Best features showcased at Tableau Conference 2025

0:00Okay, so we're back.

0:01Day two.

0:02Simpler setup.

0:03I just went with the phone this time.

0:05I completely messed up the setup yesterday, so yeah.

0:08Apologies on my behalf.

0:09We're all off camera and everything.

0:11So anyway, content is what's important.

0:13Today we're talking about devs on

0:15Absolutely.

0:16So first impressions, what do you think?

0:19Yeah, it was it was intense, right?

0:21Yeah.

0:21Like it was I think so it was a different setup this year.

0:25So Matt Mill wasn't hosting, it was giving back giving back to devs, which I think was great because again

0:30Giving more FaceTime to fantastic project managers, uh developers, people that are working on the product.

0:36Absolutely.

0:36G give the show to them.

0:38Obviously you've got the pantomime theme, dad jokes galore, like everything passed super.

0:43Good dog puns all the way through.

0:45Storytelling was on point.

0:47Um I think what do we see?

0:49So we had half an hour, half an hour of devs on stage, and then half an hour of a new thing, which was devs on stage labs.

0:56Absolutely.

0:56Yeah.

0:57Uh do you want to explain what that was?

0:58Yeah, so um devs on stage

1:01If you're not sort of familiar with the keynotes, Devs on the Stage is where a tablet developers who actually build the product show the capability.

1:08Devs on the stage labs is the same thing, but what they were doing is showing us very early builds and asking us to sort of vote on

1:15direction and so what they did is they brought three very different ideas and they asked us to sort of show our sentiment.

1:20Uh Matt Miller made a point of saying that these weren't pledges to build these things, they're they're very much just hey if we built this what would what would your vibe be?

1:29And I think it was a nice

1:30And no problem.

1:30Because I think we finished devs on the stage thinking, Oh, this is gonna be it.

1:34And then they sort of did a one more thing kind of thing at the end.

1:37It worked nicely.

1:38I think it worked really well 'cause like f you had half an hour of like quick fire feature, feature, feature, one after the other.

1:43The thing I liked about um

1:45devils will say his labs, I think Alteryx have done something similar previously.

1:49I think the really interesting thing that I like about it is it reminded me of what used to be called Octo that Tableau had, the office of the CTO, like who worked on like moonshot projects that were like

2:00Let's just see how far we can get with this and see what w is there something here?

2:03Yeah.

2:04Um and I think a lot a lot of the things we saw came from small things like hackathons I think we do internally.

2:09And I think servicing that great work is really important.

2:12It's super good.

2:13I I think it was just a nice um

2:15sort of homage because with Devs on Stage was now separate again.

2:18I don't think it was last year, was it the end of the keynote?

2:20Yeah, probably.

2:20So we got more time.

2:23I think to talk about Tableau core products.

2:25There's a bit of Tableau Next in there.

2:26The we looked at the semantics connector and a few other bits but broadly speaking

2:30This was 90% core, 10% sort of let's say the next product.

2:35But what I did think was nice is that they infused a lot of AI capability with core, which was very much like I think it was like a subtle message to say look AI will change your workflow.

2:45regardless of whether you love next or not.

2:46Yeah.

2:47Let's just soften you up with a bit of this.

2:48Exactly.

2:55Surrounding a lot of the tableau messaging and things like this over the last year has been like why

3:00should I care about agents?

3:02Why is everything agents focused?

3:03Yeah.

3:04And I think today with a lot of the features it's like and this is why you should care about agents.

3:08This is why the role of the analyst is changing and like the jobs we do are changing and everything's transforming.

3:14And I think

3:15It it this dovetail for me really nicely with the keynote.

3:18I saw a post from a Tableau uh like uh the owner of a Tableau Partner, I won't name them, and they said they made a post by saying, Look, uh I know AI is great, but look, I can still do analysis really, really quickly.

3:30the type of analysis that was being done, I genuinely think a lot of people don't arrive at that same analysis quickly.

3:35I kind of felt like it missed the point.

3:37Yes.

3:37It's a very privileged position.

3:40Yes you could do the same analysis but I think fundamentally you're not getting

3:45it there is you're not doing the kind of analysis you should be asking AI to do.

3:49The kind of analysis you should be asking AI to do is the analysis you don't have time to go and do.

3:55So go down twenty paths exactly and see what comes back.

3:59Do it

4:00for me.

