Unlocking Tableau's Potential: Supertables in Tableau 2024.2 by Infotopics with Merljin Buit

0:00As a Tableau developer, there is no doubt you'll get asked to build a table inside of Tableau.

0:05And most users ask for very simple tables.

0:08They'll ask you to do pivot tables.

0:10They'll even maybe get you to build something more complex.

0:12with a hierarchy but eventually every single user who's familiar with excel or other tools out there on the internet will ask you for some more advanced capabilities with tables and unfortunately tableau doesn't let you do those things at least not not yet

0:27In this video, I'm talking to Merlin from Infotopics and he showed me a product called Super Tables.

0:33It essentially solves that problem.

0:35It's a more enhanced version.

0:37of the table capability that you get in Tableau.

0:40It behaves like a pretty modern web excel slash spreadsheet tool and

0:45It's also got some really incredible hooks into the Tableau ecosystem and he spent essentially an hour walking me through how it works.

0:52Now the reason I'm doing this is because 24.

0:552 has literally just come out hours before I recorded this video

0:58We had the discussion last week, but this is a pivotal release because it actually opens up the platform to so many more ideas

1:06and ways of working.

1:07And so I wanted to spend time talking to Merlin about that exact topic.

1:11We talked about AI, we talked about what Salesforce can do to help developers.

1:15Grow on the platform.

1:16But more importantly, I also asked them, hey, is Tableau just handing over the keys to developers to do the things that it doesn't want to do?

1:23I went there with all of those topics and we covered everything in detail.

1:26At the bottom of this video you'll find some time snaps calling out each of those moments throughout the course of this video.

1:31So as ever, get stuck in, enjoy the video, and let me know what you'd like to know about 24.

1:362.

1:37Merlin, how you doing?

1:38Good?

1:39Good?

1:39Are you?

1:40Good.

1:41Yeah, I've I've been I've been super excited to talk to you because uh um there's some exciting stuff going on with Tableau and I think your name keeps coming up.

1:48Last week I was speaking to Tristan and he mentioned

1:51something that you'd built a long time ago that inspired him and I was like perfect I'm talking to Merlin this week.

1:57So yeah, really good to have you on board.

1:59Um where where are you coming to us from?

2:01Are you in the Netherlands at the moment?

2:03I'm from the Netherlands, yes.

2:05Good, good, good.

2:06So y let's let's maybe um let's maybe give people some context.

2:10We were talking about football earlier on.

2:11It's the Euros going on at the moment and we're both

2:14disappointed with our teams.

2:16I'm obviously disappointed with England, you Netherlands.

2:18But um because this is about Tableau, let's maybe give people some context about like your experience and your history with Tableau because I think a lot of people watching

2:27might not know.

2:27So yeah, um what was the first version of Tableau, I guess?

2:31Uh version seven dot what was it?

2:33Three, I think.

2:34Ah okay.

2:35Same as version.

2:36Love it.

2:36Love it.

2:37Yeah it's a long time ago.

2:38I can still remember.

2:40Yeah.

2:40Yeah, yeah, yeah, yeah.

2:41Yeah.

2:41I started using it at the police department where I did a yeah uh research about what was the best BI tool back then.

2:49Right.

2:50Only one one partner in the Netherlands at that moment.

2:53One Tableau partner and uh had to like figure out Tableau or ClickView or

2:58What was it again?

2:59I think MicroStrategy, Kognos even?

3:01Like these kinds of tools.

3:02Like what was the best?

3:03BobJ.

3:04Yeah, exactly.

3:05Yeah, that's how we got to know Tableau at least.

3:07Yeah, but for everyone else, Bob J is a nickname for business object

3:11So if if if you're not sure what I was talking about, but there you go.

3:15Yeah.

3:17Good stuff.

3:18And and then sort of h how did you get to sort of

3:22like the interests you have today, right?

3:24'Cause it's that's obviously like um we're talking ten years at this point.

3:27So like the thing I've always asked selfishly is how do you keep passion with Tableau and how do you keep yourself sort of

3:34dialed into what's moving, why not another tool?

3:37Why not try all the new things that are coming out?

3:39Um well it started actually at the police department where I

3:43I was the police was using Cognos, Cognos, and they were showing me how to build a table with huge queries and it took such a long time.

3:52And I got to

3:54uh play with tableau where I could drag and drop a whole dashboard based on I think what's what it called some kind of crime thing and there was the whole department standing behind me and I was just like let's drag and just see okay what is then

4:06theft and just do something amount and over time and they were completely blown away by that.

4:12And I was also blown away by how easy it was to just

4:15Literally drag and drop uh data on your canvas.

4:18And that was the moment I fell in love and and yeah, I didn't find it in a different tool until today actually.

4:23That that ease of use

4:25Yeah.

4:26Actually that's quite it's quite an important um thing today.

4:30I was seeing someone posted a meme about um

4:34Power BI versus Tableau and they had uh Power BI as Drake and um they had uh uh Kendrick Lamar as Tableau and and uh it's always interesting

4:47Because one of the comments that there was, oh, if you search for jobs on LinkedIn, like Power BI is like three times Tableau.

4:55And I was my response to that, I actually responded to the comment uh on LinkedIn and I said

5:00Isn't that a red flag, right?

5:01Isn't isn't the fact that you need a job?

5:04So many people doing that job, a red flag about ease of use.

5:11Oh yes.

5:12I I have a course on LinkedIn and the people who do the course are not they're not data analysts.

5:18They want to know about Tableau because they have to use it to consume data.

5:21And so

5:22in a good way, you shouldn't need a dedicated job to be a Tableau developer.

5:27Everyone should be using Tableau.

5:29Totally agree.

5:29I think it's a it's a it's a great point.

5:31Like we have other jobs beyond just you know sifting through data.

5:34So yeah

5:35Yeah, totally agree.

5:36Yeah.

5:37Interesting, interesting but it's the small things like like the double click, right?

5:40If you double click a measure, then double click a dimension, it's done.

5:43You have a bar chart.

5:44Yeah.

5:44It's just these small things really yeah.

5:46Makes us reason.

5:48There's a lot of those in Tableau where you know there's a lot of thought.

5:51And I think, you know, if you go use Power B, I'm sure there's something there as well.

5:55These aren't unique to Tableau, but

5:57I think it's the ability to stitch these experiences into a cohesive journey that makes them sticky.

6:03Like when you when you pick up your favorite piece of technology, when you pick up your favorite music

6:07There is something about those things all the way through the experience that keeps you kind of wanting there.

6:12Like every every even the worst design product and have a good element.

6:17but the best design products have good things all the way through.

6:20So I think that's a super important part of um uh part of the experience.

6:24Um and so if we if we go back to Tableau, um like what do you do today?

6:28Like w what's your role and and and yeah

6:31I guess we can maybe slowly move into like, you know, what are you here to show people?

6:34Because I was I was keen to get you on to show us more stuff.

6:36So yeah, over to you on that.

6:38Sure.

6:39Um so today I am the CTO of Infotopics at For Tableau

6:44Still very technical on a daily basis.

6:47But I also lead the development team and a bit of the marketing team.

6:51But my main passion still is develop and build things, like innovate.

6:57And that's what I try to do for as many hours a day as possible.

7:03So I'm still heavy into the code actually.

7:06Okay, okay.

7:06And how long have you been doing that role?

7:08Um just for context.

7:09How long have you been in CTR?

7:11Well it started f five years ago just with one other death

7:15Um we grew a little bit, grew a little bit.

7:17And I think the last three years I have this role.

7:21First with like two, one, one developer as a co-developer, and now I'm here with like

7:26Ten developers plus.

7:28Yeah.

7:29But they're all different.

7:30Like I've all developed I've even had a developer of 17 years old.

7:33He's way smarter than I am.

7:34Yeah, he's yeah

7:36The amount of wisdom he has, I'm sure, an experience it brings to I you know, when you have someone with that much wisdom, it brings everyone up to a level because they've seen things and they've they've they've tried things in

7:46in places we've not tried.

7:47And they've also remember a world maybe where these technologies weren't as good.

