Video |

Tableau | AI & ML | Data visualisation

Generative AI in Tableau Desktop 2025.1 - 8 Language Audio Track

The AI agent now lives inside Tableau Desktop, and for once it wrote my first-time-customer LOD faster than I could copy and paste it.

Tim Ngwena •21 March 2025 •1 min read •5 min watch •13K

Part ofWhat's new in Tableau 2025.1

Watch on YouTube

The Tableau AI agent in Desktop requires a Tableau Plus premium licence and a Tableau Cloud sign-in; it is not coming to Tableau Server on any current roadmap.
Once signed in, an Einstein icon appears top-right in Desktop, opening the same generative AI side panel available in Web Edit.
Asking the agent to classify first-time versus repeat customers produced a correct LOD calculation and added the field straight to the data pane, saving the manual work of researching granularity and FIXED expressions.
Follow-on suggestion buttons propose sensible next questions, such as comparing average sales between customer types, and build the charts automatically.
Data is sent to the Einstein Trust layer rather than directly to the LLM, keeping real data secure during querying.

0:00In Tableau 25.

0:011 you can now use the agent inside of Tableau Desktop itself.

0:05I'm going to show you how that works in this video

0:07Let's get stuck in.

0:08Okay, so the first thing you're going to need to do is to sign in to a Tableau Clyde instance that has the Tableau Plus license enabled.

0:16It's not going to work unless you have Tableau Plus.

0:18Tableau plus is the premium tier license for Tableau.

0:22To do that, you need to go up into the server settings and make sure you sign into the tableau.

0:27cloud instance it's only available for tableau cloud this won't be coming to tableau server I get that question a lot there's no roadmap really for any of these ai features for Tableau server so until that changes

0:38Don't expect it in Tableau Server in any upcoming release.

0:41So once you've signed into Tableau Cloud, you can go ahead, open up Desktop as normal, and over here on the top right hand side you'll see you get this little Einstein icon.

0:49This is the Salesforce icon for using generative AI.

0:52When you click on it, it opens up the side panel and you pretty much have the same panel you have in WebEdit.

0:59Now there have also been some improvements done to this agent.

1:01So

1:02You can go ahead and ask a range of questions and what I'll do is at the end of this video I'll link you to a great session that Tableau had showing you the upcoming capabilities in this particular area.

1:11But if you have Tableau Plus and you're using this agent

1:14you're now able to use it inside of desktop.

1:16So we can get it to go ahead and do a very simple thing.

1:19An example of a calculation I saw them writing was asking Tableau to classify first-time customers and repeat customers.

1:25So I'm just going to go ahead and repeat that uh question here.

1:28Can you

1:29uh classify first time customers and repeat customers now before I hit enter the way I would do this is I'd use an LOD to look at the customer level for the

1:42lowest order date in the transaction list and once you found the smallest transaction item you can basically check every date against that date to say true or false for uh the specific first time or last time customer and then you're pretty much good to go so let's go

1:57Ahead and hit enter and see what it suggests.

1:59Now it takes some time because what's happening here is Tableau Desktop is querying the cloud for this answer.

2:05And the way that this is working

2:06is it sends the data over to the Einstein Trust layer.

2:09Now I've actually done a whole course on LinkedIn learning that covers how Tableau AI works and it has a section specifically

2:16on this bit but what i'll do is i'll put a diagram on screen on how the einstein trust layer works you can go ahead and check that out and it will break down how your data stays secure without

2:26you having to send any real data to the LLM.

2:29But anyway, you can see it's done the calculation and I think it's broadly gotten the question right.

2:33So if the order date is equal to

2:36the minimum order date fix to the customer level then it's a first time customer else repeat customer end and actually what it's done is not only is it created the calculation it's also put it here in the data pane have

2:48I was doing this demo previously, which is why there's already an example right here.

2:52It's just done the same thing twice.

2:53But you can see it did the sort of whole thing and it actually renamed it uh number two as well.

2:57So that

2:58That's exact same calculation.

2:59Now the nice thing is you can then go on and do some follow-on questions.

3:02So okay, the new customer classification field was added

3:06to the data pane if you go ahead and click the suggestions on the bottom it thinks about what would make sense to ask next and it gives you those sort of three options here so how does the average sales value compare between the first time customer

3:17and the repeat customer, if you go ahead and click on that, the analysis then gets done.

3:22So even though this wasn't great at launch, I think this is starting to become a useful feature, especially for new analysts who are just looking for quick ways to ask questions.

3:30The critical thing here is

3:32Look, if you were trying to figure this out yourself, you would have had to have gone to Google, correctly found the information to let you know that an LOD was required.