4:00So the reasoning capability they showed was really nice.

4:02It went off, did a bunch of analysis for you, gave you some clues, but still left you with the agency to take action on where you think the story should go.

4:11And and this is where humans will will have to be part of that process where

4:15Like I think the the uh one uh what is it, the one fly in the ointment is like AI and everything that's agent focus assumes that people are critical thinkers and they're not.

4:26They're not.

4:26Right right like w we just uh we live in a second best world where like

4:30people are aren't as smart as we think they are.

4:32So they will that there will be a lot of mistakes made.

4:34So therefore AI and agents have to have guardrails set up by knowledge curators, subject matter experts, all of those sweet things.

4:41Exactly.

4:42Right.

4:42Let's get it.

4:45So what we'll probably do, I'll try and cut some cut some slides from the real thing into this video.

4:50So um yeah I've got the list here on my iPad.

4:53So first of all we had a first first group of presentations.

4:56So we had Prep in DB.

4:57Prep in DB nice.

5:00keep getting fired at Ulrich, I'm telling you.

5:04And the funny thing is Prep started in the cloud, right?

5:06It's got a cloud capability first.

5:08So I always wonder, will Prep get to where tablet uh Altrix desktop is now faster than Ultrax can take Ultrix.

5:15to the cloud.

5:15I think so at this rate.

5:17At this rate, yeah.

5:18Then we had custom Python scripts.

5:19So this is a subtle one.

5:21Running in Tableau Cloud.

5:22Running in Tableau Cloud, which means Tableau themselves are hosting the scripts you feed them.

5:27That's an interesting one because I think that's going to use consumption.

5:30It's consumption but also like it's considering like any packages, will they need to be double double entered, right?

5:35Yeah, exactly, exactly.

5:37And then um the other thing was uh prep output to Google Drive.

5:40So being able just to export a data source right to Google Drive and being able to

5:45I think that's a really nice touch.

5:46It speaks to the continuous of the connectivity of prep.

5:50I think it really speaks to what I would call like the data prepper rather than like the the data engineer.

5:55The very different hobbyist.

5:56Yeah?

5:57It's the ho it's the hobbyist, right?

5:58Like some someone who wants to push to to do

6:00Google Sheets uh as a as a hobby, but you should you can if you can pull pull, prep and push, um that that works well as well.

6:06Exactly.

6:07Okay, um then we have the Tableau Semantics Connector.

6:10Now this is sort of important and I don't think it should be on Devs and Stage because I think it's on this

6:15necessary for the platform to succeed.

6:17So this allows you to essentially connect to um uh wait a minute, Tableau Semantics connector.

6:24Gosh I'm getting confused here.

6:25So is this a connector It's connecting to the semantic layer on

6:30your your um Tableau Next platform.

6:32Right, so okay, thank you.

6:33There we go.

6:34I got so confused for a minute there.

6:35It's a Tableau Desktop connector that connects to the semantic le uh level in in and you can make changes in desktop on top on and into that model and

6:45Exactly, exactly.

6:46Um then we had viewport parameters.

6:48Viewport parameters is gonna that was that was my that that was my higher.

6:53That that was my fire emoji, right?

6:55Someone's gonna use the map uh map layers with make it like a visual

7:00View zooming functionality.

7:02Exactly, exactly.

7:03I can't wait for someone to build that.

7:04But I think so.

7:05In my world, that's gonna be amazing.

7:06Right?

7:07We use XY coordinates for so many things.

7:08It's gonna be so so so from a sports perspective, definitely.

7:12Um analytical page extensions, another fire emoji.

7:15This is subtle, so I'll explain this.

7:17Analytical page extensions are the tab that you don't often go to called analysis next to the data pane.

7:22If you go in there, there's sort of analytics capabilities like median mode, box and whisker.

7:27At the very bottom of that pane is analytical extension.

7:30here they showed us a couple of good examples.

7:32For example the traveling salesman is an example or being able to draw contours on the map.

7:37Very nice.

7:38So really good, really good setup.

7:45feature the community breaks it and does whatever they want with it and then they build a better feature right so set actions then was followed by parameter actions because everything you're bracing set action for you now do a parameter action because it's better easier faster.

7:56We just kicked them away.

7:58That's right.

7:59Someone just

8:02It's all good.

8:03It's all good.

8:03Right.