7:50So uh they have a more genuine sense of how to do things the hard way as well, which is valuable when when things get tough, right?

7:56And just going back to, you know, your experience as CTO

8:00You've been CTO, I think, at a very pivotal point in Tableau's journey because you've probably seen Tableau open up the platform a lot more, right?

8:08And sort of

8:09uh creating a space for you to have ten developers in the first place, right?

8:13Like ten like ten three years ago maybe didn't you didn't think you'd need ten developers now.

8:19um the story is different right?

8:21Totally, totally.

8:22Yeah.

8:22It's the um I I st I my my first presentation I had a uh I said okay Tableau is this toolbox just a small toolbox and I said

8:30And you have like a saw and hammer and that's Dableau.

8:33But what if you can have like this big massive uh um construction with all the drawers?

8:38And I said the drawers are the APIs.

8:41And inside are all the applications from partners and other things you can't imagine and you can build for yourselves.

8:47Um and um I think first up was the JavaScript API

8:51Now embedding API.

8:52That's what I used a lot to do things outside the dashboard.

8:56Actually also like dashboard extensions, but then outside the dashboard.

8:59It was a quite kind of a similar API.

9:02I used the web page element a lot and then with parameters, which was super hacky, but it was some kind of interactions with outside applications.

9:11Um and then they came and then it it went quite quickly actually.

9:14The REST API of course was already there for a long time, but

9:18Yeah, that it's a different use case.

9:20I always found it the most interesting to be as much as possible to the front-end part.

9:26Like the dashboard extensions are like to the end user.

9:29That's where my yeah, where my passion is basically what I love the most

9:33So that's also why I tried to be with JavaScript API or the web page element always inside the dashboard.

9:38Yeah, and now then in 2017 I think they announced the

9:42Extensions API and that was the moment like okay, I think this is the thing.

9:47This is it, yes, this is it.

9:49Yeah, and I still remember yeah I I the the first extension uh I made together with Challeng, the the death

9:56I have to mention he's uh he's a legend.

9:58Um we made show me more.

10:00Um that was the first extension and um with a Senke diagram that was the first visual you could make.

10:05And I still remember I sent it internally to Tableau and I said, Okay, don't spread it yet, it's gonna be like a thing we want to sell or something, we don't know yet how what.

10:13And then I saw a tweet from Francois Einstadt, I think it was in Australia or New Zealand, he was already presenting, like you can make a Sanky in Tableau.

10:20Bam on the screen.

10:21And I laughed that because it was like, okay, this is really a thing.

10:24It's really big.

10:25Yeah.

10:25Yeah.

10:26Yeah.

10:26Yeah.

10:26Yeah.

10:26Yeah.

10:27Like I'd I made a video about Viz extensions last year because I got excited with the the test they were doing.

10:34And the video I made was about Sankey.

10:36It got 5,000 views in two days.

10:39And like that is the most amount of like instant traffic I've got for any video, any video whatsoever.

10:46And it broke me a little bit because I was like, oh, but it's not out yet.

10:49Like now five thousand people looking for it last year.

10:53And now, like the the the most common comment on that thing is when is this out?

10:58When is this out?

10:59And then

10:59I kind of played into the the stereotype about Tableau cause Tableau does have this reputation of showing us stuff but never releasing it.

11:06And so like ten of the

11:0710 30 of the comments are like, oh Tableau showing stuff again that you can't use.

11:11And like now finally, it's only like 7th of June, almost a year later

11:15It's finally out in public.

11:16So anyway.

11:17Yeah, definitely.

11:18Well you could already do it with the dashboard extensions, of course, the the sneaky diagram and the others, but this is like a whole different user experience.

11:24It's a much deeper experience, right?

11:26Because your Tableau's giving you a space inside of its platform, whereas with with dashboard extensions, it was letting you sit alongside the the platform, right?

11:35So completely

11:36It's a completely different paradigm, I think.

11:37And and that's an important distinction because I think it's a change in mindset from Tableau about how it's going to build the product going forward, but then also

11:46um the role it wants developers to play, right?

11:49Very clearly not giving you the keys, but giving you the runway to do the things it it it can't.

11:55But at the same time giving them the freedom to take the platform in a new direction and to let more people be sort of part of that shared vision.

12:02So yeah.

12:03I've always been pushing for uh since I actually talked to people from Tableau

12:08I think since my first conference actually I've been pushing for opening up the platform.

12:13Because yeah, having like your own developers is nice, but having like thousands of potential developers outside the company doing things you can't imagine is like

12:22That's the dream.

12:23Yeah.

12:24It is.

12:25It is.

12:25Let me put some some some a question back to you, though.

12:28I get a lot and I hear a lot about that specific point, which is um

12:32You know, is is Tableau being lazy?

12:34Shouldn't they be building this stuff nately?

12:36Shouldn't they be making it easier so, you know, people don't have to buy products th through you?

12:41And I think

12:42Uh last week Tristan showed something that I think was super interesting.

12:45He showed a piece of innovation that Tableau haven't done, and I was like, yes, that's that's an example of why devs exist.

12:51But from your perspective

12:52How would you sort of respond to that um sort of critique, I guess, that people s sometimes level?

12:58Well, we we get a question also, like why didn't Tableau do this?

13:01But I think the answer is the same for the iPhone.

13:05When they announced the App Store, that was the moment the thing went, the world changed basically.

13:11And you can do everything you want.

13:13And you cannot expect one company to build

13:15all of those apps.

13:17But also Tableau will probably never like specialize into a finance topic because they have to be like a general analytics platform.

13:26You will never go heavy into a PL

13:29Customize statement exactly following the rules.

13:33You will need to have partners who are able to who are specialized in that

13:36Yeah.

13:37Yeah.

13:37Yeah.

13:37Exactly.

13:38Exactly.

13:38I think it's super it's a super important point.

13:41Focus is actually why you need uh extensions.

13:44It's why you need developers because you can go deep into niches that customers want you to go into.

13:50And actually those customers are happier, right?

13:52Like if you can give a a an a a a table that meets all the PNL requirements or does all the filtering in the in a nice way, then uh people are happy.

14:00And we see that request.

14:01Tableau's great from getting you to zero to sixty, but then you get these really tough questions about that sixty, seventy, ninety, eighty percent, right?

14:09And then

14:09Uh you start doing hacks at eighty percent and people are like, Oh, you can just do this and like you get to ninety-five percent, it goes to the CEO and the CEO says

14:18But can it have rounded edges and you're like ah okay I'm done Yeah exactly yeah Well yeah I will Yeah I think that's um that's fair and I think if you ha want to build everything in the tool you're also gonna be building a monster probably.

14:33Yeah, exactly.

14:34True.

14:35True.

14:35And you focus on the core.

14:37Yeah.

14:37Yeah.

14:38Exactly.

14:39So anyway, I cut you short.

14:40You were telling us about your your role as CTO, Infotopics.

14:47Oh right.

14:48Um well what do you want to hear?

14:50What do you want to know?

14:51I want to hear about what you built, what what was shown at uh Tab.

14:54Oh of course, and and and and let's

14:56Let's dig into the software.

14:58I took you off back, so you Okay, okay.

15:00Yeah, so um yeah, I'm actually very proud of this, but Tableau showed um

15:04super tables during the Tabla Keynote together with Power KPIs.

15:08And they showed a new capability we are adding to super tables, which is the ability to

15:14Actually, like you're used to from Excel, type into the table.

15:18Just type anything that you want.

15:19You can add a new column for a comma, for example.

15:21You can add a new column for a budget.

15:23You can actually edit existing cells.

15:27And all that information will be stored into a separate table, which you can then just join back to your primary data source data source, and you can you have basically your primary data source, your

15:36Updated information and you can do all kinds of new things with it.

15:40Incredibly powerful.

15:41Many customers asked for that for a long time.

15:44We already have Ripex Stream, of course, doing the Ripex stuff, but doing things inside Super Tables makes it like

15:49It's like Nextel, it's like it really like Excel.

15:52Well, who doesn't love Excel, right?

15:54Um like Excel.