3:41Having done that, you would then have to understand

3:44the concept of granularity in a lot of detail and then understand which LOD to use fixed versus include versus exclude

3:53Once you've done that, you then have to correctly put together this formula logic.

3:58Now in most cases what you do is you look for this formula and you just copy and paste it in.

4:02But even this was faster than doing that.

4:04So that's actually quite a nice uh quality of life improvement if you have the license that is.

4:08And it's a really nice thing to have.

4:09So you can just go ahead and ask these questions and have that running inside of Tableau.

4:13Now you can see here it's built the charts, built the analysis

4:17and we can essentially start using this in our work and we're pretty much good to go.

4:20That's the feature in a nutshell.

4:22Now I said I'd show you um a a session that they had that showcases more of what's coming.

4:27So let me just bring that here on screen.

4:29If I bring that in you'll see that they have this um

4:32uh new series called inside track on tableau and essentially what this is a series by tableau that's going to be looking at specific features in the product it's actually a user group so you can sign up to this user group and get involved but they had uh this exact demo and they basically walked

4:47through parts of the capability so you can see here that they've got some improved capabilities coming here in April and May and June.

4:53So first on cloud

4:55And then subsequently in the follow-up release.

4:57We know here that the next release for Tableau will be around June 25, because that's what they're saying here in terms of general availability.

5:04So there you go.

5:05That's um all I have for you.

5:06Thanks for watching and I'll see you in the next one

Rollup

Tableau Desktop 2025.1 brings the Einstein generative AI agent directly into Desktop, letting you type a plain-language request and get a working calculation added straight to your data pane, rather than researching and writing it yourself.

The demo uses a transactions dataset and asks the agent to classify customers as first-time versus repeat, a task that would normally require writing a level-of-detail calculation by hand.

The Breakdown

Licence and sign-in requirements 0:09

You must sign in to a Tableau Cloud instance with Tableau Plus, the premium licence tier, before the agent works in Desktop at all. It only works with Tableau Cloud; there is no roadmap for bringing it to Tableau Server.
Opening the AI agent in Desktop 0:41

Once signed in, an Einstein icon appears top-right in Desktop, opening the same generative AI side panel used in Web Edit, now available inside the desktop application itself.
Asking it to build a classification calculation 1:26

Typing a plain request to classify first-time versus repeat customers sends the query to the cloud, which returns a correct LOD calculation checking each order date against the customer's minimum order date, and drops the field straight into the data pane.
Data goes through the Einstein Trust layer 2:05

Your data is sent to the Einstein Trust layer rather than directly to the underlying LLM, which is how Tableau keeps real data secure while the agent works out the answer.
Using follow-up suggestion buttons 3:00

After a calculation is created, the panel suggests sensible next questions, such as comparing average sales between the two customer groups; clicking a suggestion builds the chart automatically without further setup.
Why this saves real time 3:22

Manually solving this problem means researching that an LOD is needed, understanding granularity and choosing between fixed, include and exclude, then writing the formula correctly; the agent skips all of that even for an experienced user, and is especially useful for newer analysts.
What's coming next 4:22

Tableau's own Inside Track user-group session shows further capabilities landing first on Cloud in the following months, with the next general-availability release for these AI features expected around June.

Worth Knowing

Requires both a Tableau Plus licence and a Tableau Cloud sign-in — it won't function on Tableau Server, and there's no roadmap to change that.
The agent occasionally repeats or renames a calculation if run twice in the same session, so check the data pane for duplicates.
Even for someone who already knows how to write the LOD themselves, asking the agent was still faster than typing or copy-pasting the formula.

Use It When

Reach for this when you want a quick calculation or chart without stopping to look up LOD syntax or granularity rules, particularly if you're newer to Tableau or just want to move fast — provided you already have Tableau Plus and Cloud access.

How this Rollup was made provenance & method

A Rollup is drafted by AI from the video's transcript, then reviewed and edited by Tim. Everything used to produce this one is listed below — the model, the exact prompt, and the source video — so the process is transparent and reproducible.

Transcription: On-device — NVIDIA Parakeet v3 for recent videos, OpenAI Whisper large-v3 for earlier ones. The transcript never leaves the machine or gets published.
Drafting: Claude Sonnet 5 in the cloud, from that transcript.
Prompt: The exact Rollup prompt (v2) — the full system prompt, unedited.
Source video: Watch on YouTube
Drafted: 5 July 2026 at 09:38
Reviewed & edited: 5 July 2026 at 09:41 · by Tim Ngwena

Model + prompt + video is everything you'd need to recreate a Rollup like this yourself. The one thing we don't share is the transcript.

Rights. The video and its transcript are the property of TN Media Ltd. Unauthorised use or download is prohibited. © TN Media Ltd.