as ever.

3:06Let's get stuck in.

3:07I want to get right into the product right from the beginning, but as you know with every Tableau showcase like this, Tableau will do some sort of a scene setting.

3:15And I think in this session, Ryan actually opens this the session

3:19And he talks a bit about Tableau, the community.

3:21So I'm not actually going to sort of play through this, so sorry Ryan, but I think it's important to go some of the key highlights that he talks through.

3:28And they're the same highlights that I think we've all seen before.

3:30So if I just um

3:31So to go back we have the forward looking statement which is absolutely fine.

3:35And as every tableau the name event like this, there's always a thank you at the very beginning

3:39And then uh as we go through, there's obviously a homage to the community.

3:43Um, I think Salesforce already understands the community has opinions about the product.

3:47I think this slide is actually useful because if you pay attention over time, you'll notice these numbers are changing and I think they show the continued momentum

3:56I know it's very easy to dunk on the incumbent.

3:59It's very easy for um people with cheaper software to say, oh, I don't understand why Tableau's

4:04It's so expensive.

4:05But at the same time, there's also this community that actually enjoys using the product and gets a lot of value from it.

4:10And I think that's a healthy tension to have.

4:12If you have really passionate users

4:14they personally challenge the product even more.

4:17And that's what we have um in this in this session.

4:19That's what we're doing in this video.

4:21So um whilst the community always comes up, I think it's an important dynamic that keeps Salesforce honest in here

4:27So um let's go ahead and skip forward a little bit.

4:30Um Ryan talks a little bit about the community and then we go to um a little bit of frame setting here.

4:36So

4:37Um Ryan has uh you know ChatGPT Windows very much sets in the context.

4:41I doubt he needed ChatGPT to answer this.

4:44But you know, what is the ultimate goal of business intelligence?

4:47And he's using a modern version of ChatGPT here, ChatGPT 4.

4:500.

4:50Um kind of interesting.

4:51Why can he ask this inside of Tableau Agent?

4:54Anyway, if we kind of skip ahead, he comes up with these values and these values make a ton of sense.

4:59Let me move this little play bar down here.

5:01And um I think this is fine.

5:03You know, Tableau did this in the previous keynote.

5:06They talked about multiple waves of analytics.

5:08Um we were talking about the fourth wave, that was sort of their code name for Tableau before they had a name.

5:13And the name changes unfortunately have been quite frequent, so it's been hard to keep up.

5:18We'll talk about this a little bit later.

5:20But um once he sets the scene here, he then sort of tees up Salesforce's understanding of this.

5:25Uh so if I fast forward a little bit here.

5:27We have the slide that me and Ravi like to call Mind of the Gap.

5:31This slide, this slide with like two cliff faces on the other side.

5:36Um, you know, challenges remain, AI on the right hand side.

5:39This idea of the gap has been in like every tablet conference for the last like four years.

5:45So whenever I see this, um and whenever I'm chatting to Ravy always go Mind the Gap, because in the UK that's the sound they play before a train comes through the platform I'll put some sort of audio on here.

5:55But nonetheless, um we have the slide.

5:57Uh it comes up, it basically portrays the gap between, I guess, uh the reality and the desire.

6:03Um, you know, where we are today in terms of analytics, in terms of teams, how we use data.

6:08And where we want to get to, and actually the promise of AI being that thing that can get us across the other side.

6:13I'm not here to sort of discuss whether that's true or not.

6:15I'm just sort of sort of uh relaying the story.

6:18Okay.

6:18And so we then go to um uh I guess

6:22The proposition, the proposition that Agent Force, this digital labor platform, this is a Salesforce-wide initiative, is this thing to get us there.

6:30Now you have to understand that

6:32Um, Tableau is a product that is part of the Salesforce ecosystem.

6:36So whenever Salesforce has a big push for the year,

6:40Um that becomes the push for all products.

6:44All products have to sort of find their home underneath that umbrella.

6:48And this year it's Agent Force.

6:50And so as a as a result of that, um we have everything framed in the context of Agent Force.

6:55Agent Force sits behind this idea of agendic AI.

6:59I recently did a video breaking down sort of Tableau's journey to AI up until I guess the beginning of 2025.

7:05Um go ahead and check that out because I I really break down some of I think the weaknesses of that perspective, but also some possible strengths and where I think that could go in the future.

7:14But in essence, Salesforce frame everything under that banner.

7:18So we have these sort of core tenants, trusted, easy to deploy, humans in the loop, zero whole time, integrated and open.

7:25Um and they really kind of uh want to set this sort of um vision of Tableau being at the heart of this ecosystem.

7:32And so that's essentially what this slide is.

7:34Whether you're a customer or data analyst, wherever you are in this sort of data cloud or this ecosystem, the Tableau playing an important role in enabling you to achieve your goals.

7:44Now

7:45The thing about this is it's actually probably the first diagram.

7:49You might not sort of be familiar with this diagram, but it's quite a common sort of slide we see in the Salesforce ecosystem.

7:55It's actually a platform diagram.

7:56It's just not laid out in a very easy to understand way.

7:59So across the bottom, you've got sort of the fundamental and core parts of the Salesforce platform, the work through all of the Salesforce platforms.

8:06You have Data Cloud.

8:08Data Cloud is this concept within Salesforce where essentially you're able to hook into your datasets in situ.

8:14You don't have to move them around, whether it's Databricks or Snowflake.

8:17Everything works in a sort of a zero copy architecture here.

8:21That's a data cloud system.

8:22You're able to build out those uh semantic models um behind the scenes.

8:27And you know, according to Tableau and Salesforce, the semantic model is now available.

8:31We'll come to a slide about that in a second.

8:33And then you've got Tableau that wraps around that, this part of the customer 360, essentially the analytical part of that customer 360.

8:40And then you have Agent Force, which is essentially a bunch of agents that sit around this platform.

8:47And I think an important thing to understand is that Agenda KR can sort of work in in two in two core uh

8:55frames.

8:56You can have the kind of agentic AI where the AI sort of goes from the very beginning to the end of the experience.

9:03And you can have this other more granular agentic AI, which is uh some parts of the experience are delivered by AI.

9:10And I think where we are today is the latter one.

9:13You can see here there are uh this sort of narrative of lots of different types of agents which are customized and good at doing specific things

9:20You have a support call agent, you have a chat bot agent, you have an analytics agent, all these agents doing small parts as part of the bigger ecosystem.

9:29And so I think that's important to call out.

9:31So if we if we carry on here, what we then get is a little bit more of a a framing of when to expect.

9:38And I think this is actually a pivotal slide because

9:41For me it's the first time I start to see when I'm gonna be able to play around with a new tableau.

9:47And it also sets the um

9:50the context of what we we we should expect this year.

9:53So the first thing to understand here is you've kind of got like three pillars.

9:58You've got the left hand side, the middle bit, the marketplace, and then the agent force on the right hand side

10:03Now, if I start on the left, we've got action, which is general availability.

10:07GA stands for general availability.

10:09That's June 25, that's uh right after conference.

10:12You've got the visualization right now.

10:15Tableau Pulse is a little bit of a part of that.

10:17And what they mean by this is that the visualization pilot is actually taking place.

10:21A few customers already have access to the agent force element of this where they're able to

10:27Build visualizations and charts as part of this new designer tableau.

10:30And we're doing it using AI and assistance throughout the whole sort of process.

10:34Again, we'll come back to that what that workflow looks like

10:37Following that you've got Tableau Semantics.

10:39This is already generally available in February 25.

10:43To be brutally honest with you, I don't know what this is.

10:45So I'm gonna do my best to find out what that is in the coming weeks

10:50But it's available apparently, so um we can go ahead and start to um maybe discuss that a little bit more soon.

10:57Um it's not part of the 25.

10:581 release, so I think

11:00More clarity and that would be great.

11:01And then you got the data layer, which is essentially what I had talked about before, which was um the data cloud essentially.

11:06The the the

11:08the zero copy architecture and the ability to build semantic models within the data cloud that you can then call upon and you know in in one click create Tableau data sources from.

11:17That's that's kind of available now.

11:18It's not a workflow that many people use because you have to do it through the Salesforce ecosystem, but it's strictly speaking, it's available and you can start to use it.

11:25You've got the marketplace in the middle.

11:27Um that's coming October 25th, so much, much more towards the end of the year

11:31And so everything on the left is happening very soon, marketplace happening towards the end of the year, and you've got agent force on the right hand side without any date.

11:39We have to assume that's also coming this year.

11:41And on the right, we've got

11:43potentially these components of taking actions with pre-built skills.

11:48So the four skills are data pre, prepare and visualize.

11:51A concierge, enable trusted data question and answer, um inspector, deliver proactive insights, so the ability to look into your data and sort of

12:00um give you insights uh ahead of time.

12:03Analytical apps uh build solutions for action.

12:06So these are essentially dashboards in the new framing, but they're calling them apps here in this specific uh context

12:12Then you've got the ability to enrich agent force, so you've got the sales cloud agents, the service cloud agents, and then other cloud agents from the Salesforce ecosystem feeding into this ecosystem.

12:22Okay.

12:22This is sort of a framing of

12:24um where and when all of these things are are sort of going to to come in.

12:29So if we go ahead, yeah, this is sort of where we have Matt Miller.

12:33Now Matt Miller is senior product management at Tableau

12:36Um he typically has been doing the demos for this new version of Tableau.

12:39So anytime you see Matt, you can guarantee he's uh you know pitching the new thing, the new platform.

12:45Um the new vision for Tablet.

12:46And so for the next uh 36 minutes, uh he basically goes through a demo.

12:51And this is essentially what I want to play through.