15:55I think there's a company that competes with Tableau that like to call themselves uh spreadsheets uh

16:01We'll come back to that in a second, but yeah, you I'm sure you know who I'm talking about.

16:04Yes, I know.

16:05Yeah, yeah, yeah, definitely.

16:06Well and and what uh what we uh we thought like hey if we have super tables um and we give the give it like write back capabilities, we you get like a whole new product.

16:16And that's what was being shown on the on the keynote.

16:19Super cool.

16:20But also very short and a very short demo.

16:23Not really.

16:24Yeah, yeah.

16:25And it also went, there was something on the screen not going well, so you saw of course the demo on the on the

16:30Watch along.

16:31Um so I think it's cool to show death here and maybe go a little bit deeper into what is actually doing, what you can do with it

16:38And then last week we have been doing a hackathon with the team, thinking about okay, what do we want to go next for as a company?

16:45And we have added AI capabilities to this piece of software.

16:48Nice, nice.

16:49Um, which is gonna yeah

16:51Give you like as a dashboard designer, yeah, but also as a dashboard viewer, giving you new capabilities you didn't know you could do in a tablet dashboard.

17:00Yeah.

17:00I think that's gonna be a uh that's what I want to show to you.

17:03Yeah, yeah, amazing, amazing.

17:05We'll put a pin in the AI piece until after you've shown stuff because I think

17:08AI is such a big topic.

17:10People people are getting so passionate about it.

17:12But um let's see what you've built and then we will uh we'll we'll touch on the sort of chat chat afterwards.

17:17Um

17:18So what we have here, this is a super tables table as a this extension.

17:24And we have a region, a province, and a city, and like two columns.

17:31This is all this is for my demo in a minute.

17:33So this is this is just a flat table.

17:35It's actually sample superstore data, but I just renamed a bunch of things.

17:38Okay.

17:39And this is this is already um

17:41Yeah, this is super table.

17:43So maybe give you a little introduction to it.

17:45Yeah, go for it.

17:46It's basically like the ultimate table for Tableau where you can easily group columns by just tracking it on the group bar.

17:53You can expand it like this.

17:55You can uh inline filter, so you can for example just search for Los Angeles and you get Los Angeles.

18:01Um

18:02You can also remove columns by just dragging them off the canvas or you can like pull them away or reorder easily.

18:09There's also an option to pivot, so you can just hit the pivot icon and you can make your own pivot tables just like you're used to from Excel.

18:16Yeah.

18:16And well it it has a lot of capabilities that people love from a table, like just copying some cells, uh paste it over, for example

18:24Things like that.

18:25But a highly requested feature is the ability to write in write basically like an Excel sheet, like type into this.

18:32So what we have been adding is the writeback capabilities right there.

18:36And with this, you can set up a writeback connection with our writeback extreme application.

18:43And all the magic is happening over there.

18:46Meaning, if I'm going to type in this table, information is being sent to our product right extreme.

18:54That's handling that, saving it to a database.

18:57of your choosing of course and then we pass it back to the super table and show it in the super tables.

19:03So I'm just going to show you how that looks.

19:06So what we can for example do in this table is add a column

19:09For example for comments.

19:11Let's call it like remarks.

19:14And it's a single line text.

19:15And we can just edit.

19:16It will reload and we get a new column in our table right there.

19:20And I can actually just start typing in this

19:23And this is all real, so I can just say uh um this is a comment about New York.

19:29Just hit enter, and it's done.

19:31This information is now written back to a database.

19:33Nice.

19:34Um

19:37F for pla peop people who want to do feedback.

19:39Like um I think insurance is a good example.

19:42Need to add notes on a claim or um if you're going through an IT support ticket, you need to add notes as you go through.

19:47You could do this in the dashboard, right?

19:49Exactly.

19:49Yeah

19:50And um so this is now in the sheet, but of course if you add it to a dashboard, it's part of your dashboard.

19:55The next thing we will add is right-click and the ability to just add a note here

19:59Just like you used to from an Excel sheet.

20:01That's not here yet, but that will also be there.

20:03It's like a comment, yeah.

20:05Exactly.

20:05And you can also um do things like

20:08Bulk updates, which is also an Excel feature.

20:10So if you say okay, we want this comment for all the East Region cities.

20:15Let's let's the only thing you have to do is just select all the records we want to update

20:20Ah nice.

20:21And then maybe type something like a comment, hit enter and it writes it back for all these records

20:28I just want to show you that it's actually doing this.

20:30So what I did already is I create a cross-data source joint with my superstore and this table.

20:36It's called WBE with AI.

20:38That's my writeback table on the left here.

20:40And whenever I hit this blue refresh button here, you will see that it refreshes my data source.

20:46We will see a new comment column popping up called remarks, right there.

20:50Okay.

20:51Just refreshing it.

20:52Here we go.

20:53Remarks.

20:53There it is, yeah.

20:54I can create a new worksheet and just add it to my worksheet.

20:58Maybe add cities to it.

21:00And I already have it.

21:01You can see all of them are there.

21:02Nice.

21:02Yeah, exactly.

21:04That's it.

21:04So it's extremely easy.

21:06Um and I can of course put it on the map or things like that.

21:09Yeah, do whatever you can with Tableau, right?

21:11It's just in the discourse.

21:13But you can go a little bit further.

21:14So it has now one c

21:16common uh one column called remarks but I can also type into uh the the the cells let's just undo this

21:24And that's the demo that was being shown on the Tableau dashboard.

21:26This is not a use case you always want, but I can just type 5000 here.

21:30Yeah, to correct values or something like that.

21:33It could be stock takes, right?

21:34Yeah, yeah, exactly.

21:35And then um

21:36It will refresh the dashboard and what you will see on the left is that it will create that's that column, the number of companies amount.

21:44Here, there it is.

21:45It's a measure.

21:47And it will have a value for New York with 5,000.

21:52And then I can create a calculation saying that hey, if I have uh a value in my

21:56like my WBE with AI table, then use that one, other one use the original data source value.

22:02Right?

22:02So I can create an if-l statement calculation to to um or actually you can do things like do all kinds of things with it

22:10I can also create a for example a budget column if I want.

22:15Just add a budget column here.

22:16And then I can also start typing like a budget.

22:19Things like that.

22:20So it's all six thousand, maybe seven thousand.

22:23Okay, I think add budgets.

22:25Yeah.

22:25It reloads.

22:27Very good.

22:29And this is how easy fast and easy.

22:31Yeah, it's very fast and easy.

22:32Like I've used I'd use like a version of this in the past, but what I'd not appreciate maybe I think my setup was poor because um

22:39I I guess I was sort of one working my way through it, but what I'm noticing as you're using 'cause you're you've obviously got a lot of experience with this

22:47You're pushing it a little bit more than I did 'cause I think I was just going through and slowly adding stuff.

22:52But yeah, you're just working at the pace that people would work at and it's it's native inside of Tableau.

22:58There's no sort of

22:58fun and g funny games or anything.

23:00It's just it's just native capability.

23:02You've got the marks pain, you've got your pills on the left hand side and you can you can kind of

23:08Bring a bit of your own juice with this uh additional table.

23:11And I think an important thing to highlight, you've said it already, but you're not changing the source data.

23:16You

23:17you're adding the changes to a separate table, right?

23:19And that kind of exactly keeps the sanctity of the original data source.

23:23You're not sort of messing it up.

23:24No, no, I'm not touching the real data.

23:26I'm only writing it to a separate table

23:28Yeah.

23:29And then just join it back on a key, basically.

23:31Which in this case is city, but it could also be like an order ID or anything else, what you what your unique key is up to table.

23:37Yeah.

23:38Yep.

23:39And as you can see I just it just reloaded.

23:41It's a VIS extension improvement that needs to be done, but sometimes it reloads.

23:45I just reconnected it, it reloads, you see all the information is still here.

23:49So it it also when you open a super table, look at the

23:53uh data from your worksheet, so what we're using here on the left.

23:57And it also looks in the writeback API and retrieves that data, joins it together and shows it right here.

24:03Right.

24:03Yeah.

24:04Perfect.