8:03But th this this is a great example of people whose map layers now can should be able to code their own analytical page uh analytical page extensions.

8:12Pane extensions.

8:13A APS.

8:15APE.

8:15The apes.

8:16The apes.

8:17Right.

8:17So if you want to create an ape, you drag the ape in instead of a the um the uh map player.

8:22I think again this is gonna the community's gonna break this, you're gonna find some great things.

8:27Again, my f fly in the orbit is gonna be paywall.

8:30You get so many things that get paywalls.

8:32So part part of um either plus or it might just be data management and it would be the marketplace.

8:38Um, the marketplace, sorry, all the actual extensions will be on the marketplace behind a paywall.

8:44But I think tablet

8:45Tableau themselves will add more.

8:47What they did with the Viz extension is they use it as a vehicle to add stuff that they were gonna build as well as the community.

8:52So I think Tableau themselves have built this so they can enable maybe two or three extensions they wanna build.

8:57But like two.

8:58But they're relying on the community to do the heavy lifting.

9:00My two favourite things about extensions in general is that they auto update.

9:03Yep.

9:03And they um backward compatible.

9:06Correct.

9:06Across version.

9:07If you can do an extension, you don't need the latest and greatest.

9:10Like so it can just be.

9:11So it'll work for twenty four two?

9:14Onwards?

9:14Yeah.

9:15Which is good.

9:15All right.

9:16Cool.

9:16That was that was dev one.

9:17That was dev one.

9:18Um then we went to Dev Two Ginger.

9:20Yeah.

9:20A bit of a a chocolate theme.

9:23Factory Willy Wonka kind of theme to it.

9:25So we start with Tableau Pulse descriptions, I've called this.

9:28I'm not sure I've got a lot of things.

9:29Pulse powered

9:30By LLM.

9:31Yeah, so yeah.

9:32It's basically hard by LLM.

9:34What is that?

9:34You explain it to me.

9:35Yeah, sure.

9:35So if you've got a Tableau Pulse, again, we talked about this yesterday in the af after the keynote.

9:40It's basically f for a very specific user.

9:43The LLM part is

9:45Extending that user to have a coven conversation with your data.

9:48Yeah.

9:48Which is actually in reality what our data could, should, would have been, right?

9:53Yeah.

9:53So it's it's solving that problem uh but it's like investigating your data and adding that that reasoning and having that conversation.

10:00Pulse advanced Q<unk>A.

10:02Yeah.

10:02That's what it is, right?

10:03Which is good.

10:04Which is cool.

10:04It's good.

10:05It's it's a n it's a natural next step.

10:07Um and I think the thing about Pulse is it's n it's it feels like a mature product already, but I think it still has a lot to go because it's so focused.

10:15I think you have to be very tasteful about what you add.

10:17I think this is a good touch.

10:18Um then we had instant viz and tableau.

10:20Viz first show me.

10:21Yeah, viz first show me.

10:22So this is very different.

10:23So in show me today, you see the charts, but what it typically asks you to do is to select a dimension and maybe

10:30First, before it will show you what to build.

10:32This time around this is different where ShowMe shows you the chart and then it's able to look at the data source and suggest uh the measures that it's going to use to build the chart.

10:41So demo they showed a bubble chart, needed four things, it went and

10:45pick those four things from the data source and built the chart.

10:47No, it's gonna be wrong a lot.

10:48Yeah.

10:53We've got all your workspaces, you've got your canvas, and it's learning what most people build first.

10:57Exactly.

10:57In Austin 2016, Francois

11:00showed a very similar recommendations thing.

11:03Then the and I think most Tableau services will have the recommenders built in.

11:06That that almost started there.

11:08We're now what almost 10 years later and we've actually got the feature that he's talking about.

11:12Exactly.

11:13It's not live but we're like we're a lot closer

11:16Much closer, much closer.

11:17Good.

11:17Um this was the first crowd pleaser.

11:20This was.

11:20Colour palette themes.

11:22I love the demo for this, it's very succinct.

11:25Like it was click it.

11:27Select it, choose it, drop it.

11:28It was like a very dark

11:30Exactly.

11:31So good.

11:32So good.

11:33It was very good.

11:34But this is this is a long standing feature.

11:36I will make a point at the end of this video about Salesforce delivering on stuff that people have been asking for a long time.