12:53So I will play through it and I'll try and keep the volume low so that you can hear me over what Matt is saying, but I'll also try and pause it as we go through this so we can.

13:02We can have a bit of time just to settle and understand sort of what we're looking at.

13:06And I will pause at various moments just to sort of uh break it down in a little bit more detail because again I have some context, but I'm going to share only the context that I know is public.

13:14I have to sort of

13:15toe a to toe a party line and and as I've said before to be just completely frank and transparent.

13:20This isn't the first time I'm seeing this.

13:23Um it's it's by no mean um, you know, the last time I'll see it as well, if that makes sense.

13:28So I have to be

13:29uh sort of very open and frank to say look I've seen some of this well before it was announced today as part of you know my role in the community but then also just um I think that will help

13:40inform the way that I break this down in a way that Tableau I think would agree is honest with the uh with the product that they're sharing.

13:46So let's um let's hit play and see where we are.

13:50Transformation that's happening recently, it lets us make a huge leap forward.

13:54And so everyone can see, understand, and act on data easier, better, and faster.

14:01So what are these

14:02new skills that we are adding in Tableau.

14:05Well, first of all, I have to say for me as a let's say kind of old school guy who learned how to program on a on a computer that plugged into a TV.

14:14It was a mindset shift for me.

14:16And so because I went through this, maybe I'll share my journey a little bit to help explain how I think about these things.

14:22Like in the old way, we when we added a feature to Tableau, we had to code it explicitly with like if.

14:29x is greater than this than else.

14:32In other words, the software didn't do anything for us until we told it to, and then it didn't do anything until we explicitly spelled out what we want.

14:40But this new technology is so broadly capable that we can leverage it simply by sometimes explaining, just as you talk to a colleague, what it is that you want to achieve and apply your business knowledge, and it can do the rest.

14:53So I'll give you an example.

14:55Uh my teenagers are responsible for cleaning the kitchen in our home.

14:59And and so they say they've done the dishes, I go and check the kitchen and like I've got a standard.

15:03It's like there's no dishes left in the sink and the knives are

15:06cleaned and put away and the counters have been wiped off.

15:09And so we have a system where if they do their chores enough times in a week, they get their allowance.

15:13And so yes, I visualize this in Tableau.

15:16Now, this may s make me sound like a great parent or a terrible parent.

15:19I don't know.

15:20I'm just trying to see, understand, and act on parenting.

15:22But recently I decided I want to play with this technology myself.

15:26And so I built them an agent.

15:28I trained an agent to help them with this.

15:30I kid you not.

15:31So I explained to it what my business definition of a clean kitchen was.

15:35There's no dishes in the sink, the knives are clean and put away, and the counters are clean.

15:40So now my kids can actually converse with this agent.

15:43They can ask it questions like, am I done?

15:45And they can actually show it a photo.

15:47And it looks at it and goes,

15:49I see a bunch of dishes in the sink.

15:50I think you're not done.

15:51But more than that, they can ask it interesting questions I would have never anticipated.

15:55Like, I want to go watch a movie with my friends in an hour.

15:58Do you think I'll have this done in time?

16:00And it might say

16:01You sure can, but you better roll up your sleeves and get scrubbing those dishes.

16:05Now, this is just a completely different way of thinking about capability about So that that story.

16:13Um

16:14You might think, oh, come on, like, what is this story really about?

16:18Um and if I sort of break the break it down

16:22I I think it's it's a genuine story, but it also serves a a purpose, a contextualization for this session.

16:30The reason we're hearing this at the beginning.

16:32Is because it's very much a framing of what this platform is going to be about.

16:36So what Matt is essentially saying, look, if you're using Tableau today,

16:41Um you probably will argue that there's a high barrier of entry in terms of understanding certain functionalities, especially the advanced things like LODs.

16:49writing window calcs.

16:50And yes, of course, there's lots of people on YouTube and blogs, the Ken's, the Andy's, the you know, Donabell Santos's.

16:57Um yes, we all sort of can teach you how to do that fairly well.

17:01But the reality is that not everyone actually discovers those people in time.

17:05And a vast majority of people are asked to use Tableau and just have to get on with it.

17:09And it's very powerful, but it's hard to discover that power.

17:12What he's essentially talking about here is

17:14a change, a mindset change essentially, where um this product is going to have a different paradigm built into it.

17:21So if you um if you sort of take that into account

17:26you you can sort of approach bigger, more challenging problems.

17:29And if you don't sort of have that mindset shift right at the beginning, then I think um you won't get as much value out of the platform.

17:36And fundamentally

17:37you won't realize the value of this platform.

17:40So this both serves as a story, a pretty good one, but it also serves as a narrative sort of

17:46tool to get you in the mindset to understand that look this platform is only going to be valuable if you buy into the philosophy that we've built into it from the ground up.

17:55And I think that's going to be an important thing that we kind of um

17:58uh experience throughout this this the rest of this talk so let's keep going.

18:02Technology solving real world problems.

18:04That's what Agent Force is doing.

18:05So I want to double click with you into these new skills that we're adding.

18:08That's why we don't call them features because

18:10it's a completely different way to think about the capabilities we're adding.

18:14And those are first the data pro for the work around analytics.

18:18Preparing and visualizing data.

18:19There's no great insights without great clean trusted data.

18:23And those are the biggest parts often of analytics.

18:26The concierge, think of this like your personal assistant, like the smartest analyst you know that you can ask and get trusted responses

18:34Not made up, but really backed by the real uh the truth of your data and your business understanding.

18:40The inspector, like Ryan said, it's the one sort of working in the background, finding insights proactively.

18:45Completely different paradigm to analytics than just.

18:48It doesn't do anything until I ask it a question.

18:50Finds insights even while you're sleeping.

18:52And then the last piece is how you scale that with analytical apps.

18:56So I get to show you all this and we're gonna jump into that first.

18:59But first I wanted to show you I thought that was a good good framing.

19:02Um

19:03These four skills I think break down um I'd say four pillars of how the new platform works.

19:10The Tableau of today

19:12Um it breaks it down into products.

19:14So if you want to go do data prep, go to Tableau Prep.

19:17If you want to go do visualization, go to desktop or tableau public.

19:20If you want to go share your work and govern it and make it secure.

19:24go to Tableau Cloud or Tableau Online.

19:26It's a very product focused sort of paradigm.

19:31This paradigm of Tableau blows away with products.

19:34and focuses on experience.

19:36This this is sort of my tagline for this whole new version of Tableau.

19:40Get rid of the idea of products, focus on experiences.

19:44Experiences being the fundamental sort of

19:48thing that you that you go through to access a capability.

19:52So features will sit behind experiences.

19:55Experiences will be what you get to know.

19:57So data pros and experience.

19:59The concierge is an experience, inspector is an experience, analytical apps are an experience, and agents work within those experiences to bring through capabilities.

20:09And the capabilities will remain the same.

20:11But you just won't be hopping between this product and that product and that product.

20:15We'll just be happening in one window.

20:17Okay.

20:18So uh let's carry on.

20:20The agent force piece of this.

20:24How do you actually leverage that?

20:25And I'll do that.

20:26So what he's done here is highlighted the two things.

20:29So the agent force piece of this is Data Pro and Concierge.

20:33So there's

20:34It's kind of not very clear on the slide, but I think it's subtle and it's obvious um how they're doing this.

20:39And I think throughout the session, um

20:43When we get into the screen share, I'll leave the slide up in the top right hand corner so you can see when we mode shift between these different experiences.

20:51So you've got the new tableau.

20:53Tableau cloud, the Tableau you know today, and Agent Force, if that makes sense.

20:57Tableau server can be replaced with Tableau Cloud as well.

21:00Although to be honest with you, if you're on Tableau Server, you're not getting any of the AI stuff at the moment.

21:04And that's a big weakness.

21:05I covered it in my AI video.

21:06But nonetheless.

21:07And that's sort of how this is working here.

21:10By sharing my screen.

21:13So let's imagine now that I am the brand uh manager for a company called DataFam Kicks.

21:18And those of you who know me know I love my shoes.

21:21So of course we had to use a shoe example.

21:23And I'm trying to hire street teams to get the word out about my amazing DataFan kicks.

21:27And I've got all these people who are applying.

21:29And I want to apply data and my business rules to be able to manage all those and make sure I hire the right street team to promote.

21:36So I've got this

21:37application from Simone.

21:39And so that's gone into a data set that's in a spreadsheet of all my job postings and their feedback.

21:45Now I want to sift through these and find

21:47the best candidates for these roles.

21:49So how can I take the power of data and trust and pair that with this agentic capability?

21:55To solve my business problem, which is finding just the right people.

21:58So I'm going to walk you through how we would do that, creating that agent that you can then leverage with data.

22:04Okay, so um immediately uh we have dropped into what I I think I would describe as a Southwest experience.

22:14So we've got um the kind of standard side menu on the left.

22:18Um we've got a bunch of different agents um and it's very clearly contextualizing this whole experience inside of the Salesforce ecosystem.

22:26So um

22:28Whilst this is not strictly Tableau, this is Salesforce, but as I've said before, Tableau is a product in the Salesforce ecosystem.

22:37So, um we'll start off in Salesforce, DataFam Kicks is the company, and as we navigate through this, I'll try and highlight what I think is Tableau and what I think is

22:48Salesforce.

22:49I think it'll be a little bit tricky there and I might get some of it wrong.

22:51If you work at Tableau, let me know.

22:53In the comments below, that's your space to correct the record.

22:57So I'm going to begin here in the agent force builder.

23:00I'm going to say create a new agent.

23:02And I'm going to start from scratch.