24:05So for end users there's no development needed.

24:08There's no uh um yeah, there's no programming out of the box, right?

24:13Exactly.

24:13That's it to point it to the right place

24:15And with that table that gets created, like what does like what does the user need to to do there?

24:21Like what if like ten people are using the dashboard?

24:23How how are those how are those sort of tables how do those tables work?

24:29Um so you mean super tables or with the writing capabilities?

24:32So with the writing capabilities, so you let's say I've got this dashboard, I've added a comment, and then someone else has got a dashboard and they've added a comment.

24:39Are they being appended to one big table of of changes or are there like separate tables that are you know being kept for each each session, if that makes sense?

24:48Yeah, um so this is the d it's it's being released in Q3.

24:52This is one of the things we're de uh designing at the moment is what do what do we want?

24:56Do we want to give them like the end user, right?

24:58We give them a dashboard.

25:00Maybe just go to this dashboard here

25:02We give them a dashboard and do we then want them to be able to like connect to ripack themselves like I'm doing here?

25:10Or do we want this ripe connection to already be set up and they can only use it

25:15Um because so they can only uh for example when they uh open the dashboard type into the remarks column and that's it.

25:22That's the only thing they can do.

25:23Right lockdown a bit.

25:25Yeah.

25:25Yeah, exactly.

25:26And maybe you can even say you cannot uh update existing cod columns, you can only uh add add data to this remarks column, right?

25:34Because there's maybe a little band behind it or something else

25:37Uh yeah, that's what we are designing at the moment.

25:39Um but the great thing is with Raggick Extreme, all these questions have been answered before.

25:44Um we know exactly what to do.

25:48The only thing is like

25:49What does the customer expect from this?

25:51Because this is a this is more like Excel, so but what do they expect?

25:54And um I can also imagine that people would love to have the ability to create their own like

26:01like their own uh like local what would you say like like local okay yeah like a little yeah yeah that makes sense yeah so that you can for example go here and say okay I want to I want to create my own use case and I don't want a remark column I want something else

26:14To store my notes and every time I open my Tableau dashboard on Tableau Cloud, I see my notes over there and not the from the dashboard.

26:22Yeah.

26:23Like a workspace essentially within the workbook, right?

26:25Like uh Yeah

26:26Uh better than the comments or better than uh you know what what people do is they actually export the data and go do the same thing in Excel, right?

26:33Like that's that's why people ask for the download button.

26:36Whereas here you kind of

26:37You're kinda catering for that need in the dashboard and that's actually convenient.

26:40That's the problem with a lot of you know these things.

26:44It's not convenient

26:45Uh and therefore people extract the data to go do it where it's convenient.

26:49So this is a nice solution to that.

26:51Thanks.

26:51Yeah, and I think Super Tables is already helping with this question a lot.

26:55I think one of the unique selling points is that it actually helps you to reduce the amount of exports to Excel because you have the Excel capabilities in your hands inside the Tableau dashboard.

27:06And this is only gonna help with that

27:08Yeah, absolutely.

27:10Yeah.

27:10It's so it's it's it's such a bizarrely simple uh product.

27:15I'm gonna call it a product.

27:16It's a b it's a simple solution and product to very

27:19common problem we've had in Tableau for a long long time, right?

27:23Because people want uh let's call it the user experience they're familiar with from Excel.

27:29But they want it brought into sort of the modern world.

27:32Like people ask for filters, people ask for dynamic search, they you know, write back.

27:37All of these things are are the things that Google have made us familiar with in

27:41Google Docs, right?

27:42Um and so you've kind of brought that experience inside of Tableau here, which is fantastic.

27:47So really nice.

27:48Yeah, thanks.

27:49Good.

27:51C so in terms of um if I talk a bit about like building this extension, like how easy was it once Tableau said, okay, we're gonna have this extension's capability to to sort of build this and and and and and

28:05bring it bring it to life.

28:06I'm not I'm not sort of talking for like uh how how was it built.

28:10It's just more how do you how quickly was it how easy was it to mobilize your team to kind of do this and get to like a V one of the product and sort of how's the journey been up until

28:20I guess 2014 is about to come out, so that's going to be the big release, I guess, to everyone.

28:25Yeah, so the the this product is now almost

28:28five years old.

28:30We we co-developed it in the beginning of a company together with a company which is a tableau user and he asked like can we have like a better table

28:39And I'm actually willing to help fund it so um in return for a free for a lifetime license basically.

28:45So that was the deal we made.

28:48And um yeah it turns out to be uh uh it was a yeah very great investment from that company.

28:55Exactly, and also something you would not expect um not expect

28:59uh directly if you would build this yourself for example.

29:03But um at that point we made it as a dashboard extension and it it was super simple.

29:09It was it was only having the like the minimum viable product

29:12features that customer requested and nothing else.

29:15And if you see how it evolved over five five years, um it became

29:20It's like hands down like a it has so much capabilities you can basically almost do anything with it.

29:26Yeah and um we just for example edited the capabilities to have a

29:31tooltip per column and that tooltip can be an individual customized HTML tooltip per column with buttons inside and images and everything you want.

29:39And then that means that you can if you only have three columns, you can have like a completely different tooltip or column with information about a customer, for example, maybe a location with an image of the location and some information about the location.

29:51and a button to go there on Google Maps.

29:53You can build whole applications with it.

29:56And that's just a recent added feature.

29:58And there's so much more we're doing.

30:00And then the next thing we have we've been working on this year is of course making it a Viz extension.

30:06And essentially this product is a hybrid, so it's a dashboard extension and it's a this extension.

30:11So it's still for us we treat it as one product

30:14And the code is smart enough to figure out like, okay, am I a dashboard extension, am I VIS extension?

30:18I have to behave a bit differently.

30:19Right.

30:20Right.

30:20That makes sense.

30:21Because I was gonna ask about that transition, like because

30:23You know, Viz Extension API stuff has probably only come into like focus in the last seven months, right?

30:28So like yeah, you've you've had you've not porting is maybe the wrong word because I think you're still working in the Tableau ecosystem, but

30:36You've definitely had to adapt the code to be conscious of the environments because a dashboard extension and a viz extension have completely different levels of, let's say, interaction expectations.

30:47Exactly.

30:48Yeah.

30:49And and and you've kind of had to work around that.

30:51Yeah.

30:51Yeah, yeah.

30:52So just for uh an example, the dashboard extension has to basically look at the dashboard and figure out

30:58To get this data somewhere from the dashboard.

31:00So we always say that you have to build a data sheet, which is like a hidden worksheet where we retrieve the data from.

31:06Like a raw data, yeah.

31:08Yeah, so it's basically like in the dashboard, you search for that worksheet and then you get the data from there

31:12But on the this extension level, you don't have a dashboard.

31:15So you're always you're already on the worksheet level.

31:17So you're like one level deeper in the API.

31:20And everything is happening there, but also things like filtering, for example, is working completely different.

31:25Normally we have to say, let's add let's uh um apply a filter to that worksheet on the dashboard.

31:31For example, but it's not possible because we cannot access that.

31:34So you apply it like to yourself and then through a dashboard action that's being deployed to the other worksheets.

31:39So it's like a whole different way of thinking.

31:42Yeah.

31:42Okay.

31:45Oh no, it's ne it's never too technical for this challenge.

31:47Uh like uh anyone who's still listening after thirty-three minutes is definitely in the right place, so don't worry about it.

31:54Um I was I was gonna say

31:56The nice thing about the Viz extension is it can also access the data model, which which I think is richer than maybe the dashboard extension, because the dashboard extension essentially presents like a single table with everything you need

32:10for the Vizic for the dashboard extension.

32:13But the Viz extension has way more access to I guess what's coming behind it of the data model.

32:18And with

32:19with all the new energy going into the data model, um, that's good.

32:22I is that a fair assumption to make?

32:24There's there's more you can do with the Viz extension and the data source than you could with the dashboard extension.

32:30It both it both essentially is a getSummary data async function.

32:34So it's both still the same function.

