Salesforce launches Einstein Copilot for Tableau data analysis tool

There are a lot of different things an enterprise user can do with a business intelligence and data analytics tool like Tableau. Often the biggest challenge is not about having more features, it’s about users staring at a blank page and not knowing what to do.

That’s the challenge Salesforce is looking to solve with the debut of Einstein Copilot for Tableau, which is entering beta today. The new AI tool for Tableau is an extension of Salesforce’s broader portfolio efforts with Einstein Copilot, which brings a generative AI set of capabilities to Salesforce’s applications. Einstein Copilot was made available to Salesforce CRM users as a public beta last month, providing users with a conversational assistant to help with workflows.

Einstein Copilot for Tableau is specifically tuned for data analysis and isn’t just about asking natural language queries about data. The new copilot integrates a series of capabilities to help business users and data analysts alike get past the dreaded ‘blank page’ situation, where they aren’t exactly sure what to do next, or how to perform a certain operation. Einstein Copilot for Tableau integrates recommended questions to help prompt users with what can be asked about a particular data set. There is also a conversational data exploration capability to help users dig deeper into results.

Fundamentally it’s all about providing business users with a tool to accelerate data insights, without the hassle of first learning to be a prompt engineer.

“We don’t want to come out and say to the user – you need to write better prompts,” Southard Jones, Chief Product Officer, at Tableau told VentureBeat. ” So we spent a lot of energy and time making sure that when an analyst is in the traditional Tableau pills and shelves experience and they ask questions that they can get responses back or we guide them to very specific answers.”

In Tableau,  pill refers to data types that a user pulls into a workflow while shelves refers to the columns and rows of data being analyzed.

Einstein Copilot goes beyond Tableau Pulse

The integration of AI-powered tooling into Tableau is not a new thing.

In February of this year, Tableau debuted its AI-powered Pulse tool,  which helps users surface data insights and build data visualizations. Even before that, Tableau has had multiple iterations of tools that enable varying degrees of natural language queries and AI-powered insights as well.

Where Einstein Copilot for Tableau is different is in its focus as an actual assistant to help users with data analysis and data exploration operations.

“This is going to actually recommend things for you,” Southard said. “It will assist you as you build your analysis to do your analysis and that gets people comfortable using the interface.”

The interface is also driven by feedback, enabling users to identify if the recommendations were helpful or if more is needed. Southard emphasized that multiple user studies that Tableau has done found that people are more likely to use a feature if there is an easy way to submit feedback about it.

Guided Calculation Creation helps to prepare data for analysis

In addition to data exploration, Tableau also has data preparation capabilities as part of the core platform. 

One of the toughest things for a new data analyst is often figuring out how to create data calculations, which is now an area where Einstein Copilot for Tableau can help.

“When you prepare data for analysis, you are often doing things like adding columns or creating a calculation,” Southard said. “Creating a calculation in Tableau, just like with any other tool, sometimes you have to know some language and it can be hard to learn so we are now allowing you to write in human language and we’ll turn it into the machine language.”

Writing an email with gen AI isn’t the same as doing data analysis

While Salesforce has already been deploying Einstein Copilot for CRM and uses the same core foundation to enable the Tableau capability, Southard emphasized it has been trained and optimized for a specific use case.

With Salesforce CRM, the context of the data is understood as all the data is already stored and collected inside of Salesforce. With Tableau, the context for how the data is going to be used is not just about CRM it could be used for any type of data analysis. 

“LLMs can do amazing things, but in order to get the prompt to return a good response and to interact with it positively you’ve got to know the use case well,” he said.  “Writing an email is far different than having a service call conversation and is far different than asking a question about random data.”