23:04And the thing is So this to me is

23:06In Salesforce.

23:07This is not in Tableau.

23:09So we're building an agent inside of Salesforce ecosystem and the agent force setup.

23:13This is the agent builder.

23:15And it's going to feed into the analytics story inside of Tableau.

23:19So let's keep going with this.

23:20This is so awesome because I learned to write C<unk> for DOS.

23:24That's how old I am.

23:25It is incredible.

23:26You can just say

23:27You're a recruiter and explain to the agent this in whole language, this is what you do.

23:33You review job requisitions, you analyze profiles, explain fit, and you make recommendations.

23:38When we do that, it goes and finds topics.

23:40And topics are areas of capability that you're applying to this agent.

23:44And so this is where to leverage some of those things that you saw in the previous slide, like data pro and concierge as well.

23:50So it's recommended both Data Pro and Concierge for this.

23:54Right.

23:55So you've kicked off an experience in Agent Force, um, and it's recommended two of the skills

24:02that relates to Tableau.

24:04Okay.

24:05And you can actually see they've both got this sort of Tableau icon just here.

24:10And um

24:11It's also in context of other agents.

24:13So you'll see if I go over here, you've got candidate outreach, which has an email, a Slack, and an SMS component.

24:19And we've got appointment scheduling, which has an email and a Slack scheduling component.

24:23That's the service agent

24:24got lead development age and that's coming from sort of the pipeline gen element of tab of Salesforce.

24:30But fundamentally here you've got um

24:32Two channels um and and Slack as well and six actions.

24:37So the Asian force sort of ecosystem has

24:41um skills the skills can be used by an agent um and your agent can employ a series of these skills to achieve actions

24:49throughout the Salesforce ecosystem.

24:51That's sort of the best framing and understanding that I can give you today.

24:55I hope it's right.

24:56I hope it's um sort of true to the to to to the to the to the product.

25:00And anyway let's keep uh let's keep going

25:03Makes sense.

25:03There's some chewing of the data and processing the data that has to happen and also the ability to answer questions about it.

25:09And I can tune exactly which of the actions I want

25:13this agent to be able to do from right, right.

25:17So we talked about um skills.

25:19Um the actions I think are slightly different.

25:22So that I like

25:24It's not clear to me off the top of my head, um, but the the way the way some of this agents work is they have to be given very narrow confines and say what

25:36What these actions suggest is that they are specific things.

25:40So generate a visualization is a type of action.

25:43Build dashboards, a type of action, analyze dashboards.

25:46type of action.

25:46Analyze metric.

25:47Use semantic data model.

25:49Send Slack notification.

25:51That's a different type of action inside of a different sort of ecosystem.

25:55And so you can tick some of these uh actions um and so

26:03Actions sit within a scale.

26:04I'm I'm already struggling to explain this, but that there is a little bit of a hierarchy that's developing here.

26:09Let me sort of finish playing off this section and then maybe it'll become a little bit more clear.

26:12Yeah.

26:12Um building dashboards, generating visualizate visualizations.

26:16My goal is to expose this in a way that my colleagues can communicate with it through Slack.

26:21And so probably using the semantic data model and generating visualizations are the actions that make the most sense.

26:27And it has taken my really minimal prompt and turned that into these instructions, which I can then tune and refine.

26:33So I'll say next because those look good.

26:35That's what I want.

26:35And I now define.

26:37how we're going to interact with this agent.

26:39And I'm going to do that through Slack, but that could be through any other channels.

26:43I think this interaction thing is actually quite powerful because if you're a Tableau user

26:47Actually getting some of these interactions into your analytics workflow has actually been quite difficult.

26:53So I think this this actually does present a quite of a huge opportunity.

26:57Um, you know, I know not everyone uses Slack, but I know they announced a relationship with Microsoft Teams.

27:02But specifically, APIs are going to be a huge point of call here.

27:06Uh things like SMS, Apple Messages, WhatsApp, Messenger, these are all pretty important.

27:11Even emails, getting email workflows to sort of work through here is also going to be quite useful.

27:16That's all said.

27:17So what I've done now is I've created this agent, I've given it the topics and the actions that I want it to do, and I can then make sure that it has the data.

27:26This is the power of Tableau.

27:27These are Tableau data sources on all that information, but they can also be other things like my document about what defines a clean kitchen that I wrote up in a Google Doc and attached to the agent.

27:37You can attach the qualitative stuff that defines here's the kind of profile we're looking for, here's the kind of culture we're trying to build.

27:44So I think this is where we start to see the the vision of Agent Force.

27:48being able to bring data sources into the ecosystem just natively, but also then being able to mix the data along with the

27:58um you know business context that you might have living in Salesforce.

28:02And a lot of companies have their pipelines in Salesforce.

28:06I've seen insurance companies, I've seen investment companies.

28:09I've seen obviously sales companies.

28:11Um they have a lot of what I would call um the way they work inside of sales, even seen fundraising work inside of Salesforce.

28:18So um all of that context

28:21alongside the data that supports it is actually quite useful.

28:25And I think the idea here is that AI can be the the the sort of the magic uh between the different bits.

28:31So let's um let's see how this goes

28:33Inclusivity principles, for example.

28:35You sure you got the right data, review it and click create agent.

28:39Now it's time to validate

28:40that this does what I want so I can ask it a question here in this test environment.

28:44Let's just ask it a simple uh descriptive question here once I uh

28:48Click in and the simple question I want to ask is how many active candidates do I have?

28:54So this is just a seems like a simple piece of math.

28:56Like this should be easy to do.

28:58But actually, there's a little bit of nuance here.

29:00So how many active candidates is actually an answer that differs.

29:04I was talking with a colleague yesterday who worked in HR Analytics.

29:07How do you define what an active candidate is?

29:09It actually differs from business to business.

29:12Depending on whether you're hiring for a flex workforce or you're hiring for uh W-2 or full-time employment, maybe your definition of an active candidate is one

29:21who has already gone through a phone screen and a panel interview and that's an active candidate.

29:26Or maybe your definition of an active candidate if they've done a phone screen and you've been in touch with them in the last seven days.

29:32Whatever your business definition is, that gets applied here so that when we say active candidate and somebody else sees an answer for active candidate, we're talking about the same thing.

29:41We're singing from the same sheet of music.

29:43And we know this is trusted.

29:45because it's backed by certified data in Tableau.

29:48So this looks okay, so that's quite uh sort of important context.

29:51So we've asked the question, the agent's gone and found the data source inside of Tableau and

29:57What we get here is uh think of this like a pulse card.

30:00Um Tableau Pulse has these sort of metrics.

30:03It's very similar to that.

30:04Um it maybe isn't that, but it's it it could easily be a pulse metric.

30:09with an additional bit that we don't have today, which is a description, a semantic description of that specific metric.

30:18So active candidates predictor to reach 55 at the end of the current month

30:22More details in recruitment dashboard.

30:24The semantic description is what we get up here.

30:26The brand ambassador has been open for two months.

30:28Since then you've received 68 applications with 57 active candidates.

30:32Notice it didn't open a new window with a chart.

30:34That's sort of the traditional tableau.

30:36You ask a question, boom, you get a chart out.

30:38That's pretty much what's happened inside of Slack.

30:40This is what has historically happened throughout the tableau apart from it builds a chart inside of the uh the view.

30:46Um instead we're staying in the conversational context and we're still only just testing the agent in this particular case.

30:52Fantastic.

30:53I want to activate this agent.

30:54So now how do we use this practically and deploy this?

30:57Well, we set this thing up as a uh Slack-centric

31:01Agent, so let's go take a look at how that looks.

31:03So now I'm in Slack and I've got a channel where we're working on recruiting these brand ambassadors.

31:07And I could simply just ask this agent, what are the top three candidates for this brand ambassador role?

31:14Send that message and right there in the chat in front of everybody else, agent responds immediately with all the information.

31:19It's a data-driven recommendation.

31:2192% match sounds good, but I want to know what's behind that.

31:26So that's interesting because um obviously in the um agent building experience we saw the pass card kind of thing.

31:33Here we see a a sort of native experience inside of Slack.

31:36Where it's um I think digging into the data source and bringing in more context through that data source that we have inside of Tableau.

31:43So

31:44It's essentially pulling out um this contact.

31:46And I'm assuming these percentages would have been pre-computed in the data source.

31:50So it's not like it's generating them on the fly.

31:52AI isn't doing some sort of

31:53magic here, but it's essentially just pulling them through, but the agent has made it easy for someone in Slack who's maybe not familiar with the Tableau data source

32:01to just go and get an answer without having to go build a dashboard uh and bring that context immediately here.

32:06Now I'm up for this because uh frankly a lot of business users don't care for dashboards.

32:12They don't care

32:13for having to go into the platform and extract the answer and build a nice shop.

32:17They don't even care for opening a dashboard to get the answer right there for themselves.

32:21They might not even care enough to even open up Tableau Pulse

32:25and go and get it.

32:26Instead, just having it immediately go to them in the contacts is great.

32:30Now not everyone works in Slack.

32:31I totally get that.

32:32Microsoft Teams is probably where a lot of people

32:34have to work in that business context.

32:36Even in companies of Microsoft Teams, honestly people just don't know how to use Microsoft Teams.

32:40It's kind of like a a a bugbear of mind.

32:42But nonetheless, um wherever people are working, whether it's Microsoft Office or Microsoft Word or whatever it is

32:48If they're able to ask and get the question in context, then I think this is super powerful.

32:52So let's see where this goes.

32:55I and my colleagues might say, okay, now why do you say 92%?

32:58Why does the agent say that?

