Tableau

Enterprise data and AI architectures are moving faster than ever. As every tool and data platform introduces its own semantic layer to power AI transformations, a fundamental challenge emerges: semantic drift. When your analytics dashboards, ad-hoc query tools, and AI agents interpret business metrics and dimensions differently, trust breaks down.

Organizations rely on dashboards to report important insights, but dashboards are just the start of the conversation. A chart shows revenue is down 12% quarter-over-quarter and the follow-ups immediately stack up: Which region? How many product lines? Those questions pile up in an analyst's queue, and by the time the answers are prepared, the moment to act may have already passed. LLMs seem like an obvious fix, but an LLM alone creates a new problem: it doesn't know your business.

Analytics works best when knowledge is accessible: delivered to the people who need it, exactly when they need it, and proactively surfaced before they even have to ask. But as organizations scale and the AI landscape continues to change what users expect out of their platforms, their analytics strategies need to evolve. Audiences grow to include partners, customers, external stakeholders, and now AI agents running automated workflows.

Get the right insight to the right person—before they ask Picture this: It’s 8:25 a.m., your morning coffee is brewing, and you have a massive forecast review meeting in five minutes. Instead of scrambling across tabs, sifting through many dashboards, and filtering through a mountain of noise, a clean text brief pops up directly inside Slack. It outlines your team's current pipeline health, flags critical deal risks, and shows you the Salesforce opportunities that need your attention today.

Ask any analyst what they consider the major pain point of dashboard authoring. Chances are, it will be the “last mile”: styling and formatting. Now, think about why. You talked to your stakeholders, cleaned your data, and curated your views. In comparison, tasks like view placement, adjusting fonts, configuring margins, and matching the corporate style guide are all easy and minor, but are also manual and repetitive fine-tuning work that you’ve already done for a dozen other dashboards.

Modern data exploration tools often struggle with a basic problem: people typically know the pattern they are looking for, but cannot easily express it through SQL, filters, or natural language alone. In this project, we built a multimodal research prototype that lets users search data using sketches, annotations, and natural language together within the same interaction space.

For the past 16 years, the Iron Viz competition has captivated Tableau Conference attendees and viewers worldwide. The ultimate data showdown viz off where data viz experts compete in a 20-minute dashboard creation challenge. Let’s relive the energy and highlights of the 2026 Iron Viz show. Hosts Archana Ganeshalingam and Cate Guyman kick off the 2026 Iron Viz Championship.

At Tableau Conference 2026, the message is clear: in a world where AI makes generating answers easy, acting on the right answer is everything. We are witnessing a fundamental shift from reporting what happened in the past to orchestrating what happens next across your business. This isn't a vision for a single solution. This is the moment that groundbreaking agentic analytics capabilities are available across the entire portfolio—including Tableau Cloud, Server, Desktop, and Next.

AI can generate an answer in seconds and agents can automate entire workflows, but none of that matters if the underlying data is siloed, out-of-date, dirty, or missing your business context. Ultimately, the integrity of your data determines the accuracy of every insight and the reliability of each agent action. The real question isn't whether your organization is ready for AI, but whether your data is even ready for it. The response?

Tableau Research presents Lexara, a user-centered evaluation toolkit for Conversational Visual Analytics (CVA). This research was conducted by Srishti Palani and Vidya Setlur at Tableau Research. This work was presented at the ACM CHI Conference on Human Factors in Computing Systems, the premier venue for research in human-computer interaction.