Business intelligence is currently moving to bridge the gap between curious business users and data through AI-powered agents. The transformation promises to eliminate the traditional barriers that have plagued analytics for decades. We spoke with Matthew Miller, VP of Product Management at Tableau/Salesforce, about what is called Tableau Next.
Miller, who has worked in and around analytics for 27 years, describes what he calls a pivotal moment in the analytics world. “Data is traditionally hard. Business people can’t get access to it,” he explains, referencing the promise that has defined the industry since the late 1990s. “We’ve been chasing this dream of closing the gap between the curious people who have questions and the ability to actually get answers.”
Despite decades of innovation, this fundamental challenge persists. Miller recalls working with a Dutch medical technology company where no fewer than eight layers of bureaucracy separated business users from the people controlling the data. That complexity created bottlenecks rather than solutions, keeping valuable insights out of reach.
The new version of the Tableau platform, branded Tableau Next, directly addresses this issue by embedding AI into the fabric of analytics, rethinking how organizations interact with data and how insights flow throughout the enterprise.
Capturing institutional knowledge through semantic layers
One of the biggest barriers to data democratization is the loss of institutional knowledge. During a customer visit to Arendonk, Belgium, Miller encountered a veteran employee who had spent 25 years developing an intuitive sense of what the data really meant. “As the dashboard is loading, this old guy is going ‘you want that table? No, no, not that one. You gotta exclude code 999. That’s a test code,’” Miller recalls. “How would I have known that?”
It’s precisely this kind of tacit expertise that Tableau wants to capture and encode. The goal is to ensure that key knowledge doesn’t remain locked in the minds of long-tenured employees. That’s where the semantic layer of Tableau Next comes into play.
Tableau Semantics forms the foundation of the platform’s intelligence capabilities. It enriches raw data with context, allowing users across departments to speak the same business language. Through metadata modeling, it provides clarity on how data is generated, processed, and presented. Thus, building trust in the analytical process and establishing a common ground for interpretation. “Essentially, we want to find that old guy in every one of our customers and have him dump everything he knows into the semantics and train it into the model,” Miller says.
An agentic architecture with skills and semantics
The backbone of Tableau’s AI-driven approach is what Miller describes as “skills within an agent.” The architecture is structured to reflect how humans naturally ask questions and perform tasks. Skills contain topics, topics contain actions, and actions are deterministic in nature. This layered setup allows the system to handle complex queries while remaining predictable and compliant.
“The power of large language models is that these huge statistical models can figure out the imprecision of language,” Miller explains. Whether users are retrieving information or updating records, the agent understands intent, even when phrased in natural, conversational language.
Crucially, the agentic framework respects the rules that organizations must follow. “If you’re dealing with government compliance, the agent will still obey policies on table joins or revenue recognition,” Miller emphasizes. The goal is not to bypass process, but to intelligently navigate it.
This is also where Tableau Next’s architecture shows its enterprise maturity. It includes data lineage tools that visualize the journey from raw input to finished dashboard, supporting transparency and governance. Combined with Salesforce Data Cloud and MuleSoft integrations, the platform delivers real-time insights across hybrid environments without requiring organizations to move or duplicate their data. This is precisely where the advantage of being a Salesforce company comes into play. Tableau Next is here for companies that also use Salesforce, and for those who don’t, there is Tableau Cloud or Tableau Server.
Ephemeral analytics and the end of dashboard fatigue
As part of its broader transformation, Tableau is challenging the conventional reliance on static dashboards. Miller estimates that up to 50% of today’s dashboards could be replaced by what he calls “metric-centric experiences.” These experiences surface insights dynamically, in response to business questions, rather than through predefined visual templates.
At the heart of this shift is Tableau Pulse, a system that generates on-demand visualizations based on certified metrics. Users can subscribe to metrics much like following people on social media. When a change occurs, for example in sales pipeline or supply chain delays, the system pushes a tailored update directly to the user. “For an executive who says it takes me two months to get a new dashboard, forget that,” Miller says. “This will build it for you.”
These visuals are transient by design. “If I’ve got a question about my supply chain and we generate a new visual that answers your question so you can make a decision, we might not need to keep it,” Miller adds. This idea of ephemeral analytics challenges long-standing conventions, offering more fluid, mobile-friendly, and context-sensitive insights that better reflect the dynamic nature of modern business.
Making insights actionable, not just accessible
Even the most insightful data is worthless if it doesn’t reach the right people in time. That’s where Tableau’s use of AI agents becomes transformational. These agents operate inside the flow of daily work, proactively surfacing trends and recommending actions before users even formulate the right query.
Tableau Next combines generative and predictive AI to forecast future states and suggest next steps. Users no longer need to pull reports or dig through dashboards. Instead, insights arrive contextually within the tools they’re already using. By pushing contextual, AI-powered insights directly into the flow of work, Tableau aims to make it easier for everyone in the organization to act on valuable information.
Embracing platform extensibility and reuse
To support enterprise-scale collaboration, Tableau Next also introduces a component marketplace that allows teams to share reusable assets across departments or even organizations. Dashboards, data models, and custom applications can be published and reused, accelerating development cycles and enforcing best practices.
The API-first architecture supports extensibility, enabling Tableau capabilities to be embedded within third-party applications. Whether through Slack, CRM systems, or industry-specific platforms, data insights should become a native part of business workflows. This composable approach to analytics reflects a broader trend in enterprise software: build once, deploy many times, and meet users where they are.
A realistic view on the adoption curve
Despite the transformative potential, Miller is clear-eyed about where many organizations currently stand. Having spent nearly a decade living in the Netherlands, he understands the cautious pace that often characterizes European IT adoption. “I have a customer who still uses Microsoft Access and they’re a big company,” he notes. “I helped a global company move off of Lotus Notes to Tableau in 2018.”
That’s why Tableau emphasizes freedom, flexibility, and choice in its rollout. The shift to AI-powered, conversational analytics won’t happen overnight. Traditional dashboards and reports remain relevant, especially for organizations still maturing their data infrastructure. “We’re not ripping those out of your hands,” Miller says. “We’re giving you a way to evolve at your own pace.”
Perhaps the most profound change is what this all means for the analytics profession itself. As agents take over basic query construction and visualization tasks, the analyst’s role shifts toward semantic modeling, prompt design, and quality control. Analysts won’t disappear. They’ll become stewards of the system, ensuring that models reflect business logic, outputs are trustworthy, and prompts deliver value. The semantic layer becomes their canvas, and the AI agent their tool of expression. “The idea is not to eliminate analysts,” Miller concludes. “It’s to elevate their work and bring them closer to the heart of the business.”
Also read: Tableau keeps business intelligence (BI) alive and kicking