32:59We gotta be able to trust and verify and understand what's behind it.

33:02So we can not only just get that answer, but ask why is Simone the best candidate?

33:07Please reply in that thread.

33:09And here you see the agent goes and says, all right, I'll drill to the next level and I'll tell you

33:13uh how I did it, show my work, so to speak.

33:16So here we get that qualitative and strategic side.

33:19This is capability that just was, like I've said, science fiction two years ago.

33:23It's incredible.

33:24So I get all the qualitative points pulled from CV and pulled from our our profiling and our our configuration, but I can also go and give me the breakdown of the quantitative side.

33:35Let me understand how we got to that score of 92%.

33:37And our agent goes and queries Tableau and checks our trusted data, gives us a breakdown, and of course builds us a great visualization as well.

33:45So that's just a glimpse of how you configure an agent, how use it to leverage Tableau.

33:50In the next sections, we'll go into each of those one by one.

33:54So that's that to me is a very nice canned user flare.

33:59Yes, it's grounded in Slack, but you could easily see how that experience can be grounded in some other m communication service within a company, whether it's

34:08Microsoft Teams, you could even see how potentially this could work inside of like a team-centric version of WhatsApp or whatever it is

34:16Um to to be honest, I don't know where people do their conversations.

34:20I I please don't say emails because that's not a conversation platform.

34:23But nonetheless, you you get the context.

34:25It's um I think it's quite easy to set up.

34:27So I think this has set up the flow.

34:31You build an agent, it gets access to your Tableau data sources.

34:34Those Tableau data sources are trusted and verified, hopefully by a data engineer who knows how to build those data sets.

34:41Once those are trusted and verified, the AI is able to peruse that data source with some guardrails about what it wants.

34:48You obviously gotta ask good questions.

34:50You've then got to understand what a good follow-up question is.

34:56But you get that sort of interaction here in Slack being able to r reply specifically in a thread and it pulls off a side piece and you get this sort of

35:03Candidate match score 92% for this one person again.

35:07Looks a bit like Tableau Pulse with a bit of semantic description, an actual chart

35:11And it's the same sort of let's say pattern that we've seen in Tableau Pulse.

35:15So really easy to follow.

35:16And if this is how the rest of this demo goes, then great.

35:20But there's of course a ton more about the Tableau ecosystem that's not in here.

35:24So let's find out how that all works.

35:26Okay, so what we're gonna do in the next section is we're gonna look at that data pro skill specifically.

35:31How does that show up and help me in the work that I'm doing?

35:35Okay, so immediately I'm just gonna stop here and talk about this.

35:39Because um this is the Viz Building Experience, they call it the Data Pro experience.

35:48Um

35:50Inside of this new ecosystem.

35:52So when you go to build something yourself, you'll be building it in this experience.

35:58Now

35:58It's a much, much lighter experience.

36:00Um and this could be one of two things.

36:02This is still an early development version, so it hasn't got all the sort of density of all the features that you'd expect.

36:08And I think that's going to be deliberate.

36:10I think this this experience, as m as Matt has said before, isn't designed to be this sort of um, you know.

36:17um Swiss Army knife to every single problem.

36:20Instead, it's a much more refined vision of right, people are just coming in here to build quick solutions, build answers.

36:26Let's take away all that sort of complexity in how people build stuff.

36:30Let's bring in some interactive elements like images and make it easier to do certain chart types.

36:36Let's just design nice charts on behalf of everyone so they don't have to do all that work themselves

36:40But let's still use columns and rows and keep everything familiar.

36:44So you've still got the marks pane.

36:46You've still got the columns and rows up here.

36:48You've got filters, pages, marks.

36:51There's no parameters area here because maybe we haven't created one and that's maybe not contextual to this specific view.

36:57But at the very top you see this new concept of something called a workspace and I think Matt will go into that in a minute

37:02But you can uh see that the workspace sets the context.

37:08So you've got mentions by channel, and you've got a few sort of an action bar here at the top.

37:12And my hunch is that probably changes based on the context of what you're building.

37:16You've got metrics, calculated field, and cloud objects as well.

37:20So cloud objects are essentially tables in the data cloud.

37:23Calculated fields are probably imputed from those objects.

37:27And you've got metrics which I think uh basically build on what we've seen in Tableau Pulse predefined by semantic models inside of the uh ecosystem.

37:37Then on the right hand side you've got sort of the insights pane and this is sort of that sort of communicative element.

37:42So this is the data pro um experience.

37:44This is in the new tableau.

37:46The slide has sort of gotten cut off, so I should have done a better job of that.

37:49Um but this is sort of that whole new um experience.

37:52So let's see how it works and then maybe we'll we'll we'll talk a bit more about it after that

37:56So I'm here in the new Tableau and it's an experience that many of us are familiar with.

38:00It should look quite familiar if you've used Tableau before.

38:02We've brought forward the brilliance of our industry leading

38:05exploratory experience and we've taken it to the next level with this agentic capability.

38:10So I'm having a col conversation now with one of my colleagues and we're trying to understand our street team performance as compared to and see if it's really influenced by

38:19social media traffic.

38:20And to do that, because we were having a conversation about that in Slack, Tableau has picked up on the context.

38:26This is the data pro going, hey, I see what you're trying to achieve.

38:30You're going to need some additional data that you don't have.

38:32Go find that data and I can help you answer that question.

38:35So it's so that T off is actually really important.

38:38So we've gone from Slack and it's gone it's gone right into the editing experience.

38:43And it's it's brought in different context from the different sort of elements of the platform.

38:47So obviously the data cloud is already there, got the metrics, calculated fields, and it's sort of set up a context immediately based off that conversation

38:56So that ability to just set that up for us, I think, is also AI AI driven.

39:00And it's actually quite powerful.

39:02Sometimes a starting point for analysis is actually a really hard thing to understand.

39:06You need some additional data that you don't have, go find that data and I can help you answer that question.

39:11So it's Okay, we're back.

39:13We're gonna carry on through this experience and see how we basically connect to a new data source

39:19guiding us through the process of doing that analysis.

39:22And we can connect to any enterprise data source or just use ad hoc data.

39:26Something like one-third of all connections to Tableau today are still two spreadsheets.

39:30We know a lot of the world's data lives there, so you've got that ad hoc as well as the other thing.

39:35So third of all data sets are still two Excel spreadsheets.

39:38It just shows we've got all these data engineers out here building semantic models and people are still exporting them and putting them back into analytics tools.

39:46It's just it's just the way things work.

39:48So

39:48I think it's um it's it's good of Tableau to think of that workflow rather than trying to block people because there are there are some tools out there that don't support this workflow natively.

39:57So it's really good that, you know.

39:59Tableau's not forgetting what most of its customers are doing.

40:05So we'll pull in that spreadsheet, and the first thing we see is that it's a mess as many spreadsheets are.

40:10Now one really subtle thing, um up here it created a new tab.

40:14So something that I've noticed with a lot of these um demos is that actually

40:18And as you're going through different parts of the workflow, it's creating tabs for each experience that you're going through.

40:26So here we're in the data connection window as you would be in Tableau.

40:29And so this has its own tab.

40:30And what we're essentially about to see here is essentially the agent supporting Matt in cleaning up this data set, essentially.

40:37So let's have a look and see how it does that.

40:39It's obviously got like

40:41uh different sheets and i think it's identified different data sources within each sheet so it's it's able to do the thing that tableau can already do which is pull out uh formatted tables

40:50from an Excel file.

40:52And you've got the data interpreter feature up here on the top left hand side.

40:55So that's that's still quite useful.

40:57And this allows you to just you know be familiar that this is

41:00This is the same experience we've seen in Tableau, but it's just been redesigned here for the web and it's a little bit more native.

41:08There's missing columns and headers, things are shifted.

41:10But that's okay because this data pro agent is partnering with us.

41:15It knows what we're trying to achieve and it recognizes that this data could answer our questions with a little bit of cleanup.

41:20So this is really interesting because I don't think we have this capability in Tableau Agent today.

41:25I think you could type a question, you could ask it, but I haven't seen it suggest something similar like this in Tableau Prep.

41:30Um, we've seen the uh agent write calculations, but what this is suggesting is look, it's it's got a couple of suggestions, it's actually going to go and execute those suggestions for you.

41:40Which is, I think, one step beyond where we live today.

41:43Um so if you know if this was um part of the world today, I would agree that this is like an agent.

41:49It's helping you kind of

41:50uh do a bunch of things but it's not quite there yet so um yeah hopefully hopefully we see a more clearer path between where we are today and how we get to this.

42:00It's one tap to accept that question.

42:02Just to be clear, this isn't a wizard at all.

42:05This is an agent.

42:06We've given it the outcome we want to achieve, and it's figuring out what the steps are.

42:10to get us there.

42:11We've cleaned the data up and it looks great.

42:13We can add it to the model we were working with.

42:15And that model that we were working Okay and so uh we've cleaned the data, we've prepped it, we've added to the model and then it's opened up yet another experience.

42:25So the first one was like an agent builder, the second one was a data source.

42:30You've got the little um spreadsheet here at the top you can see here.

42:33And then the third part, when we select add to model,

42:36We're adding it to a semantic model that sits within the marketing team.

42:40So I like that once you've essentially um uploaded a flat file of some sort, you're able to add it to the business context.

42:48And the thing is

42:50The value of doing it this way is that as a data engineer you get to hopefully see the metadata and see what people are connecting to your data source

42:57And you can figure out how to bring that data in and build it into the set and then go back to that person and say, hey, you don't need to do that.

43:03What I'm hoping is they have an AI that does that.

43:05But um this is really quite nice because this is also the first time we're seeing a very deliberate placement of the semantic model.

