Coming Soon workshop
Build Self-Service Analytics You Can Trust with Claude + MCP
Claude
AI
Data Analysis
Data Governance
Notify me when registration is open
Data teams spend much of their time on data wrangling and ad hoc requests, which leaves little room for deep analysis. Connecting AI tools directly to your data warehouse can finally make self-service analytics real, so non-technical users can answer their own questions without writing a line of SQL. But without the right business context and guardrails, off-the-shelf AI tools can expose sensitive data, return wrong or misaligned answers, and burn through tokens.
In this hands-on session, you'll build a complete AI-powered analytics environment from scratch using Docker, open-source Metabase, and Claude, including a basic semantic layer and a Model Context Protocol (MCP) server that generates reliable queries and dashboards on the fly. You'll walk away with a tool-agnostic framework you can apply to any modern BI and AI stack.
What you'll learn
Configure a local AI analytics environment with Docker, open-source Metabase, and the Claude API
You'll build it step by step during the session, so you leave with a working setup on your own machine.
Deploy an MCP connection that safely links a large language model to your database and BI tool
A controlled connection is what lets AI work with your data while keeping sensitive information protected.
Define a basic semantic layer that gives AI the business context to write accurate, token-efficient SQL
Without that context, AI tools often misread your tables and metrics and return misaligned answers.
Spot where text-to-SQL goes wrong and apply prompting and model-selection best practices
The right prompts and models improve accuracy and reliability while keeping costs down.
Who should join this workshop?
Data analysts and BI developers who field a constant stream of ad hoc requests and want to give business teams a reliable way to answer their own questions
Analytics leads and data managers who are being asked to "add AI" to the reporting stack and need a practical, low-risk place to start
Data practitioners who want an AI analyst of their own that can write queries and build dashboards
Prerequisites
Basic knowledge of SQL, databases, and data analytics workflows
A paid Claude subscription (for API/inference access)
Docker Desktop (free version) installed on your machine, if you plan to follow along live
Some experience with open-source Metabase is useful, but not required
Stephen Tracy
Founder & Principal Consultant, Mozaik
Stephen Tracy is an award-winning data and AI expert whose 15-year career has taken him from agency analyst to Chief Data Officer, and from Singapore to Toronto. He has built data teams at global firms like IPG, Publicis, and YouGov, and co-founded Milieu Insight, one of Southeast Asia's fastest-growing customer intelligence platforms. Today, he runs Mozaik, a Toronto-based intelligence consultancy, and is building TalentQuill, an AI fluency and skills assessment platform. A passionate educator, he has also built best-selling courses on data, AI and leadership that have reached over 50K students globally.
