Building AI-Ready Semantic Foundations for Enterprise Analytics
Webinar Speaker: Donald Farmer, TDWI Research Fellow
Date: Thursday, November 5, 2026
Time: 9:00 a.m. P.T. / 12:00 p.m. PT
Conversational analytics and AI agents are changing who asks questions of enterprise data. People who never had a BI license can now ask questions in plain language, and agents can run a series of queries without a person reviewing each step.
Yet generative AI produces a fluent-sounding answer whether or not the number is right, and new users are often the least able to spot an error. Because an agent uses each result as input to the next query, errors compound: 90% accuracy per question falls to about 59% over five dependent questions.
TDWI research finds that 66% of organizations call a semantic layer critical, but only 19% have one. Most enterprises already have several semantic models, each built into one BI tool or platform, and they often give different answers. What is missing is one governed semantic layer that computes metrics the same way for every tool and agent. Using this research and real-world examples, this webinar shows how organizations can build and govern that layer.
In this TDWI Solution Lab, research fellow Donald Farmer gives an independent view of semantic layers for AI and the relevant TDWI research. An AtScale and Precisely specialist will show demonstrations of a governed semantic layer giving the same answers to BI tools and AI agents, and will join Donald for a discussion of how organizations are building these foundations today and what comes next.
Data and analytics leaders who need consistent, auditable AI answers will learn:
- Why a fluent AI answer can still be wrong, and why errors compound across agent queries
- What differentiates a universal semantic layer from catalogs, glossaries, and BI-specific semantic models
- How a central team and business domains can share ownership of metric definitions