43:13Now it's worth calling out, it's very easy to miss.

43:15On the left hand side we've got metric definitions.

43:18Now this would

43:18typically sit in the semantic model.

43:21You've got parameters which are sitting here, calculated field, logical views, data objects.

43:26So it's the same uh you know chart window we saw earlier, the same left hand side pane we saw earlier on.

43:32and but in this context we're looking explicitly at the semantic model.

43:37So what this sort of suggests is that even though we've had two different experiences, the side pane is consistent through different parts of it.

43:45So whether you're building a chart,

43:47whether you're looking at semantic model or connecting to a data source, the context upon which you're working with that data follows you through those experiences and it's a consistent experience.

43:57And so that's actually really powerful and it's a really new capability.

44:01I think of this platform.

44:02So here we have um the marketing team.

44:05We're in the marketing team workspace up here.

44:07We've got mentions and street team events.

44:09This is a semantic model.

44:10We're gonna add it.

44:11our data set to this model the tableau agent is thinking and it's about to add that context here so let's go ahead and look at that.

44:18Is a locked model that is it's it's owned by another team.

44:21Idol control it

44:22But I'm not stuck.

44:23This is governance and ad hoc together.

44:25I can build my own derived model that's linked to and leverages that model while adding this new data about street teams.

44:31So that's really again another point.

44:33So it's not us allowing us to edit the original model, but it's saying you can you can use this

44:40in your own work by adding this file to this locked model, but we'll create another instance of this model on your behalf so that when the main model changes or you get all the benefit of that

44:52But your little Excel file can live in this world out here and uh just be attached to that.

44:57Now it's got this sort of dotted line to suggest that it's not a permanent part of that model.

45:02Then you've also got suggestions on the left hand side to other data sources that I think might be useful.

45:07I think that'll probably come into play here.

45:08So you can see on the right hand side it's it's given us a little bit of a suggestion.

45:12So again you get to accept the suggestion.

45:14Let's see what it does

45:15So it's suggested that we add our data as well as add a calculated metric on the sales of those street teams.

45:21And it's suggested we get to add additional data.

45:24Now again, we can do all these things manually manually as we always have.

45:27What I know about data practitioners is they have their workflows, but I bet what you're gonna find when you start using this.

45:33is you'll go, wow, actually that's the thing that used to take me 50 clicks and two hours and maybe some Googling, the agent could just do for me so I can focus on that higher value.

45:43And so this is that sort of point.

45:45They're really drumming at home.

45:47And I think it's a really important part of this experience.

45:50I could have done all that work.

45:51And in some cases, in some cases

45:54I will argue this is a demo, so it's really hard to know how we got here.

45:57My experience of using Tableau Agent today hasn't been as few conversational points as this.

46:03My experience has been typing a lot of things, going back and forth, it getting it wrong, and you know, I'm six or seven questions in and we still haven't maybe achieved a very basic thing like building a chart.

46:15This experience of agent is, I think, a lot more aspirational.

46:18I hope Tableau can achieve this very soon, and because it it it it proves two things.

46:23It's actually quite capable.

46:24It doesn't need you to keep asking it to do stuff.

46:27And if it really works like this is suggesting, um, then what it means is that it's also quite advanced in the way that it brings context to you.

46:35For example, knowing this other data set could could be connected to it.

46:39That is uh something that you might know how to do as a data practitioner, but you might not have thought to do immediately.

46:46And the only way you sort of think to do it is by

46:48Maybe someone asking about another metric and you remembering, oh yeah, that sits in this other data model.

46:53Let's bring it through.

46:54So I think it will help with this discovery issue that Tableau's had, which is when you go into Tableau and you build a data source.

46:59you have to do all the work.

47:01But in this ecosystem, because you're working with a semantic model, it's hopefully building that sort of really large context of the business.

47:08and bringing it right through.

47:10And I think that's super interesting in terms of how um this could potentially make it easier as an analyst to know what to do.

47:17Which is actually a a a big barrier sometimes in the Tableau ecosystem.

47:20It's an extremely powerful platform.

47:23um more so than you know many of the other tools that go up against it but it's just so hard to understand where those features are.

47:29I continuously see really experienced individuals like people who've been using Tableau for a decade

47:35just not know what I consider to be core functionality in the platform that I I would think everyone is caught up with that's been released in the last sort of two or three years.

47:45It is still a huge problem.

47:47And I will still see

47:49sort of instances where people have old habits.

47:51Um and it's not to say that I'm perfect either.

47:54Like I still do some things that are quite old in in many cases.

47:58But this is the problem.

47:59Unless the product surfaces these products, these features to you in a better way, why would you change your workflow if you know how to do it and you know how to do it quickly?

48:08So let's uh let's carry on

48:10So we're guided to add some data now and now we're bringing in some enterprise data.

48:14We're connecting directly to workday and ingesting some So let's just go back there.

48:19So this is really interesting.

48:21Um these are some cloud connectors.

48:23Now I would hope that these are very similar to the connectors that we get as part of Tableau Cloud.

48:29It's interesting that Databricks is a is a beta one there, which would suggest all the others are built and ready to go.

48:35It's not a very broad set of datasets, admittedly, so Snowflake Databricks is a pretty big one.

48:41Postgres is a pretty big one to be fair.

48:43Microsoft SQL Server isn't on this list, but I guess they're not going to give credence to Microsoft in a demo, so let's get rid of all of that.

48:50You do have Azure Synapse up at the top, which is probably what most people are using now compared to what happened before

48:56You've got Azure SQL Server as well.

48:58Maybe that's actually the new name for the product.

49:00Uh Makito Slack, uh Google Cloud Storage.

49:04So, you know, broadly speaking.

49:06you do have um the big kind of um data sets here.

49:10I don't see Oracle, that's sort of another big company name in here, but maybe there's other ways of sort of accessing that.

49:16So it's good to see the support for sort of native connectors as well.

49:20And now we're bringing in some enterprise data.

49:22We're connecting directly to Workday and ingesting some information about our teams.

49:26So again, we've gone back into the connection window.

49:29You can see that every time we're in the connection window, we get these sort of orange icons.

49:32We're still in the marketing team workspace.

49:35The side pane has changed context again.

49:37Now we're looking at the tables.

49:39We've got the data interpreter that I think is saying they can work in here.

49:42I'm not sure how that works in a tabular context.

49:45Um but it might be able to, for example, pass some of these bits of information.

49:50It might do some nice data prep for you up front.

49:53But you've got these different tables.

49:54So Street Teams is the schema, I assume, and members is the tables, campaign.

49:59probably a little bit of a data model here you tick the members you see the seven fills 108 000 rows and then again you've got this agent on the right hand side so this agent is a consistent

50:09persistent part of this experience on the right hand side and I think it's super important because

50:17What it what it's constantly doing is it's showing you this agent interface rather than the native interface to do any of these tours.

50:26Now

50:26What I would be curious to understand is what does the native inter interface for these for these capabilities look like?

50:32For example, if I go to this drop-down for member,

50:34Can I split these using spaces?

50:37If I go to this uh Street Tim ID, can I split these using hyphens by clicking on this drop down?

50:42Um let's have a look and see if he actually shows us that because at the moment he's showing us everything through the agent.

50:51So let's walk through like what we've done.

50:52We had a how come we want to do it Cool so we added to the model it came back to the tab we already had open.

50:57I don't know if that's native, if it's throwing us in between these tabs natively or if

51:01If this is just the different Chrome tabs and it's just sort of managing that.

51:04I'd like it if it wasn't throwing you to different browser tabs, if it was just keeping it all in sort of one tab.

51:10But this could be sort of a side effect of the way they're doing the demo.

51:13So let's have a look.

51:14We've got sales team members yet, sales team events.

51:16And I think because we have the sales team members, we can make the connection to sales, street team sales as well.

51:23Eve, and we have all the tooling and capability to do it.

51:26Tremendously powerful.

51:27We can do it drag and drop as we always have, but we now have this powerful agent.

51:31knows what we're trying to achieve and walks us through the process.

51:35Now get back to our visualization and continue analysis.

51:38So that's a glimpse.

51:40of what the data pro skill looks like.

51:43Again, think of it very differently.

51:45This isn't a feature per se.

51:46It's a skill that you see infused throughout the new Tableau and all made possible by this new agent force technology.

51:54Now what's really interesting at the top here is you have this option to test the model and ability to set something as well.

52:01So I wonder what that is in relation to the semantic model.

52:04But it feels like that's encapsulated this Data Pro sort of experience.

52:07So Data Pro is essentially the ability to go connect to data, uh, visualize it very quickly and bring it together and build semantic models.

52:14That's essentially

52:15the capability in the tableau agent is able to take that skill and make the most of it and it can drive that experience for you

52:23So now let's move to the next uh skill that we're building, and that's the concierge.

52:28The concierge.

52:29So if you're following on the top right hand side here, you can see the slides has changed.

52:32Uh we're still looking at the new tableau.

52:34It's cut off a little bit.

52:35and we're going to the concierge, as it were.

52:39Think of that like this is the smart analyst who can answer any question, but it's so much more than that because

52:45When I've worked with really brilliant analysts in the past, over over time they learn the business model, they learn the business problems, and in fact they're the kind of people who come in to the to the office or shoot you a note on Slack and say, hey, I just had this thought.

52:58We were talking about this last week.

53:00And I was thinking, maybe if we look at it this way, look at this insight I found.

53:03That's what the concierge is doing with you.

53:06When I open this up, I'm immediately guided to the

53:10Things I ought to focus on.