32:36But what I what I do think is great about the this extension is that we can now use the native tooltip

32:43function for example so we can just ask for a tooltip right and get a tooltip I just and I disabled it here but you can also show you like native tooltips is like a big benefit

32:53In all our products we have to build custom tooltips that look like Tableau tooltips and you would also be able to customize them inside the configuration, but you all had to build that yourself in a configuration plane and how people should do that.

33:07You have to do a lot of parsing and stuff to be able to make it work.

33:10And I can just use the just use the native one.

33:14Yeah.

33:14Yeah.

33:15Yeah.

33:15That's uh that that's actually

33:18A a times over for you, but B, it means the experience feels familiar.

33:22Like 'cause that tool tip it has tooltips are the kind of one place where maybe you want them to look the same across all charts, even if

33:30let's say I'm using your super tables and I'm using I don't know Sankey from somewhere else, right?

33:35Like you want the tool tips to be consistent across those.

33:39Yeah.

33:39Exactly.

33:39Yeah.

33:40And you can also use the keep only button, the execute button, all these kinds of cool things.

33:44Yeah.

33:44Nice, nice, nice, nice.

33:46Excellent.

33:47Good.

33:47So so yeah, like where do you go next with this now?

33:50Like you've you've like I always think like how how much more can you develop tables?

33:54I'm sure there's this

33:55tons of ways I've not thought of and um you probably know all the limitations but yeah where would you go next with this?

34:00Oh man there's so much possible but um what

34:04Imagine you um like parent child tables that you can have a child table if you open a row and for example you have uh um

34:13Yeah, you have a sales manager.

34:14You can open the sales manager and see like his recent deals in a totally different format inside the table.

34:19That's a parent-child relationship.

34:22That's I think very interesting

34:24And of course the capability I'm going to show you in a little bit is also very interesting.

34:29Yeah, but there's we have a whole roadmap full of ideas and things we actually are going to work on.

34:35Yeah, I'll bring up the company we were talking about earlier, Sigma.

34:38Sigma always showcased that parent child sort of nesting.

34:42That's actually one of like their big thing, right?

34:44Like

34:45being able to drill into your data.

34:47It's kind of like a weird analogy 'cause I think that comes from Tableau in the first place.

34:50But anyway, um it's it's it's it it's that sort of ability to explore, but in this tabula format, right?

34:58There's a lot of

34:59There's a lot of uh relationships, uh, opportunities that you can kind of build into your data.

35:04And I think going back to your point about logic and being able to do different things on each column

35:10The way you've set this up suggests that you've got a lot more freedom than, you know, a typical table would actually offer you.

35:17Something like Excel, you couldn't do the kind of things you're doing, which is which is great.

35:21Exactly.

35:21Yeah.

35:22Yeah.

35:22Yeah.

35:22Yeah.

35:23Good.

35:24So, um yeah, well where next, uh more more drill down.

35:28What about um you mentioned AI earlier on.

35:30Where where where is that gonna sort of

35:32come into play.

35:33I'm sort of curious to see how how would AI help this use case as an example.

35:37Okay, good.

35:38Yeah, maybe that's a fun.

35:39So I I I brought up a little uh idea

35:43Yeah, yeah.

35:43So l two weeks ago we had a hackathon with the team.

35:46Um already thinking about this for a longer time, like how can we

35:50jump on the AI hype train.

35:53Not just to jump on it, but actually think okay what kind of value can we bring to the table with our products and when we add AI into it.

36:02And I think we have two extremely good use cases.

36:06And one of that is this one, which I think is extremely fun.

36:12Because uh during and the demo I'm gonna here do here is together with you by the way, you're gonna help me.

36:18Um I'm not gonna do a uh demo I have prepared.

36:21I'm just gonna see what happens if you get me there.

36:23Yeah

36:24We're gonna do it together, right?

36:25I love these kinds of demos.

36:26They're just let's see what happens.

36:28Um but during the keynote there was one demo where they were looking for a new location to go on holiday, right?

36:35Like what's the best location to go on holiday

36:37Yeah.

36:38Question to you is if you have like a list of cities you would love to go to when um on holiday and you you and you want to make a tableau desk for it, you have to figure out uh more columns

36:50And then also figure out the data in those columns, like like the temperature of those cities, but also other interesting things you have to think about.

36:57That's gonna take a lot of time to excel it together or tableau prep it together or whatever how you can do it, right?

37:04Yeah.

37:04And I thought how can we make that process easier and and do it into the Tableau dashboard without leaving it?

37:12by adding ideas or columns of data and then visualize them instantly in your tablet dashboard without leaving the tablet dashboard

37:21So that's what we're gonna try to do.

37:23Exactly.

37:24Yeah.

37:24So what we have here is a and I thought okay, what would be will be a fun topic?

37:28Um so I saw okay, Tableau Conference, we're here with all Tableau people in the uh in the in the video

37:34Um so I thought, well, can we together find the next ultimate tableau conference location in the US?

37:40Maybe it's San Diego

37:41Maybe it's something else.

37:42But I hope we can figure out together where that would be, based on your and my opinion of what is important for a city to host a teleconference.

37:51Yeah.

37:51Right?

37:53Shall we do it?

37:54Yeah, good.

37:54Let's do it.

37:55Yeah.

37:55Good.

37:57Alright.

37:57So what we're gonna do is let me just remo

38:01disable these columns we don't need them is use the new AI feature right here and with this feature we can ask chat gbt

38:10For an additional column of data, and we can ask it like, hey, can you create a column with and add this information to the table?

38:16Right

38:17And that's gonna, I will show you in a second.

38:20Um, what's that gonna do?

38:21Maybe give you like a little demo here what it can do.

38:24So we can say, let's do a um are you into food?

38:28Yeah, I meant to fit.

38:29I love my feet.

38:31Let's do a best food of these cities.

38:34So we we we're gonna find out like what's a good city for uh for for the conference and number one's of course food

38:41And just do it this code.

38:42So what you're gonna do here is you can just ask for give me the most popular dish for these cities

38:51Cool.

38:51Oh, okay.

38:52So reference in city and state is going to be like the column it uses, right?

38:55Yeah, exactly.

38:56And we're we we're doing this with a reason because we want to send as minimum information to

39:00the a the the API as possible because it has a it has a limit.

39:04So we're not going to send the whole table, we're just going to send one column.

39:07True.

39:08And then of course the tokens as well, right?

39:10It's the whole token thing, yeah.

39:11So you uh we already have a new version which can add one or more columns, but this one just sends one column.

39:17Yeah

39:17Let's see what happens.

39:18So technically what's happening?

39:20We're gonna send this uh this table or like this column basically back to uh OpenAI.

39:26It's gonna add a column to it for uh the best foot

39:30And then it's gonna send it back.

39:31And with the RIPEC extreme uh connection, we're gonna write that information into a table and we can start instantly start using it.

39:39Okay?

39:39Let's see what happened.

39:40So it's generating data from OpenEI

39:43I would love to have a Einstein AI API.

39:46If we're talking about new APIs, I would love to have an Einstein AI.

39:48Do it within the tablet ecosystem, yeah.

39:50Exactly.

39:52And look at that.

39:52We got new fruit here.

39:54Wow.

39:55Which is just a column

39:56And the pyramid.

39:59I just saw barbecue and I'm like, I'm right there.

40:02North Carolina?

40:04Oh yeah, so Carolina, sorry.

40:07Um

40:08Where was the Charlotte?

40:10Yeah, we can't get it.

40:12Right, so that's just a very simple uh demo of what you can do.

40:16But you can also, well, you can iterate on this, right?

40:19So you can also say, okay, um, but what's the main ingredient?

40:22Because I don't know all of these, uh I don't know what Tex Max is.

40:26Nobody actually

40:27Maybe it's like what's the um give uh what the main ingredients uh give me the main ingredients um present in this dish

40:39Just check and then we can actually select the best food column with the new column and just add another column to it.

40:51Uh so it's like feature engineering, isn't it?

40:54A little bit in data science where you're kind of adding adding adding quality to your data, but you're also

41:00Doing data prep at the same time, it's kind of what you're doing, right?

41:03Because you you're using source data, but you're adding value to it as well.

