Once your company data sits in a well-modeled lakehouse, AI assistants like Claude, ChatGPT, and Copilot can build board decks in your branding, run what-if scenarios on pricing and demand, and pull in outside context like weather to explain a dip in sales. In this episode, Brick Thompson and Landon Ochs walk through the architecture we use to connect those assistants to client data through a custom MCP server.

Landon covers each step of the flow, starting with sign-in through Entra, where security groups determine which tools and data each person can reach. Business users get curated views with the company’s metric definitions built in, so a question like “show me sales” returns the right view for their role. Analysts and developers get read-only access across the bronze, silver, gold, and platinum layers to accelerate their Power BI and Excel work. Every query runs under the user’s own identity, which carries row level security through to the AI and records the person’s name in the audit log.

The conversation also covers searching unstructured sources like Gong transcripts, PDFs, and emails through Azure AI Search, why midsize companies with multiple ERPs or recent acquisitions benefit from consolidating data before connecting AI, and how a custom MCP server can be repointed from Fabric to Snowflake, BigQuery, or Redshift when your platform changes.

Blue Margin builds and manages data platforms for midsize and PE-backed companies, including the custom MCP servers that connect AI assistants to governed company data. Learn how we can help you put your data to work with AI: https://bluemargin.com

Once your LLM is connected to a data lakehouse, it can reach every table your permissions allow. Kate Eberle, Director of Sales at Blue Margin, and Brick Thompson, Blue Margin’s CEO, talk about how to govern data for BI and AI analytics so the right people see the right data and read it correctly.

Kate recently presented to a group of analytics practitioners in the legal industry, and governance took over about a third of the hour. She and Brick cover data governance committees, training for self-service users, role-based security in the lakehouse, and how documented context helps an LLM return reliable answers to business users.

Chapters
00:00 Governance for BI and AI
00:40 What legal analytics teams are asking about
01:46 Permissions in a single system vs. a lakehouse
02:55 Data governance committees
04:40 Training self-service users
06:17 Supporting business users with AI analytics
06:51 Building context for reliable LLM answers
09:07 Curation and the data team’s role
10:09 Why access control matters more with LLMs
12:33 Getting busy leaders to participate in governance

About Blue Margin
We build and manage data platforms for mid-market and PE-backed companies. Our managed data services cover everything from bringing your systems together in one place to reporting and AI analytics, all built on a single governed foundation that your whole leadership team can trust. Learn more at bluemargin.com.

When your team asks the same question in Power BI, in Excel, and through an AI agent, the answers should match. That consistency comes from defining your business rules once, upstream in the lakehouse, so every tool reads from the same place.

In this episode, Brick Thompson and Landon Ochs walk through the platinum layer, which takes the business logic sitting in your semantic models and the cleansed data in your gold layer and bakes both into materialized tables an LLM can query directly. They cover what stays in the report for runtime behavior like security and date slicers, how to settle a definition of revenue that sales, finance, and the CEO all recognize, and how role-based context can serve the right version of a metric to whoever is asking.

They also get into authentication, since MCP servers can run on the same Entra security groups you already manage, and into the other agents that can read your data once it is in a consistent state.

Chapters

00:00 Same question, same answer across Power BI, Excel, and NLQ
00:53 One place where the logic lives
01:40 Where the platinum layer sits in the medallion architecture
02:20 What belongs in the semantic model at runtime
03:00 Combining gold layer data with semantic business logic
03:39 Materialized tables and defining revenue for your business
04:46 Handling competing metric definitions, including role based
05:30 Context files, MCP servers, and measures in the lakehouse
05:54 Entra authentication, tool permissions, and agents beyond NLQ
07:35 Updating business logic in one place

We're Blue Margin, a managed data services firm working with mid-market and PE-backed companies.

Learn more: https://bluemargin.com