11:41I think people critique them a lot for not doing that and yet I can't

11:45watch devs on stage and actually support that opinion ever but nevertheless let's keep going colour palettes and themes which are simple yeah so the the hack previously you have to go into the preferences pane and change it all of this that was literally the documented approach right the document approach and now you can just

12:00Do it yourself.

12:00Not only that, you can ask an LLM to generate a colour palette for you in that pain.

12:05Of course that comes with Tableau Plus um plus I'm sure.

12:07Yeah.

12:08But nevertheless, being able to create it, save them, style it and apply it.

12:13I fell into that trap myself.

12:15With an interface, super powerful.

12:17Okay.

12:18Um next one, dynamic colour range.

12:20This is Project Patchboy.

12:21We've seen Patch Bay.

12:23This like a shout out to Philippos.

12:24Yes, Philippos.

12:25He's no longer a sales for the record.

12:27So I think about three years ago we saw a preview

12:30of a framework that allowed for this capability.

12:33Further back?

12:34Yeah.

12:34Maybe at your last conference.

12:36And I think it was your last conference we still we talked about this.

12:39And it also linked to the viewport parameters.

12:41Because the thing about the viewport parameters is you needed a mechanism for a

12:45Essentially the different sheets that talk to each other and Patch Bay was eventually a framework to enable this to happen and so we've we've seen it in the it's data driven everything.

12:53Exactly.

12:54And now we're seeing the dynamic colour range in the same context.

12:57So basically it's dynamically changing that sort of legend.

13:00to contextualize based on the metrics that are available.

13:03So um I think we'll see more things like this, the build-off patch bay.

13:06Yeah I think the the the really interesting thing to say here is it's dynamic colour radius based on powered by parameters.

13:12Yes.

13:12Right?

13:13So like basically it's creating a parameters

13:15Yeah.

13:16But but it's paid from the selection you've made.

13:18Yeah.

13:19What I hope is you don't have to go and create these parameters yourself.

13:21That would be tableting to do.

13:23But I I th you probably will.

13:26Yeah, I know.

13:26I know.

13:27I know I know.

13:27It's a it's a little bit of like annoying US.

13:30But I think it should be should be more natural.

13:32Okay.

13:32Um server and cloud, accessible navigation and iC actions.

13:36Very good.

13:37So hear and understand your data.

13:39Oh, there you go.

13:40So this is the voiceover capability, uh being able to go into a chart, click a data point, and then not only

13:45just see the tooltip but it will narrate the tooltip as well very very good and the other half of this was being able to navigate a viz entirely using the keyboard no mouse involved so going across using tabs up and down all that I I really like this yeah because when you're doing demos

14:00Yeah.

14:00When you it was yes for absolute accessibility, but even for people that don't require accessibility.

14:05But I think everyone needs to be accessibility to make things better, right?

14:09Um but this i th it's a really clean way of visualizing.

14:12Like if you're telling a story with your keyboard, like

14:15It is it's like uh you got a lot of nice symmetry, everything's working quite nice.

14:19Like it just is a bit smoother, a bit quicker and a bit richer.

14:23I I really like that.

14:23Um random corners that's got the bigger chairs, this was the crowd.

14:28I feel like we're in like ninety ninety five.

14:30when rounded corners have just arrived.

14:31Yeah.

14:32Tableau has arrived 20 years later.

14:35So I I I mentioned this yesterday, right?

14:38I was saying like if everyone watched like you know spell checker rounded corners, spell checker rounded corners.

14:42But honestly, like rounded corners

14:45is great.

14:46Yeah.

14:46It will be used.

14:47I think the um the really cool part of this actually is um you can now set the radius of the rounded corners to create a circle.

14:55Now that's cool because if in my world if you have loads of shapes you can

15:00Just like spotlight the middle.

15:02So good.

15:02You could do circular maps.

15:03Circular maps?

15:04Yeah.

15:05Yeah.

15:05That little viewport parameter, circular maps.

15:08Zoom in.

15:08Mm-hmm.

15:09Yeah.

15:09Look at the big chart, the big thing on the bottom and then the little bit of a little bit of a little bit of a little bit of a little bit of a little bit of a little bit

15:12Little charties of the context.

15:13Yeah.

15:14You can get

15:15You can get a GTA map in the cloud.

15:16Oh love it.

15:17Who's gonna build a GTA viz?

15:19Right, okay.

15:20Dev three.