53:11Think of this like a a business user focused digest of all the metrics I care about most, tuned and prioritized by AI, so that when I tap into this, I get

53:21of the metric that I care about.

53:23So this is essentially Tableau Pass as it is today.

53:26It's being sort of phrased around this concierge.

53:29um skill, which makes sense because you can already do questions and answers in Tableau Pulse today, but I think this framing of a concierge as a skill makes a little bit of sense.

53:39So let's see how this sort of plays through

53:42automatically built and it's beautiful with best practice and crucially it's prioritized because it's forecast the future so it's predictive and in fact this is what you're seeing here is a partnership between the inspector agent which found this insight behind the scenes

53:55concierge agent which lets me go ahead and do exactly what everyone would do when they see hey something's forecast in the really subtle clay um

54:06This yeah, so let's break this down a little bit.

54:09So this is Tableau Pulse, it's a Tableau Pulse metric.

54:11Today it doesn't do any forecasting.

54:14So that's a big hint as to what's coming here

54:17But additionally, the inspector here is doing some computations in order for it to be able to guess that forecast, to be able to say, hey, this is what's going this way.

54:27It would be good to have a little bit of the data behind that because it typically to do that kind of um sort of predictive analysis, you'll have certain drivers and factors that you'll be tracking.

54:38And so um you'd be able to see that now.

54:41I would love to see if, you know, this forecast, it could tell you what's driving that forecast in that in a more sort of clear way.

54:48Just showing it like this says, okay, this is where it's forecast to go

54:51But I think the the big question people have is why?

54:54Why is that?

54:54Like what products or what behavior is going to drive this negative um sort of element?

55:00We can see a breakdown just off-screen here of what I think is happening with the minus 22%, but I'm not sure it's telling us about sort of the future.

55:09So let's carry on.

55:10About automatically built.

55:12It's beautiful with best practice.

55:13And crucially, it's prioritized because it's forecast the future.

55:17So it's predictive.

55:18And in fact.

55:19This is what you're seeing here is a partnership between the inspector agent, which found this insight behind the scenes, concierge agent, which lets me go ahead and do exactly what everyone would do when they see, hey, something's forecast in the wrong direction, is to ask.

55:31What's driving that?

55:32You can just ask why in your own language.

55:35And we get this new visualization built automatically on the fly that answers that and shows us that this is happening because of a decrease in inventory.

55:42And notice it's certified.

55:44This is the power of trust.

55:45This is where Tableau says you can model your data and your business, certify it and trust it so people So this is um

55:53This is a an interesting one and let me just give an analogy here to something that's happened recently.

55:58Recently, um Apple started doing these called news summaries.

56:02And the thing about these news summaries or these summaries in iOS 18 or whatever is that they take your notifications and they summarize them.

56:10The problem is there's some notifications which if you summarize them correctly

56:14drastically change the meaning of the context of the thing you actually try to read.

56:18And where that relates to this is that it's obviously done some analysis on this data set

56:25uh chromatics, gama tech, whatever, and it's saying look, these are gonna run out of inventory.

56:30Great, that's a good analysis.

56:32But the data source is certified.

56:35But that's been badged onto the chart that explains the AI.

56:39So if I'm a lay user, what I read is that this is a certified piece of analysis.

56:45When in actual fact it might not be.

56:48No one's certified this.

56:49The data source is certified, but not necessarily the analysis.

56:53And so I think

56:54there is a distinct difference between those two, especially if you've got AI involved.

56:59And I think it's it's really important to say data source certified.

57:04or analysis certified.

57:06Analysis certified could also be another batch where if someone does the analysis and goes and looks and says, okay, that's actually something that is happening.

57:13Yeah, someone with authority like a data engineer or a manager can say, yeah, I'll certify that that is what's happening and they can hit that and then this analysis becomes certified.

57:22That could be another way of looking at this.

57:24So that's a little bit of sort of product feedback um in situ there.

57:28Um but I think this sort of little box that it tries to explain it's still quite a nice idea.

57:33So they can they can count on the answers.

57:36This is actually a the right answer.

57:38And you can ask your follow-up questions just like you'd ask a colleague.

57:41Note, I'll ask, is this everywhere or only a few markets?

57:44And even though market isn't in my data, it still figures out that I mean region.

57:47Yeah, same problem.

57:48And shows me that actually there's some regions where that is

57:52uh trending down, we're actually in risk of a stock out where there's others where we have too much inventory.

57:56So I want to take an action.

57:58Ryan mentioned that actionability is a huge focus for us.

58:02So what's essentially happening in this concierge experience is on our phone we're just having this chat back and forth.

58:08We've discovered some insight and now we're going to make a request to change the inventory.

58:12And this is quite a powerful button to have inside of an application.

58:15But if it's an action that's been configured and certain people have the ability to do it, then that's great.

58:19Here's a summary of the inventory change request.

58:21So task.

58:24Adjust distribution, product, chromatics plus one more, region, New York metrics, New York Metro plus three more.

58:32There's actually five more suggestions.

58:34And there's a submit request button.

58:36And again, this must be a pre-configured thing that the AI can do.

58:41Essentially just sending an instruction to the database to say ship more products.

58:45uh to this particular place.

58:46It's normally a little bit more complex than that, but um let's just assume that it's it's it's simple.

58:51And we realize that action is how you translate insights into value.

58:54This is taking action, making a call to an external system, whether that's any Oracle system, any external system, Oracle, SAP, Microsoft, you name it, it's one tap to take that action.

59:05And that is tremendous power of this counciers.

59:08You can also, of course, go further to broad, strategic, open, root cause analysis.

59:14And this has found based on a new feature we are shipping called metric relationships.

59:18It's found an interesting correlation between two very different data sets and driven us to the next step in our analysis.

59:25So that's the all new concierge experience.

59:27I showed it here on mobile.

59:28but it will be available anywhere.

59:30You can leverage this in Slack or Teams.

59:32You can use this embedded in any application or Salesforce.

59:35Use it directly in Tableau or in the all new Tableau Mobile.

59:38So the conservation is basically this stream of let's say consciousness between you and the data source and the ability for the AI to take actions on your behalf.

59:47I like this little kick out to talk to an analyst.

59:50Um I think that's a really powerful capability.

59:53So often business end users don't have access to the analysts who build the dashboard.

59:58And so just just just cutting out the managers, I think it's a really, really good thing.

60:02Because often you find the end users have a different perspective of how they like to use things compared to how their managers think they should use those things.

60:10So it's a very interesting thing being able to talk directly to the end user and having that as part of the product feature.

60:17I'm sure someone will ask for that to be turned off, but nonetheless.

60:20Nice to see it.

60:21So this stream can happen in lots of other places.

60:24I think they've done it in mobile just to show there's a mobile narrative here

60:27Um but I can easily see how that works in other places.

60:30Now in the Tableau ecosystem, we don't really have a natural place for this.

60:34We'd like this would be mobile, this is the Tableau Mobile app.

60:37there's obviously slack but inside of the tablet cloud place we don't really have like a concierge chat window the closest thing to that is the search box so potentially the home page will need to change to reflect

60:50these sort of insights coming to you and then you kicking off this discussion, this concierge sort of Q<unk>A session over on the right-hand side, or maybe as a central part of the window.

61:01All right, so that's our second new agent that we're working on, the concierge.

61:05The third is the inspector.

61:08So this is the one working behind the scenes.

61:10to find interesting insights.

61:12And this opens some really cool possibilities.

61:14Remember, the old model of BI, the old model of technology is the software sits there and does nothing until you tell it to.

61:19And you got to tell it the right way and get the right drag and drop to get your answer.

61:23In the new world, the insights come to you.

61:26So let's check that out.

61:30So now I'm working in Tableau and I'm dragging and dropping.

61:34I'm trying to understand this relationship that I was looking at in the previous section.

61:38between social media mentions and street team sales.

61:42And what I notice is actually that I can I can interact and drag and drop as I always have.

61:48So if I want to drill in and take this data from the product level, look down at the individual street team level, I can just drag and drop that onto detail as I always have and interact.

61:57and with the visp.

61:58So this is really interesting.

61:59I can see there's clearly some kind of a correlation.

62:01There's some street teams here in the kind of top right quadrant that are really outperforming and there's obviously whether causal or not, there's some kind of correlation between sales and mentions.

62:12Before I do anything else though, I notice that there's this insight that's been detected.

62:17And this is really, really powerful.

62:18Now, if you've used Tableau for a while, you might be familiar with the set feature, which lets you in the middle of analysis.

62:23.

62:24lasso or select a group of marks and put them together into a set or define that with a formula.

62:29And for years we've heard customer feedback saying we want to use that feature.

62:33We want to save

62:34the outcome of analysis.

62:35We found this really interesting cohort.

62:37We want to use it elsewhere, outside of Tableau.

62:40That's what a segment is.

62:42Check this out.

62:43The inspector has actually detected that segment automatically.

62:46So I'm going to say few, and it'll actually show me the segment that it's identified.

62:50It's found this top right quadrant is a cohort of top-performing street teams, and it's suggesting that we create this as a segment.

63:00So this is really nice because sets exist in Tableau today, but they kind of live within a workbook and they don't really make make it to the platform in a meaningful way if you wanted to get that set across other workbooks.

63:11you'd have to recreate it a little bit and you you know there's no clean way of sort of transporting them in between assets.

63:18This is actually a really good idea.

63:20Um again there's a little bit of iconology to suggest that the segments are going to be supported

63:25part of the semantic model.

63:27So you'll be able to add a segment and then track that segment across maybe different models where the same customer identifiers have been defined.

63:37Um, but this is a really nice thing.

63:39Now again, we're doing all of this in the side edge, and this is the thing we haven't seen.