41:07Wow, okay.

41:09And the cool thing is whenever I hit refresh dashboard, it will refresh my data source and on the left you will see these two columns appear.

41:16And I I find this like kinda magical because now main ingredient is there and best.

41:21You're building your data source in the dashboard.

41:23It's kind of it's like a very weird concept to get, but it makes sense.

41:27Yeah, right.

41:28I can just go into a new worksheet and build a viz on this and say, like, hey, let's make a uh a best food viz, what appears the most, things like that.

41:35Yeah, yeah, yeah, yeah, yeah

41:37So Christian from US, so if we would like to go to a new city for a teleconference, what do you think is the most important thing for here?

41:45Oh wow.

41:47We already have food, but what other things do you think?

41:50I think Oh, that is a great question.

41:56Um public transport, I think, is pretty key 'cause that can that can that can sort of break your experience of getting around and experiencing the city, especially if there's a lot of people there all at once, right?

42:10So

42:10Las Vegas is good at that.

42:12Um but some cities have definitely you've had to like really plan your journey like your lifts or your Ubers.

42:18I think in New Orleans I was using Lyft and the

42:24Right.

42:24So what what question should we tell uh ask AI for for this?

42:28Um rank the public s transport for this city.

42:32Um

42:33Or should we do something about Uber Lyft?

42:36Yeah, let's let's let's yeah like let's ask ask the availability of Uber or Lyft or something like that.

42:43Let's do something a bit more challenging.

42:46Something like this?

42:47Yeah, that sounds good.

42:49If both is possible, rank high, only one rank, so medium.

42:56Otherwise low.

42:58Yeah.

42:58I'm not sure if this is gonna work.

43:00Just figure it out.

43:01Let's try it.

43:01Let's try it.

43:01Yeah, yeah, yeah, yeah

43:04Cause this is interesting because this is also going to be limited by OpenAI's index of the world, which I think does stop at a certain place, right?

43:10Exactly.

43:10But actually, um

43:13Here we go, let's see.

43:14Where is where is the column?

43:15It's to the right.

43:18Yeah, yeah, yeah, yeah, yeah.

43:19So I want to make a vis about this and just say, oh, we have to refresh the data source.

43:23I forgot that button.

43:25And then just go here and just put um public transport here and then let's do a count of city for example.

43:36Yeah

43:38Nice.

43:40It's um yeah.

43:42And we made a yeah of this.

43:44Nice.

43:45With with uh with with with completely

43:49kind of they're too countless.

43:50Like what's the word?

43:51En like not engineered, but you know what I mean.

43:53We've we've gone in the source of the data.

43:54Like if you had to scrape that manually

43:56You'd still be doing it right now, right?

43:58You'd be looking for like Uber statistics from somewhere.

44:01Take you thirty minutes to find it, you can't find it.

44:03Um it's kind of interesting.

44:05And I think th the question I have back to you then is

44:08Have you seen ChatGPT hallucinating on some of these responses?

44:12So like because because what we're doing here is actually in the dashboard

44:17How do you mitigate against um not necessarily just hallucinations, but how do you mitigate against some of the issues you have with quality around Chat GPT and sort of its responses?

44:27Um is it about sort of the scoping that you showed me or is it actually a bit more nuanced than that?

44:33Well, um so this is the the thing we are uh of course also experiencing experien experimenting with.

44:40One of the things what you can do is give it a temperature.

44:43That's what it's called, I think.

44:45Higher is like you have to be you can be more creative and lower is or it's the other way around, lower is you have to be like more strict to the answer.

44:53Yeah.

44:53That's one of the things.

44:54One of the other things you can, for example, say is like, hey, if you don't know, just don't give an answer and not make up something.

45:01Right.

45:02That's a thing.

45:03But we also today actually I was playing with this as well and I was looking at conference centers for the cities.

45:12Just try trying like is this a question

45:14And one of the cities was Springfield, and that city is actually in the US I think four or five times, if you look at Google Maps.

45:22And it was actually picking the wrong one.

45:23So it's also about your uh your

45:26Your prompt.

45:27I was just saying give me the uh does this city have a conference center?

45:31And then I said Springfield, but I was not saying Springfield in

45:35Well I'm not sure what it was again, but Springfield in this state.

45:38Right.

45:39Right.

45:39And then it gave me the right answer, but previously it was giving me the wrong answer, which is still the right answer.

45:44Prompt engineering as well.

45:45It's prompt engineering.

45:46I think it's gonna be a new skill

45:49People have to learn?

45:49Like how do I really write a good prompt?

45:51I'm also learning it.

45:53So probably here there's also gonna be answers that are a bit hallucinated.

45:59But I tried, for example, if you look at population, airport, conference center names, I cross-checked a lot of those and they all match up.

46:10Yeah.

46:10But the thing is that uh to me, I think this just focuses on where you use this.

46:15And a good example is survey data.

46:17How many people are and doing analysis of survey?

46:20And they have like two paragraphs of text where it's someone's written some qualitative feedback about something and you want to do quick analysis on that, but you need to do it on like, you know, ten thousand survey responses.

46:31This is potentially a really good solution to that because on something like that where it's about summation, like summarizing, ChatGPT is very good at that stuff like where it's not having to get into the world and

46:42find like facts, this you're giving it some of that information, you're telling it to sort of synthesize it into a new format.

46:48Now AI is very good at that and that's oh yes one of its strongest use cases at the moment

46:53Yeah, and that's also how we use it in Rabbit Extreme.

46:56It's the other uh example where we are into implementing AI features.

47:01Where you have to do, for example, summarization.

47:04So you have to write, for example, a quarterly report.

47:06That's what you can do with Rapid Extreme on your dashboard on a specific data point.

47:10And it can help you to analyze a whole chart

47:13And and concise it into like four bullet points.

47:16Put it in the write back application.

47:17You can hit save.

47:18That's one of the new things we I'm also gonna can show you in a minute if you want.

47:22Okay.

47:22What works as well.

47:23And so there's a lot of uh uh yeah locations where you can add this uh and where it really helps you.

47:29Yeah, and this is I think one of the use cases I found like as a Tableau desktop developer, I find this like extremely fun to play with

47:37Right.

47:38And the the devs also have this feature, so you can say create multiple columns.

47:42And it's a bit buggy, but I can for example um

47:46Shall I am I gonna do it?

47:47Um it's probably not gonna work, but you can do things like this.

47:50You can just say give me some interesting columns which helps me to decide where to host the next double conference.

47:56It's a convention center, blah blah blah, and then you do the city column, and then it just generates like five columns with average temperature.

48:02It was something like it has a conference center, yes, no.

48:05uh something about an airport and some other things.

48:08It figures like some random columns you can also be more specific of course.

48:11And it just adds those columns.

48:13You have it to your you have it uh available in your left side if you just start analyzing it

48:19Wow.

48:19That's amazing.

48:21And i you you made a comment about Einstein AI earlier on.

48:24What's the context for that?

48:25What's um is it is it instead of using ChatGPT using something

48:29There's a little bit more trusted in the ecosystem.

48:31Is was that sort of where you're going with that?

48:33Yeah, yeah, definitely.

48:34So um there's a reason why Apple is not launching OpenAI, new AI features in Europe, right?

48:39We're from Europe, we're not getting it

48:41There's also a reason why they're asking like do you want to send this to OpenAI in their iOS 18 update if you want to do something outside the Apple intelligence system?

48:50Right.

48:51Because it's all about trust that and that system using your data.

48:55And I know that one of the things Salesforce talks about is about trust

48:59and their trust layers to make sure that your data is safe and it's it's it it it remains safe.

49:06And I think like if we can talk to the Einstein AI, which is also part of Tableau of course now.

49:11uh in many places and then we can just we can um yeah basically instead of going a different route in tableau we can take the same route which benefits

49:20both of us and it also benefits the customer because it knows okay they're using a safe API and I can use this with with confidence and trust.

49:27Yeah and I think

49:29There's a lot that Einstein is doing to remove sensitive data um from the customer, right?

49:36So you can you can piggyback on that innovation, right?