15:21Dev three, Patrick.

15:22Um, Skin.

15:23I don't know no idea what the acronym means.

15:25By the time I make a video about this I will know it.

15:27But what I understand of the feature do you know what the acronym is?

15:30No.

15:30Identity management.

15:31So don't trust us for that kind of thing.

15:34Essentially what this allows you to do is previously when you added users to Tableau Server or Tableau Cloud, you used to have to go and add them manually one at a time.

15:41With this, what's actually happening is when you add them to your identity provider

15:45They're automatically brought in.

15:47So essentially you can use something like Okta or I think what's the one on Windows?

15:52Windows Active Directory to manage your groups and then Tableau through Skim will just automatically go and update those groups and get them working.

16:00system for cross-domain identity management.

16:02Oh wow there we go cross domain that's a great name yeah cross domain identity does what it says on the tip all right yep um then super useful next part please a recycle bin this is actually already in 251

16:15but has not been pushed out to public yet.

16:17So they it was supposed to make the feature drop for 25.

16:191.

16:20It was in the release, but is not yet available anyway.

16:22So they did get away with a free hit at this conference because they got to say, hey, this is coming.

16:28Yeah, it's coming like in two weeks.

16:30It was contextualized by we it's on the ideas exchange, we listened, we built, yeah.

16:35Again, these are very simple features.

16:38We shouldn't be really celebrating these hard, but they're here and they've finally been delivered.

16:42I mean I've never deleted a viz.

16:45I just keep pushing stuff to an archive because the delete doesn't exist.

16:48Now, the thing is I think a lot of young people, young users, new developers, I say that in a very respectful way.

16:54I've seen like developers who are starting the last five years, I'm laughing at young

17:00They they delete content because they expect to recycle them.

17:02Yeah.

17:03And it makes sense.

17:03You've you've never had to like think about that.

17:06So yeah, um don't delete stuff on Tableau unless you have this feature.

17:09And more importantly, make sure you upgrade to this version before you go deleting content.

17:14Weekly.

17:15Backups.

17:16Take daily backups.

17:17Oh my god.

17:17Alright, DBT connector.

17:19I think this has been announced for Oh what did I miss?

17:21What did I miss?

17:22Activity login tableau server.

17:23Oh activity log.

17:24Oh I did miss one.

17:25Wow, there we go.

17:26Let me add it to my list.

17:27Activity log Activity login tableau server.

17:29Super excited.

17:30advice.

17:30So we were talking to a uh director yesterday, uh going through some future future Tableau stuff.

17:36The really the really fun part here is I mentioned this and he was like it was a strange reaction, which is probably because he knew this was coming out.

17:43He was like, this is already out.

17:44Yeah.

17:45No That's what happens when they think something's been related.

17:48So so this is basically saying what is the activity that a user is taking on my server by clicks.

17:53Yeah.

17:53Like really specific.

17:54I think.

17:55I think it's by clicks.

17:56Yeah.

17:56Because when we scroll through the the list that they showed on stage, like

18:00some of those things are things I'm really interested in.

18:02Now my part of my job is data fluency, adoption, enablement, right?

18:06This will really help understand how our visas are being used and what we need to focus on and like what is it the action someone's taking or is it

18:15that no one takes actions or visits.

18:16Yeah.

18:16Right.

18:17So really, really interesting.

18:18Oh or or it I'm and I'm hoping it also is the navigation, like how do they get into my server?

18:23Yeah.

18:23How do you then navigate around and how do you leave?

18:25Yeah.

18:25So absolutely.

18:27I think it's gonna be good.

18:27I think it's gonna be good.

18:28Yeah.

18:28Right?

18:29DBT connection.

18:30This I did not miss.

18:31So um very simple.

18:32This has been talked around a lot of time.

18:34I think we saw a bit of it uh with the announcement with DBT towards the end of last year that we this was sort of stated as an example of something that would come out.

18:43Essentially being able to

18:45Connect to a data source based off a D V DBT model.

18:47Very straightforward.

18:48However, I think this would behave more like a cube because you wouldn't be able to change the model inside a tableau.

18:53Powered by Tableau Bridge.

18:58What does that mean?

18:59I don't know.

19:00What is why do you need Tableau Bridge?

19:01D Bt is a cloud product.

19:02Why would you be needing Tableau Bridge in Tableau Cloud to connect to D Bt?