63:42And maybe this is why I struggle to sort of connect with this, which is that I'd love to see the manual way of doing these things as well.

63:49Because if I have the choice

63:51Then I think this is going to be really powerful if I have the choice.

63:54Because there's some things here where I know what to do.

63:56So I can do it quicker than I can type in this box on the right hand side.

64:00So I can just go boom boom boom boom boom done.

64:02But in other cases I don't know what to do.

64:05I can go ahead, click, click, accept, boom, and the same outcome is there.

64:09So I'm able to switch modes, switch user interface types.

64:13to be able to get to the end solution a lot faster.

64:16And these prompts on the right hand side can help trigger thoughts as well.

64:19And maybe interesting thoughts, maybe trigger conversations.

64:22Say,

64:22You can see here you can connect with marketing on Slack to say, hey, can I create the segment?

64:26Does it tend to work?

64:27I think that's what he's about to do here.

64:28This is actually quite smart.

64:30So we can continue that conversation now outside of Tableau, which is exactly what would happen.

64:35You'd use Tableau, you'd find this insight, and then you'd probably go make a presentation about it or put this in a document or Slack your boss and go, look at what I found.

64:43Now you can leverage that.

64:44So this segment that we found of top-performing street teams, we can bring into a conversation in Slack or Teams or wherever, and we simply say exactly as we'd ask.

64:53uh Carlaiden would say, hey, um I was analyzing the data on street teams and I found this top performing segment.

64:59Should we publish it so we can better target them?

65:01Well let's ask that question.

65:03Our colleague says, yes, that sounds great.

65:06Let's use that segment to target them with a bonus and leverage that in some other system like Workday.

65:11So you can see we used the conversational experience to find some interesting insights.

65:16We used the drag and drop experience powered by an agent

65:19The inspector agent then found this interesting segment that we should use.

65:24We brought that back into conversation and we're going to use that somewhere else to drive some uh gratitude bonuses for our street teams.

65:32Really seamless

65:33the data and the analytics working together with all the rest of the ways and places we work.

65:37And so we can publish that segment directly here from Slack and go on and use that in other applications or back in Tableau as a cohort for further analysis.

65:46Really, really cool.

65:47I think we're bringing together the world of analysis with the rest of the ways and places that we work.

65:53All right.

65:54Last.

65:54Uh but not least, let's talk about scaling this out with analytical apps.

65:59So the inspector is basically this thing that's able to look at your data.

66:03and pull out insights and then like kind of act on your behalf.

66:07So it's essentially monitoring and looking out for trends.

66:10It's a bit like explain data today.

66:12and where it's able to sort of dig behind the data source but it's doing it proactively without you having to click on a data point or click on a on a on the set of values and then get it to explain what's going on

66:25Very often when we find some kind of an interesting insight, the next step from that is we gotta roll this out to people.

66:31Let's say I find some

66:33interesting insight about in-store sales.

66:34Well I probably want to put that in some kind of an analytical app that all my store managers So this is a interesting term.

66:40Um and this is not just Table doing this.

66:42Quite a lot of companies at the moment are pivoting away from the term dashboards because they've got a bad rep.

66:47and moving towards this concept of analytical apps because they want to imbue this concept that uh data is not just for consumption, it's also for taking action.

66:57And an app

66:58Suggests that context for being able to take action.

67:01So this is an observation.

67:03You know, Sigma is another tool that's moving towards analytical apps

67:06um uh and applications as they like to call it but yeah um it's it's essentially what we know of today is a dashboard.

67:13Whether or not that name will live on, who knows?

67:16Store associates can use

67:18How do we do that?

67:19Well, here I'm going back to this visualization we just uh landed on, powered by the agent.

67:25And I want to make this part of an analytical app that I roll out to my colleagues.

67:29To do that, we've created a whole new capability, a whole new experience called the workspace.

67:34And it's this brilliant canvas that lets you see in one place all the ingredients of analytics from data sources, transformations, metrics, visualizations.

67:42You name it.

67:43And so from here.

67:44So this is a a sort of a running theme and they've come to it last.

67:48Um, but I think it's a really important thing.

67:49Because this to me

67:51In a video I did ages ago about the the like perfect version of Tableau, I had a very similar concept, nowhere near as good as this, but it was the idea of atomizing the whole Tableau experience.

68:03and allowing you to take those components anywhere you wanted in the ecosystem.

68:07And the workspace is essentially that instantiation.

68:10It's this idea that

68:11There is a little bit of a hierarchy between all your assets and you can actually make them modular regardless of where they live, what they do, and what stories they tell, and be able to reuse those assets in a lot of ways.

68:23Partly by design of how clean this whole setup is and partly by actually removing and stripping out a lot of the identity that we tend to try and build into dashboards by just making this all modular, making these metric sort of cards

68:35and making these charts a sort of individual item, you're able to take these visualizations and bring them together into this sort of idea of a dashboard.

68:43And you can see that as he scrolled sort of down here, let's just go to the next thing.

68:46We have this hierarchy.

68:48So at the top you had analytical apps.

68:49then you had visualizations, then you have data models, then you have prep flows, then you have data sources.

68:55So this whole hierarchy.

68:58of assets um and your semantic models essentially live here in the model so we've actually been working through this uh concept a little bit and you can see all the different sort of datesets here at the bottom

69:09We added this Street Teams events uh element, this was modified two minutes ago.

69:12Actually very good accuracy on the demo here, because that wasn't two minutes ago, but it was

69:17quite a bit a quite a bit ago and all of these modifications that we've sort of been uh applying here are actually reflected here in this chain so good attention to detail there

69:25But if we just um go up you can see that's this is basically the hierarchy.

69:29And you've got this ability to filter and set things out.

69:31So um there could be multiple analytical apps here at the top

69:36um for the marketing team.

69:38And another thing that's sort of not obvious here is that what we're looking at here is actually an expanded view.

69:45linked to one asset at the very top.

69:48So the reason we're seeing these visualizations is because we've selected this marketing team

69:55um uh analytical app and everything else linked to it is essentially an exploded view of how that asset has come together.

70:04So the idea is if I was to, let's say, click on something like mentions, I would see all the analytical apps that use the term mentions, but then here at the bottom I'd see only the models that feed mentions and you sort of get this hierarchy building out.

70:19It's not just a static thing where you have to see everything all at once.

70:22It's a dynamic living page where as you click through the experience and you sort of browse different parts of the ecosystem.

70:29Um it's coming to life and exposing different components throughout this whole hierarchy to you in a way that's really deep and meaningful.

70:37And I think this is great because as an analyst, another thing you find is that

70:40It's actually quite hard to keep track of sort of what's in what sometimes you'll just go to a workbook and you want to know what the data source is.

70:45It's really hard to find that answer unless you have

70:48the data management add-on so it's really good to just be able to view that whole experience go all the way down and just see this all here so this obviously suggests that you know in order to use this experience without sort of the pricing models that we have today

71:01um you'll need to have sort of the greatest and best of Tableau for this to really work.

71:05So Tableau Plus is essentially going to be the license that you'll need to be able to get this because it has everything baked in.

71:11So anyway let's carry on.

71:12We've only got a few more minutes, three more minutes.

71:14Let's have a look at this um workspace.

71:16transformations, metrics, visualizations, you name it.

71:19And so from here, we can go and build that up.

71:21But we can do that manually as always.

71:23And we can drag and drop.

71:24configure freedom, flexibility, and choice, but I'm going to use the agent and I'm going to actually ask it exactly what I want to achieve.

71:32I'm going to say

71:33Can you build me an app for managing Street Teams?

71:35This is that analytical app builder agent.

71:38Okay.

71:38We've trained on how you build analytical apps.

71:42It turns out that we have, in the years we've been doing this, 21 years, we've observed millions of dashboards.

71:47And so we've learned a thing or two from that.

71:50So we asked the agent, can you build an app for managing street teams?

71:54And it does.

71:54It goes and builds us a starter analytical app.

71:58Let's take a look and see if that's what we're looking for.

72:01So it shows us a preview of this app and it looks pretty good, but I want it to be styled in my corporate Gate of Fam Kicks theme.

72:07That looks so great.

72:08I love that gradient.

72:09So I like this um sort of nod to a few things.

72:12First of all you've got layouts here at the top right hand side.

72:14You've got themes.

72:15This is actually available in 25.

72:171.

72:18This is the sort of the groundwork coming for this

72:20Um then you've got the metrics, the specific metrics that's gone and pulled in.

72:23So this is actually quite easy for it to do because if you have a decent semantic model, then you should be able to say

72:29Well, what kind of metrics make the most sense to support the management of this?

72:33And there's already some context because I'm already looking at the data.

72:36So if I asked this

72:37in the aging context, it should be possible for it to sort of narrow in just using the semantic model on what makes sense to see and then give some sort of hierarchy based on metadata.

72:48of what people actually consume.

72:49And if you pair those two things together and you're a little bit half smart, I think you'll be able to figure this out very easily.

72:55The heart the tough thing is the layout.

72:57And I'm not sort of

72:58dismissing how hard it is to set something like this up because there's obviously many different layouts and you could end up with something that's just a like massive big mess.

73:05So these sort of patterns on the right hand side that give you some sort of

73:09idea of the ways you can break down the layout I think is quite good because it confines you a little bit and that's the thing sometimes this these sort of confinements can just focus your mind on getting the outcome rather than just playing around with colours which is

73:22Unfortunately what we tend to do a lot in Tableau ecosystem because it's so hard to get repeatable results without sort of themes and styling.

73:29So yeah, this is this is really nice.

73:32I love those rounded corners.