49:39So

49:40that that that allows for more confidence and sort of a deeper use of this um potentially than than than more.

49:46I'm sure it's coming.

49:47Um

49:48Because if you think about the way Tableau's doing this, they they're gonna open it up eventually.

49:53I I think they want things like Viz extensions.

49:57to be deeply tied into the platform.

49:58So if you can create a use for those tools and the company has those tools and no Tableau announced like a new pricing structure with Tableau Plus.

50:07So

50:07That naturally suggests that they're going to want to see people use those credits and use those interfaces a little bit more.

50:14So it's a really good way of kind of adding value in that sense.

50:18Yeah, exactly.

50:19Yeah.

50:20But I think trust trust is the most important one.

50:23Yes.

50:24Yeah.

50:25Well Tableau Tableau has to earn some new levels of trust because you know as a 20-year-old company

50:34I think for a long time, much of that trust has been built over time.

50:38But with the acquisition of Salesforce, there's potentially this new challenge.

50:42of having to renew some of that trust.

50:44You can't just assume that customers trust the brand anymore because of AI is making everyone question like

50:51uh people's people's values around data and people's perceptions of what a analytics tool should be.

50:57I think all of those things are up for grabs at the moment and people are re-evaluating the the decisions they make uh and the answers they have

51:04Yeah, totally agree with that.

51:06Yeah.

51:06Yeah, yeah, good.

51:08Okay.

51:09Um that's super awesome.

51:11I really love this.

51:11It's very um

51:13Very interesting.

51:14I I guess my only question with um with the AI point, um this is sort of what I was gonna ask earlier, is would s would Salesforce sort of push into AI?

51:24Do you feel like there is um there's mileage in this?

51:29Because the the the the the the challenge I've always got with with not with not with AI specifically but

51:35with any trend like this that catches all industries by surprise, is that everyone has a need, uh has this sense of needing to be talking about this capability.

51:45So everyone pivots to it, everyone runs to it.

51:47And I think there's this discourse within the AI community that actually the la large language models, the LLMs we have today, are actually at their peak.

51:56And it's only with tremendous resources can you get them to the next level.

52:00But even then the benefits are quite small.

52:02And then if that does tran transpire to be true

52:05Actually, you know, people are putting a lot of time and energy into something they think has more runway than it actually does.

52:10So I'd love to get your sort of take on

52:13you know, sales forces move into AI, tableaus move into AI, and then sort of how you think that that might play out over the coming months.

52:20Not necessarily years, no one's a profit, but

52:23You know, we're about to see the fourth wave announced at Dreamforce.

52:27They're talking about new interfaces for Tableau.

52:29How do you think this is all gonna come together?

52:32Yeah.

52:32Big big question I appreciate.

52:34So yeah.

52:36I think I um I always remember um and that's also what's being shown about AI

52:41every time it says video of um Bill Smith, I think, eating spaghetti.

52:47It was like the the one year ago video where it was completely uh um

52:52random and not good and then one year later they showed the same prompt basically and it was like a super good video already

53:00Still some fingers missing or double fingers or whatever.

53:03Yeah.

53:04But it was already good.

53:05So I think if that's one year, um what what about and it's uh the the the whole technology is always like it always it it it

53:13It always goes like this, right?

53:14So if we're already here, but what is the name next year?

53:17It's gonna be like four times as good maybe?

53:21And then we cannot really understand anymore, like is this real or not real or

53:25Um so I think the same is happening here and the same is happening to I think we just touched the the the tip of the iceberg of the iceberg indeed of AI.

53:36in this the the last year basically.

53:39And now we're also going to see more like really useful features and capabilities of it instead of all the gimmicks

53:47uh like the really useful things.

53:49And I think I'm already used, I'm not sure about you, but I've used it already on a daily basis.

53:54Yeah.

53:58Dutchy English.

53:59But it's just it's a simple use case, but it I think it's like a great sentence, but yeah, if you rewrite it a bit then it's

54:06I use a tool called OpenAI Whisper on a daily basis to transcode my to get sorry to get the transcript of my videos.

54:14out of um the video and into subtitles for for YouTube.

54:18And then recently I've started actually uh narrating.

54:22So I just skip my voice recorder, I'd just start talking and I'd record an record an audio clip and what I'm doing is recording my thoughts.

54:29But then I'll put it into OpenIO Whisper and it will generate the transcript of my thoughts as I was talking, very human sense.

54:35But then I'll put it into ChatGPT as a final step and I'll say, right.

54:39These are my thoughts about this topic.

54:41Can you like synthesize this into two paragraphs for an email, right

54:47And it does re really good job.

54:48Again, it's that thing that it's good at is synthesis.

54:51It's not good at it's not good at going to search the facts.

54:54It's not good at being really genuine with thought, but it's very good at

54:58taking something and summarizing it.

55:00That's why it's good.

55:01It's it's sort of a language engine, right?

55:03It it can do that very, very well.

55:05Too many people push it beyond that language area.

55:07And I always think, ah, you kind of

55:10Like I was trying to explain to people, like a large language model is like a big mathematical model for what comes next, right?

55:16That's all it actually is, right?

55:18Like uh think of the world's biggest probability engine

55:22Like and it's just percentages about what comes next.

55:26So if you ask it, hey, as we did here, what's the best food in San Diego?

55:31is gonna go and look at all of its data and it goes, hmm, typically when people talk about the best food in San Diego, fish taco comes next, right?

55:39Like and and and that's how it gets its answer.

55:43I find that.

55:52Yeah.

55:53So there's one one one other use case we have been uh playing with.

55:56We have a tool called the Meal Scheduler.

55:58where you can uh create PDFs or PowerPoints or Excel files from different dashboards on your cloud, combine it and send it at the specific time to people.

56:06And one of the things we we have obviously think like is can AI analyze that image and then give me a summarization of it?

56:13And I I'm imagining like, what if you're blind, for example?

56:16Oh yeah, yeah.

56:17And I cannot re I cannot see that image, but what if I it can just read it for me and then give me the highlights of it, right?

56:25Like what's the latest and it can actually do it

56:27So that's what we also have as a demo with the new OpenAI 4 model, you can give it an image and just say that can you analyze it with a good prompt?

56:34Yeah.

56:35And it gives us gives us like uh um three uh uh highlights uh based on the latest sales data and three like things that went worse.

56:42based on the charts.

56:43It was pretty good.

56:44I'd like okay, but if this is possible, there's so much more capabilities for disabled people.

56:50Um in any I think there's an app, um I forget the name, but there used to be an app where

56:57Essentially, um uh blind users could sign up to the app and if they needed help, they could open the app and then point their camera

57:08And what it would do is it'd start a video call and on the other end a volunteer would answer the call and tell you what was on the app, right?

57:14Wow.

57:15And they they completely integrated OpenAI um uh

57:21uh the the sort of the four-row model.

57:24Yeah.

57:24And they've basically go into a place where they don't need volunteers anymore because this technology is very good, right?

57:30Yeah.

57:31And that's that's so that's literally what you're just describing there.

57:34Like

57:34um being able to give audio narration for people who are visually impaired um or to be able to augment uh with narration, right?

57:43So some people who can see or can hear, but

57:46for for whatever you know challenges they have in their life need the additional context of what's in front of them to be able to sort of uh experience the world.

57:55So

57:55I think it's um yeah super powerful technology nonetheless.

57:59Definitely, definitely.

58:00Yeah.

58:01Yeah, I'm very excited about it.

58:03Uh yeah, this is awesome.

58:04I think this is really incredible.

58:06I'm super excited to see how people sort of pick up your features and pick up your product.

58:10I think that's going to be an interesting journey.

58:12It probably also takes time, right?

58:13Because not everyone updates the newest stuff straight away.

58:16So

58:17W you're probably thinking about like a six six month timeline to really start to see, okay, what's the uptake like?

58:23And then I think there's this final point about Tableau Cloud.

58:26Viz extensions feel like they work the best if you have Tableau Cloud, right?

58:30And um we you're gonna get it in desktop, which is fine, but it's it's very clear that that

58:38Web experience is sort of what Tableau is pushing for.