19:07Like what local machine are you gonna use to access I d I didn't understand that.

19:12That threw me.

19:12That threw me.

19:13I I like I don't understand why you

19:15need to have like most cloud customers that I'm familiar with with DBT, have DBT Enterprise.

19:20Maybe the context here is you might have DBT core.

19:23So therefore you do need to have What about D Bt next?

19:26I have no idea what that is.

19:28Oh sorry, good sorry

19:30That flew right over my head.

19:37Oh my word.

19:38Apologies, apologies.

19:40Um so no, DBT Core is essentially the the let's say the open

19:45source version of DBT.

19:46So maybe if you have DBT core that works.

19:49But they were showing a demo of DBT cloud.

19:52So I have no idea.

19:53I have no idea.

19:54Right.

19:54Ask to your local dev.

19:55Um publish data sources in Tableau semantics.

19:58So this is the idea of the case.

19:59Zero copy.

20:00Yeah, being able to take the published from Tableau Cloud into Data Cloud through to Tableau Semantics.

20:06That's a step they kinda missed.

20:07Data cloud is needed to get this to work with zero copy.

20:10And that's just a basically a connector between data cloud and um necessary.

20:15And Tableau Cloud.

20:16And then this was a fan favorite, Tableau in Google Sheets and Tableau in Google Slides through uh Google Workspace, Tableau.

20:23Connector.

20:23Plugin.

20:24Yeah, plugin, yeah.

20:25Very cool.

20:26Very nice.

20:26Um I think that's all I've got to say on it.

20:28It's like cool.

20:30Google Sheets thing is nice because I think sometimes it speaks to this idea that some some organizations are Google optimized and what people want to do is to be able to take something into Google Sheets to then share that with the business.

20:39What I hope it does is it updates the asset as well.

20:42So if people keep coming into the Google Sheets.

20:45Well not even just that.

20:46Like any time someone goes to it it updates the data, right?

20:48And you and you see an updated data.

20:54I'm gonna introduce you to a dev I spoke to yesterday on this.

20:56Yeah.

20:57Um You think it's a one time copy?

20:58Like a one time connection?

21:00I think it's all refresh.

21:01You have to I so like if you think about this this data source, I think it's actually a fork off the data source.

21:06It's a fork off it and you don't even put it back in.

21:09But what about getting new updates from the data source?

21:13Yeah, that happens.

21:13Yeah.

21:14Because it's a fork, right?

21:15The the you have the source and then you're forking off and then adding in any.

21:18But the minute you change it the minute you change it, it's no longer representing the main data source.

21:23And you can't change it again?

21:28Fine, fine, fine, fine.

21:29Interesting.

21:30I think yeah, that there's a lot of detail there.

21:32Yeah, we have to go through.

21:33Good.

21:33Then we went into DBT not DBT Labs, Devs on stage.

21:37Man, I'm at DBT.

21:38It's been a long time.

21:41It's only midday.

21:43Um so the first

21:45devs on stage labs we saw was something called authoring um extension so this is essentially an extension that allows developers to build charts on your behalf.

21:54But basically the way where when you hear extension it's a bit confusing because it can either mean it's something to

22:00do with charts or is it something to do with data sources.

22:02In this case it's to do with charts.

22:04You've had um Viz extensions which allows you to build Vizes but auto extensions allow you to control the way Tableau builds sheets and it does it for you.

22:13It drives the interface.

22:14So

22:15Dragging and dropping.

22:16We saw a demo from various companies Biz3, Lad Dataviz, Infotopics and Information Lab.

22:22And all four of them basically showed a demo of things you could do.

22:26Just to give an idea, one was using OpenAI to build

22:30charts and do some exploratory data analysis.

22:32The data vis was doing a layout generator to basically generate a nice themed style.

22:37The other one is infotopics handle automated locate localization as well as um translating.

22:44As translating

22:45And then I think Infolab had the um what was the word?

22:48Automated walkthroughs automated walkthroughs with pop ups and videos.

22:51And I think that's a very nice touch.

22:53Very very good customer focused.

22:55And I think the the money shot was when everything each extension talked to each other.

23:00Correct.

23:00When they were all in the one one thing and one thing updated, the other extension which shows you the modularity.

23:06The extension has a role to play and they all feed into that big picture, right?

23:15that's the most powerful way forward, right?

23:16Which is like um all these APIs, we're just letting let people go build it.

