Salesforce's business intelligence platform, Tableau, is getting generative AI features in the form of Tableau GPT, built on the company's proprietary Einstein GPT AI engine, which has also been integrated into other products such as Slack.\n\n\u201cTableau GPT can enhance and automate things like analyzing data, exploring it, sharing it, consuming it. The generative AI engine introduces a number of really exciting use cases where for example, analyzing data feels more like a conversation via a chatbot as opposed to drag and drop,\u201d said Pedro Arellano, head of product at Tableau.\n\n\u201cOther use cases include the engine anticipating questions that users might ask based on what's already in the data or taking hundreds of insights and explaining them using very easy to understand summaries,\u201d Arellano said.\n\nEinstein GPT, the foundation for Tableau GPT, comprises various large language models (LLMs) including those from OpenAI, Cohere, and internal, proprietary Salesforce models, noted Sanjeev Mohan, principal analyst at independent consulting firm SanjMo.\n\nThese internal models were \u00a0driven by Salesforce\u2019s investments in companies with natural language processing abilities, and insights about how enterprises conduct data analytics, according to Amalgam Insights principal analyst Hyoun Park.\n\n\u201cTableau previously acquired Narrative Science, a natural language generation solution for analytics. In addition, Salesforce has made strong investments in data science over the years such as BeyondCore, Metamind, and Datorama and has hundreds of data scientists in house as well,\u201d Park said.\n\nIn addition, Tableau GPT has been given a data security and governance layer in order to protect enterprise data from internal and external data leakages or unauthorized access, according to Arellano.\n\nThe addition of the governance and security can be attributed to Salesforce's effort to build trust among customers, especially at a time when companies are banning the use of OpenAI\u2019s ChatGPT over data leak concerns, analysts said. \n\n\u201cThese layers protect users who are afraid that their prompts will be used to retrain LLMs. Also, it can guard against LLM hallucinations,\u201d SanjMo's Mohan said. \u00a0\n\nTableau GPT is expected to be available in pilot later this year, the company said.\n\nProactive data analytics with Tableau Pulse\n\nSalesforce has also released a new flavor of data analytics under an offering dubbed Tableau Pulse, which the company said offers proactive analytics.\n\n\u201cIt is sort of a personal guide for your data, where it knows your data. It knows the goals you're trying to achieve with your data. And it helps you reach those goals,\u201d Arellano said.\n\nTableau Pulse will also use Tableau GPT to help enterprise users make better, faster decisions using automated analytics on personalized metrics in an \u201ceasy-to-understand way,\u201d Arellano \u00a0said, adding that that Pulse can surface insights in both natural language and visual formats.\n\nUse cases include alerts when there is an unusual change in data or metrics, and help for users to drill down to the reason for the anomaly, the company said.\n\nThese insights can be further shared with colleagues via collaboration platforms such as Jira or Slack in order to find a resolution, Salesforce added.\n\n\u201cThe automatic nature of the analyses provided by Pulse increases productivity but also introduces consistency and comprehensiveness since the same analytics are applied wherever necessary,\u201d said David Menninger, research director at Ventana Research.\n\nHowever, Tableau might be playing catch up with other vendors, analysts said.\n\n\u201cA number of vendors have developed and are refining ways to look at the graph of individual and user behaviors and interactions with data and then glean insights and make recommendations based on changes,\u201d said Doug Henschen, principal analyst at Constellation Research.\n\nCloud-based products, according to Henschen, tend to have a leg up in analyzing user behaviors and data interactions at scale.\n\n\u201cProducts that started out as server-based products, like Tableau, have typically taken longer to develop graph and personalization capabilities that can be delivered consistently across the both cloud and on-premises deployments,\u201d Henschen said. \n\nThough many vendors offer automated insights, the addition of generative AI-produced narratives \u201cwill help make these insights more complete and more easily delivered in multiple languages," Ventana's Menninger said.\n\nTableau Pulse is expected to be available in pilot later this year, the company said.\n\nData Cloud for Tableau to unify data for analytics\n\nIn addition to Tableau Pulse, Salesforce is offering Data Cloud for Tableau to unify enterprises\u2019 data for analytics.\n\nThe plan is to layer Tableau on top of the Data Cloud, which was released last year in September at Dreamforce under the name "Genie."\n\n\u201cWith Tableau, all of a company\u2019s customer data can be visualized to help users explore and find insights more easily. Data Cloud also supports zero-copy data sharing, which means that users can virtualize Data Cloud data in other databases, making it instantly available to anyone,\u201d the company said in a statement.\n\nData Cloud for Tableau will also come with data querying capabilities, the company added.\n\nThere are many business advantages that Data Cloud for Tableau can provide, according to Henschen.\n\n\u201cAdvantages include bringing together all your disparate data, separating compute and storage decisions, and enabling many types of analysis and many different use cases against the data cloud without replication and redundant copies of data,\u201d Henschen said.\n\nSalesforce\u2019s move to combine its Data Cloud with Tableau can be attributed to Tableau having reaching a ceiling in its core analytic discovery capabilities, according to Park.\n\n\u201cIt is being pressured to increasingly support larger analytics use cases that push into data management and data warehousing. Although Tableau is not going to be a full-fledged data warehouse, it does want to be a source of master data where analytic data is accessed,\u201d Park said.\n\nData Cloud for Tableau, however, is part of a strategy to compete with data lakehouse, data warehouse vendors, and an effort to own or control more data, Menninger said. The integration of Tableau and Data Cloud will lead to direct competition with the likes of Qlik, Tibco IBM, Oracle, and SAP, analysts said.\n\nData Cloud for Tableau is expected to be made available later this year.\n\nOther updates includes a new developer capability, dubbed VizQL (visual query language) Data Service, that allows enterprise users to embed Tableau anywhere into an automated business workflow.\n\n\u201cVizQL Data Service is a layer that will sit on top of published data sources and existing models and allows developers to build composable data products with a simple programming interface,\u201d the company said.\n\nSalesforce woos new users with Tableau generative AI\n\nGenerally, the addition of generative AI features to Tableau can be seen as an attempt to attract customers who are not analytics or data experts. Business intelligence suites face a problem of adoption as at least 35% of employees are not willing to learn about analytics or data structures, Park said.\n\n\u201cTo get past that, analytics needs a fundamentally different user interface. This combination of a natural language processing, natural language generation, generative AI, and jargon-free inputs that translate standard language into data relationships provides that user interface,\u201d Park added.\n\nAnother reason why the new features could attract customers is the disinterest of business users in using dashboards. \u201cThese users would rather use natural language which has context. Up until now, NLP was very difficult for computers to handle but the new LLMs changed that,\u201d Mohan added.