For years, getting an answer from company data meant asking an analyst, waiting for a report, or learning a dashboard tool. That’s changing fast. With tools like Claude, Tableau Pulse, and Tableau Next, people can ask questions of their data in plain language, and AI agents can prepare analysis on their own.
It’s a real shift, and we’re putting it to work with clients. But there’s a catch that the demos rarely mention: AI will answer confidently whether or not the numbers underneath it are right.
AI doesn’t fix bad definitions. It spreads them.
Ask ten people in most organizations what counts as an “active case,” a “response time,” or “revenue this quarter,” and you’ll get several different answers. People work around those differences because they know the context. An AI agent doesn’t. It takes whatever data it’s given and produces a fluent, convincing answer.
If two spreadsheets define the same metric differently, a person might notice. An AI agent summarizing both might just average the confusion and present it as insight. The faster and easier AI makes it to ask questions, the more important it becomes that the answers rest on something solid.
What "governed data" actually means
Governed data isn’t a product you buy. It’s a set of decisions you make and keep:
- One definition per metric. Each measure is defined once, documented, and reused everywhere it appears.
- One trusted model. Data from every source system is mapped into a single structure, so a case, a milestone, or a dollar means the same thing in every report.
- Traceability. Any number can be traced back to the records behind it.
- Clear ownership. Someone is responsible for each definition and for changing it when the business changes.
When AI works from that foundation, a plain-language question returns an answer grounded in the same numbers leadership and clients already trust.
What we’ve seen in practice
For a mass tort case acquisition firm, BRIO built a governed data model and Tableau reporting that litigation funders can trust. During testing, every figure the client questioned was traced back to the intake records behind it, and those reviews sharpened the definitions themselves. Only then did conversational and agentic AI make sense. Now the firm can ask questions in plain language and get answers grounded in the same definitions its funders see.
For a healthcare risk management firm, we designed custom KPIs and embedded live metrics in its customer portal. The same governed model now feeds AI that helps the firm prepare analysis for its monthly client meetings, using the exact metric definitions its clients already know.
In both cases, the AI was the easy part. The hard, valuable work was getting everyone to agree on what the numbers mean.
Data culture comes first
Trustworthy numbers are as much about people as pipelines. That’s why we treat data culture as part of the work: establishing shared definitions with leadership, reviewing questioned figures together, and refining definitions as the business learns. It’s not glamorous, but it’s what makes an AI answer something you can act on.
Where to start
- Pick the five metrics leadership asks about most and write down exactly how each one is calculated.
- Find where those numbers disagree today across reports, spreadsheets, and systems.
- Build one trusted model before adding more dashboards or AI tools.
- Then add conversational AI on top, and test it with questions you already know the answers to.
Agentic analytics is genuinely exciting. Just make sure it’s standing on data you’d be willing to defend. Learn more about our data intelligence and Tableau work.
Frequently asked questions
What is agentic analytics?
Agentic analytics uses AI agents that can answer plain-language questions about data and prepare analysis on their own, rather than waiting for a person to build each report.
Why does AI need governed data?
AI answers confidently regardless of data quality. Governed data, with one documented definition per metric and a single trusted model, makes sure those answers rest on numbers the organization agrees on.
What is Tableau Next?
Tableau Next is Salesforce’s agentic analytics platform, which brings AI agents and conversational analysis to Tableau and Salesforce data.