23:22Go go outside, build cool stuff, bring it back in.

23:25Yeah.

23:25Right?

23:25Like it shouldn't be you waiting for a company to go do it.

23:29You go do it.

23:30And then we'll take the best bits and learn from what you're doing, right?

23:32Yeah.

23:32Like the community is the innovator here.

23:34Yeah, exactly.

23:36Exactly.

23:36Same same thing as before.

23:38It will likely get paywalled, but the good stuff, right?

23:41Yeah.

23:42Yep.

23:42Alright, two more.

23:43Tableau pulse research agent.

23:44So this

23:45This is um the best way to describe this is the reasoning that we've seen in AI models but applied to Tableau Pulse, where you could ask it a question and it went off to the internet and got some data sources, brought that up, did some analysis and it was basically trying to get some external

24:00context to the world to explain activity internal to your organization.

24:04Um and this this is just the reasoning capability.

24:07Now the thing about this is that it didn't kind of arrive at a conclusion, it arrived at some possible places for you to go, but you then had to go to it's a sidecar, right?

24:15I I always I always see L L Ms and ch like G P T models as like it's it's like Jarvis in Iron Man.

24:22Well let's continue Marvel right It's Jarvis in Iron Man helping you like get to the arts uh doing the bits that you don't know so you can like Yeah.

24:30It's it's helping you think.

24:31I think such uh important part of knowledge work is the space to think and consider and let your brain create the neural pathways.

24:39Yeah.

24:39Yeah.

24:40If you're just getting answers it creates brain world.

24:42Yeah, exactly, exactly.

24:43Um I'd have to think

24:45This feature was a little bit um it was hard to get the value of it because I think a lot of the audience of Pulse is not actually that audience, it's sort of the executive um who's getting this, the leadership of management who are getting this thing.

24:57So not really big for data analysts, but I think worthwhile.

25:00capability for the people who are using that feature daily and the very last one is uh Tableau sketch.

25:05Were you laughing?

25:07I got this so wrong.

25:10Right.

25:11So I speak So for context, okay.

25:13Sorry, I have to I have to explain the context here.

25:15So when they mentioned the name Tableau Sketch, Ravi, literally no joke, rabbed his hands like, here we go.

25:22And I went over to him and said, Ravi, calm down.

25:24It's not what you think.

25:26Go for it.

25:28I've spent the last year complaining about the names of things.

25:32This is why names matter.

25:33Yeah.

25:33Because people will jump to conclusions that it's a fantastic new wireframing tool that allows you to sketch it.

25:41And then it would apply the template.

25:43Oh, that's magical.

25:44Yeah.

25:45And the only reason I had that contest is because I'd seen a preview of those on stage, so I knew what it was gonna be.

25:50But to be fair, I would have thought exactly the same thing.

25:52If you'd said tableau sketch or tableau anything

25:55with drawing what I would have assumed is take my ExcaliDraw and turn it into an actual finished visualization.

26:01Excalibur draw commonly quite likely with uh some of the other altering capabilities, so it's a bit of a stretch actually to combine these two.

26:09But

26:10What this feature was instead is was the ability to look at a data set, very noisy data set, and draw the shape of the data that you wanted to find to to do some very specific analysis and you were able to describe it just using a line chart and

26:25some annotations exactly and that was used to basically go and find the data and it worked.

26:30It showed the sort of general shape being found across multiple things and it had some variance to it.

26:35It was finding things with like a reasonable spread so it wasn't just like across a mark.

26:40Yeah, doing a best fit.

26:41It it really was a good visual way to go and find stuff.

26:44Um it's also I think a really interesting interface element because that's really ultimately what it was.

26:50Imagine browsing your data but describing what you're looking for visually and being able to take

26:55about visual analysis, not just to analysis but to asking questions as well.

27:00I think my problem with it it's a very cool pie little piece and it's a really good use of uh again agents.

27:05Yeah.

27:06An agent is doing this for you.

27:07This is why we care about agents.

27:10But in practice, how often will I ever do this?

27:14And that's the problem.

27:15You're not gonna sort of go all the way into the chat window, open it up, ask the questions and then see the pop-up to that says sketch it, then you pop it out and then it's a window.

27:23It's it's so buried.

27:24It needed to

27:25be um like a wify mode.

27:28No it just needs it to be a persistent modal.