58:41And I wonder, I don't know, maybe this is like a

58:46Uh this is like a it's a theory I'm making up on the spot here, but I wonder if like we fast forward the world of Tableau, part of the reason Viz extensions exist

58:55is because they need to build the old charts using this new technology.

59:00Does that make sense?

59:01So they've kind of they've kind of built Viz extensions to enable developers to build new stuff.

59:07But I wonder if they've actually built it to go and fix the old stuff so they can bring everything into the new world.

59:15Wow.

59:16But the Viscura engine is such a powerful engine.

59:19But that but but is it now?

59:21Why why isn't why isn't um all the stuff you're building in in in in VizQL, right?

59:26Like we live in a world in a in a web we live in a web first world and I always feel like that's been the big criticism of

59:33Tableau is that VisQL is a powerful part of everything, but that it was conceived and developed many years ago, and there's probably new ways of doing things today.

59:44that that can be native across multiple platforms, not just um, you know, running through through the query engine.

59:51I think the

59:52the heart of what's happening is probably still the same.

59:54It's just it's like VisQL version two running on the web, whatever it is, right?

59:58It's just running with JavaScript.

60:00It's still VisQL.

60:01It's still the same stuff talking to a hyper, you know, a data source, right?

60:06It's just now designed for the web rather than this sort of funky thing that was sending queries in SQL, right?

60:14Um but yeah, that that's my sort of that's my little like

60:17You know, if I could put a little bet or like I always like to look at what Tableau does and I'm like, hmm, but why would you do that?

60:24Like, you know

60:25Why would you and and the thing is, the reason they can start with other charts is because those are lower risk.

60:33That people have no expectations of how Asankey works.

60:36So of course you can let it out to the community.

60:38See g let the devs sort of try all these things out.

60:41But then, you know, when you come back to build the bar chart, it's got to have all the features Tableau has today.

60:47And that's actually very hard

60:49to get right because what they then have to do is build that porting between VizQL and the new thing.

60:56So they don't have to just build the new charts.

60:58They also have to build

60:59the step in between.

61:00And that's that's the really tough thing.

61:03So um oh yeah.

61:04As a um um like show me more is now five years old.

61:08Yeah.

61:09It is incredible.

61:10We still get questions like we never thought about that

61:13So imagine if you have like Tableau 120,000 companies building the weirdest things with the bar charts and line charts.

61:21Yeah, or people do hacks.

61:22I mean this is Tableau 20 years.

61:24down the road and not addressing features that have been asked for for ten years, there's there's gonna be hacks left, right and center.

61:30And maybe that's what this this sort of new wave is about, resetting the the

61:35the the game and saying, okay, there's a new product, this is how it's gonna work and tough luck, you know, this is where we're going.

61:41Right.

61:42Yeah.

61:42Well it's interesting.

61:43I think the the whole Viz extensions is is there at least to start supporting new types of vises.

61:50Which are not supported by the physical engine or yes.

61:53Never be able to build with the physical engine.

61:55Yeah.

61:55Yeah.

61:55Like the Senke, for example.

61:57Yeah.

61:58Without data densification, of course.

62:00That's the most important.

62:02That's exactly what Tristan said.

62:03Like the please dear God, don't let me do data densification now.

62:08No, exactly.

62:09Although I'll say, like, data densification has had

62:12less of a need with the data model, right?

62:14Look you with the data model, you can achieve the same sort of results without having to like work so hard.

62:20But you still need to understand the concept.

62:21And it is a tough concept to sort of

62:24materialize excusing the sequel terminology, materialize it in your head because you're kind of doing very abstract like

62:32Abs like mathematics with data, it's not a straightforward thing at all.

62:36Well this is a good example where the web is like exceptional.

62:40Um SVG is like extremely fast.

62:44Fast, exactly.

62:45Yeah.

62:46SVGs are just one of those things where y you can just imagine like uh I yeah have you heard of Zvetel?

62:55Zvetel is like this web um Yep

62:57I forget what it is.

62:58It's is it like an API or framework?

63:00I never know which one of the two it is.

63:02It's a framework, right?

63:03Yeah, I think so.

63:04I think it's you build stuff on top of it.

63:05Yeah.

63:06Whatever however you say, I have no idea.

63:09I keep calling it the wrong thing, definitely

63:11But I know that a lot of people have done really interesting visualization capabilities using that because of its SVG support.

63:18And the value there is then you can then define

63:21like custom charts up 'cause an SVG is just a mathematical formula for for an object, right?

63:25It's it's just like a map.

63:27I I like to tell people think of it as maps in in Tableau with polygons.

63:30Rather than looking at images where you have layers and squares.

63:34You have

63:35maths that says okay a circle is this big with this wide or this has twenty-three sides and it happens to be the rough shape of the United Kingdom.

63:44And if you multiply it by 10, everything can scale infinitely and everything can go infinitely small.

63:49So it's it's a much better format.

63:51But anyway

63:51We digress.

63:54It's a bit of fun, fun, fun conversation.

63:56Some something I'm gonna do starting today, and you're gonna be the first person.

64:00I'm gonna ask this.

64:01um because I'm starting to talk to more people about the experience of Tableau.

64:04So um the first thing I'm I'm gonna get you to do is to actually ask the next person a question.

64:10And it's something that, you know, Stephen Bartlett does on his podcast, but

64:14what people get to see here is they'll know who's asked it and they'll get to see who's answered it.

64:18So I have I have two guests lined up for the future.

64:21So if there was one question you could ask that that person you don't know who they are

64:26What would it be, I guess?

64:27And it could be anything.

64:28It's completely open to you.

64:30Yeah.

64:30Oh, whoa, whoa, okay.

64:32Okay.

64:33Okay.

64:34It's gonna be a desktop question then.

64:36Okay.

64:37That's a good choice.

64:38I can give you context.

64:39I think this person would know a good answer, so yeah.

64:42Okay.

64:42So uh the questions

64:44Which new feature announcements at the Tableau conference you never forgot?

64:51Okay

64:51Yeah.

64:52You'd never forgotten the DT.

64:53You never forgotten.

64:53Yeah, that's like the most mind-blowing thing they ever announced.

64:56Memorable.

64:57That's great.

64:58Yeah.

64:58That's your most memorable feature announcement of the Disability.

65:02In the end, data devs on stage is like my main highlight of the conference.

65:06Yeah, yeah.

65:07So what's the most memorable moment for you?

65:10Okay.

65:10Yeah, we'll have to I have an answer, but I'll answer it in the next session.

65:13I've got to keep the carrot

65:15Keep the carrot coming, but when we hang up I'll tell you my answer.

65:18So you do get the answer today at least.

65:20Right.

65:20And then I'll do it again next time.

65:22But

65:23Yeah, um that's super it's a great question, fantastic question.

65:26I'd be interested to see how that gets answered.

65:28But um yeah, listen, uh thank you, Million.

65:30It's been fantastic.

65:31Um I've really enjoyed sort of what you've shown

65:33Um like what you showed was super simple but so powerful and I think you know one of the reasons I'm talking to more and more people is because there's so much happening with Tableau outside of the product.

65:44that I don't think gets sort of time in the usual space.

65:47So I'm trying to talk to more people to to sort of g give them the platform but also get them to showcase the sort of

65:54stranger sides of of the platform we don't get to talk about.

65:57And that that highlights that Tableau is a platform now.

65:59It's not.

66:01We talked about desktop, but I always say to people, I think, you know

66:04Five years ago, desktop was the biggest part of the pl like the whole p thing.

66:0860% of it was all in desktop.

66:10Today, I think the platform has exceeded desktop.

66:13So things are growing in desktop, they're growing outside of desktop.

66:16There's so many places you can get started with Tableau, whether you're a developer, data analyst, engineer, infrastructure person, there is some part of Tableau that needs your skills.

66:26And so

66:27Um yeah, I think you've you've done a good job today of showing people that you can come at it from a product perspective, from a developer perspective, and then you know, you could still bring all your desktop love and and and understanding into that world as well.

66:38So thank you.

66:40Thank you for having me.