73:33I don't think impressions is going to be really useful for my audience, so I'll deselect that and I'm going to switch the layout again because I like that kind of 16-9 widescreen landscape.

73:43And with just a couple of clicks.

73:45And with expressing what I wanted, Tableau and this new agent for analytical apps has gone and created a really awesome looking dashboard.

73:54But we want to take this to the next level.

73:55I'll hit create.

73:56Now I've got a dashboard, but we want to make this an analytical app, a dashboard that does something, and we want to power it with this new agentec technology.

74:03So I'll select

74:04This viz here in the top right, this scatter plot, I'll go to actions, and I'm going to add an action to this.

74:09This is how we take again, we translate insight into value.

74:12So Salesforce actions already exist today.

74:15Um so this is like an enhancement of that feature, if that makes sense.

74:20Oops, I went to the wrong place.

74:21Let me just get back to where I was.

74:23There we go.

74:24Board, but we want to make this a an analytical app, a dashboard that does something, and we want to power it with this new agentec technology.

74:30So I'll select

74:31This viz here in the top right, this scatter plot, I'll go to actions.

74:34And I'm gonna add an action to this.

74:36This is how we take, again, we translate insight into value through action.

74:40And I'm gonna add a recommend strategy action.

74:44This is such a different mindset than how I used to think about analytics, which is I gotta put a button that updates a record.

74:49I gotta put a button that sends an email.

74:52It's a very different way of thinking.

74:54This is my me, this is our saying from this viz, I want to ask the agent to recommend a strategy.

75:01So what does it actually look like?

75:03Let's save this and take a look at it.

75:05So in practice now I've uh oh yeah right one more thing I gotta show you what that actually looks like this flow behind this this is powered by Salesforce Flow this is

75:14powered by the agent force technology, configuring this is not necessarily a function of writing a whole bunch of code.

75:20You can do this with Pro Code, but you can build these with low code or even no code just by expressing in natural language.

75:26what the capabilities are that you want.

75:27So we'll preview this.

75:28This is the action that I want to add.

75:31And we will then save this.

75:33And let's go see how this looks when we've uh published this app.

75:38And used it.

75:39So practically speaking, I'm looking at these street teams and I realize there's a bunch of underperformers, some who are not doing a lot of digital mentions, they're not getting a lot of sales, and I want a strategy for what to do.

75:48to help those street teams increase their game.

75:50So I lasso them and I've got all the actions that I would normally take, like group, keep, exclude, publish segment, but now I have this recommend strategy action.

75:58When I tap on that, that brings open the agent.

76:01pass it that data in context and it says hey I see that you've selected these street teams based on this I recommend the following training mob modules I said that sound

76:11That's really like um complex, right?

76:14Because you've got to have all that content ready to go.

76:17You're not gonna have like uh 34 minutes of training ready to go.

76:20But maybe in the world of AI you can

76:23You can create it on the fly maybe.

76:25So that's an interesting piece of concept.

76:27I think there's a lot of tea in with workday as well here.

76:30So I don't know if that's a capability in workday to be able to do that.

76:33I'm not entirely sure.

76:35But I get I get how that action can actually make a ton of sense.

76:38You know, setting up capabilities and giving people the ability to hook those right into the ecosystems they use.

76:44Now in reality a lot of companies have dedicated systems for making these decisions.

76:49um they're not always Salesforce.

76:50Salesforce is just a porcelain to some of those.

76:53A lot of companies also export data out of Salesforce into other data systems so they can make those decisions in a holistic way elsewhere.

77:00Um that's sort of literally why day trialytics exists because part of that movement of information is also movement of data.

77:07So find them up.

77:08for that training.

77:09Everybody's trying to figure out how to manage a workforce and transform a workforce in this new world.

77:13One of the powerful things about Salesforce is we have this whole trailhead platform as well where you can build that knowledge management.

77:19Tying that together with agents and with data, tremendously powerful combination.

77:24So that's the fourth of these new agents.

77:27Okay, and that's pretty much it.

77:29That's that's sort of the end of this demo.

77:31Um and so we got all four of the uh of the agents, and I think this was a very cohesive demo.

77:36There's there's something that happens when you watch these

77:39um YouTube events and maybe it's just me, but when you join these webinars, I feel like you do have to watch them.

77:48Three times?

77:49And it sounds ridiculous because it would suggest that it's not being delivered well at all.

77:53And that's not the case at all.

77:55You have to watch it three times because the first time you hear it

77:58Your brain is is just essentially processing the new bits and you're definitely distracted and you're definitely not paying attention to the detail.

78:09On the second pass, you start to process the information you took in the first watch.

78:15And as you're processing it, you confirm your perceptions, you confirm your understanding, and so on and so forth.

78:21And then on the third watch

78:23That is when you actually start to think openly and think, wow, what could I do?

78:28How could that work?

78:28How could I do this flow?

78:30And I think it speaks to why Salesforce keep repeating some of these notes

78:34This was the second time we saw this demo in quick succession.

78:37The first time was actually a month before, and they keep kind of trying new messages each time.

78:42But I think there really is something in in what I've just said, which is

78:46I often get the clearest idea of what Tableau is trying to do on the third or fourth watch of these videos because I have to watch them again and understand them clearly in order to make content about them.

78:58So

78:59It's just some it's like a side effect of sort of what I'm doing.

79:02I know no one has the time to watch what is essentially a 15-minute uh you know webinar three times over.

79:07That's basically half a working day

79:10Um but I think there is something meaningful that if you're a partner or if you're in the ecosystem and you're supposed to be teaching other people about this platform

79:17It wouldn't hurt to watch this content multiple times to really start to get to understand why this is a big deal and why it changes because

79:27There's a philosophy in this in this in this presentation that you know Tableau as a team

79:35can tell you what it is, but it's just not the same as having that view and opinion of yourself, like envisioning that workflow in your own eyes, in your own steps, so you can convey it to other people and just to tell them what it would feel like

79:48Tableau build the product.

79:50They are not the primary users of the product they built.

79:52They dog feed it of course.

79:54But um when they explain things, they're explaining it in their vision.

79:57But they've built the product.

79:59And so I think everyone it's incumbent on everyone.

80:02um who's trying to teach this platform or share it with other people to be really this deep, you know, three or four watches in, in order to be able to really understand some of the cues.

80:11And I don't just say that because I'm doing, I say that because I j I genuinely think

80:16For the first time in 20 years in the Tableau ecosystem, we need to re-explain the Tableau ecosystem from scratch.

80:23If you've written a book in the last, I don't know how long, 10 years about Tableau.

80:27It's not going to work in this platform because you have to explain the whole Salesforce context all over again.

80:32If you've written a book about anything in the last year or written a blog about anything in the last year, unfortunately it's going to be out of date

80:41when this comes out.

80:42So um the only way to sort of stay sharp is to be really dialing into these messages and sort of separating the marketing from thing.

80:50We got into the habit in the community

80:52of essentially experience the product when it was in our hands.

80:56And I think it's going to be too late if we do that with this platform.

80:59And that's that's not our fault.

81:01That's obviously a challenge for Salesforce.

81:02Salesforce need to do a better job marketing and getting this product into our hands.

81:07sooner so we can actually talk about this essentially having like a an early preview that you know our skeeks can get into and then following that end users can start to experience now

81:17The thing about this other this platform so far, another thing we've seen, and this is now me just sort of riffing on what I've seen and what I've not seen, we didn't see a lot of ways of doing things manually.

81:27Everything was through this agent experience and I think that's because of the agentic sort of explainer.

81:31I really hope it's not the only interface.

81:34The other thing I didn't get a good understanding of is

81:37this sort of legacy perspective of having a creator, an explorer and a viewer.

81:43And for argument's sake, let's just delete viewer for now and let's just imagine that you either make content or you browse content explore

81:50So those are sort of the two paradigms and viewer is just basically a weaker version of Explorer.

81:54And I never know why the Viewer even exists.

81:57I just wish Tableau made Explorer cheaper and got rid of Viewer and then it just

82:01made it easy for everyone to just consume tablet, just give everyone the features they need.

82:06But I didn't see much of that experience.

82:09And I wonder if because there is no delineation in this new platform between

82:13someone who's consuming and someone who's making things because AI is the big equalizer that allows everyone who has access to be able to go and do things on the platform.

82:24I don't it's not something they've said overtly and it's not something they're calling out overtly in here, but I wonder if that is actually an implication of this approach, if it works, if it's actually going to achieve on this vision.

82:36You know, I'm doubtful in some aspects because you know I've not seen the today Tableau Agent.

82:41The thing that's called Tableau Agent Today is miles away from what we're seeing in this demo.

82:46So how you draw a straight line from where we are today to, you know, what is it, June 25, I'm not entirely sure

82:54But this is promising.

82:56It's a great vision to have.

82:57And if this is the aspiration, I'm here for it.

82:59And I think it's really good.

83:01I think I know

83:03More about this platform.

83:04I have a much clearer idea of it.

83:05This was my fourth watch of this video, so I've actually watched it a few times.

83:09Um this is why I'm sort of doing this.

83:11process and I think this is a really good way of doing things and I know this is a long video but this is for like the top 10% of my audience who will stick around for the full length of this video.

83:19I really appreciate that by the way

83:21So in the next video, what I'll do is I'll take the synthesis of everything I've covered in here.

83:25I'll probably process it a little bit more, probably next week.

83:28And yeah, we'll make a 10 to 15 summary

83:31We'll make a 10 to 15 minute summary of the new Tableau platform and I'd really love your help in sharing that and getting that out to everyone.

83:39Thanks for watching.

83:39We'll see you in the next one.