27:30Yeah.

27:30Um that's sitting in the top right corner, a bit like Shaming.

27:32You click it, it opens, you draw the shape of what you want and out it goes, right?

27:36Maybe it lives in explained data.

27:37Yeah.

27:37Maybe, maybe.

27:38That's a good place for it to live.

27:40But the There stuff you can do with it.

27:42I do agree.

27:43If it was if it was that but instead you drew the chart and you said uh sales, profit, draw a line, boom, boom, boom, or you draw a scatter plot, boom, boom, boom.

27:52Imagine actual feature there.

27:54I would have stormed

27:55the stage.

27:58It would take wireframing to another level.

28:00It would really sort of validate that as a process.

28:02I was saying this yesterday to someone like um a d in 2016 Adobe did a keyno where they show

28:10showed an incredible dashboarding tool.

28:13Yeah, I know the tool you're talking about.

28:15I actually had emailed the product manager for that tool.

28:19He didn't reply to me unfortunately, being like, Can I please, can I please play with this?

28:24I think it was supposed to be

28:25for Adobe Illustrator.

28:27It's infographic creator.

28:28Yeah.

28:29It's data driven infographics and always seek.

28:31And you could animate it as well.

28:33So it's very cool.

28:34But they never shipped it unfortunately.

28:36I I meshed the product manager and never got back to me.

28:39So I'm uncertain.

28:40and they never shipped it.

28:41But really cool kind of innovation.

28:42So yeah this is again that that that's the thing I'd love to see.

28:46Again, you come out of devs on stage and you're like awesome more.

28:50Exactly.

28:51Now the big question is when will these get shipped?

28:54It's all good Sharon.

28:55stuff.

28:56My hunch is not a lot of things.

28:58None of these things had release dates.

29:00Correct.

29:00And my finger in the air is about a third of these we might see over the course of the next year.

29:05I think about half of these we might never see.

29:08I genuinely do.

29:10I think you'll see everything but the runway they've got is m May next year when TC is.

29:17Do you know why I don't think we'll see all of it?

29:19Go on.

29:19Everything's Tableau next.

29:21And I just do not see them.

29:23Even all of these demos were great.

29:25The difference between getting them from where they are behind the scenes builds to being shipped and stable in the product.

29:31What if they're done?

29:33Interesting.

29:34Then why did we see dates?

29:37Last year we saw dates, we didn't this year.

29:39So

29:40It's okay.

29:41It's okay.

29:42I think that the the the thing I'll say is you can we we spoke to Southard's um after after the keynote, right?

29:49South after the keynote, yeah?

29:50Yeah, so when we spoke to Ryan and Southard, like we we gave the feedback of like yesterday was

29:55the clearest vision.

29:56Yeah.

29:57And it was very clear the direction of travel.

30:00Yeah.

30:00And something made a really good point, which was, last year you saw the vision.

30:04Yeah.

30:04This year you see the like You need to see the execution.

30:07Execution, exactly.

30:08Yeah.

30:08And that's what we're seeing.

30:09Yeah.

30:10We're starting to see it.

30:11I'm not sure we've seen it all.

30:13And I think you have to reserve judgment when you see it's like a football team.

30:17You can have a great striker, but if the whole team isn't good, you're not going to the game, right?

30:22Yeah.

30:22And so what we've seen are the striker

30:25the midfield, now we need to see the rest of the team.

30:29How do we gel over thirty eight games?

30:31Exactly, exactly.

30:31And that that we've not seen.

30:33So but but reserve judgment.

30:35But based on what we said about the the MCU multiverse.

30:38Yeah.

30:38We're converging.

30:40We've got agents doing s all sorts between desktop, server, prep, next.

30:46It's it's it's hold that thought, hold that thought.

30:48We're gonna stop this video.

30:49We're gonna explain this whole MCU concept very quickly and we're gonna do that as a separate video.

30:55I do think it needs its own space and it's it's maybe definitely a shorter thing, not linked to devs on the stage.

30:59But just to close this off, devs on stage, out of ten.

31:03Seven or eight.

31:04Alright, seven or eight.

31:05I would have said seven and a half, but we're pretty much aligned.

31:10Definitely needed after yesterday's keynote where we didn't see much of Tableau call.

31:13So a really good touch.

31:15Alright, nice.

31:16Let's go.

31:16Stop it there.

31:17Cheers.