Connecting Claude and Gemini to Mosaic via MCP: A Governed Path for AI Agents

Plugging AI Into Your Governed Data: Connecting Claude and Gemini to Mosaic Models via MCP

AI agents are only as trustworthy as the data behind them. Point a chatbot at raw tables and you get confident-sounding answers built on inconsistent definitions and unguarded access. The Model Context Protocol (MCP) changes that—giving AI tools a secure, governed path to the same trusted metrics your analysts already use.

In this one-hour session, we'll show you how to connect AI tools like Claude and Gemini to Strategy Mosaic models through MCP. You'll see how an agent discovers available models, understands their business context—metric definitions, relationships, and descriptions—and issues governed queries in natural language. Critically, every query honors the same row-level and object-level security applied to human users, so agents only ever see data they're permitted to access.

We'll walk through the connection setup, demonstrate live natural-language queries against a published Mosaic model, and discuss what makes this approach fundamentally safer than handing AI direct database access. You'll leave understanding both the "how" and the "why this matters" for deploying AI responsibly across your organization.

What you'll learn: 

• What MCP is and why it's the bridge between AI agents and governed data

• Connecting Claude and Gemini to published Mosaic models

• How agents discover models and interpret semantic context to pick the right metrics

• How row- and object-level security carry through to every AI query

• Why a governed semantic layer beats direct database access for AI reliability

Who should attend: 

Analytics leaders, BI developers, data governance teams, and anyone evaluating how to bring AI agents safely into their data stack.