Luzmo MCP server: governed analytics for AI agents

2:00 min

Expert

The Luzmo MCP server exposes analytics capabilities to an agent, so it can discover available data, ask analytical questions, or generate visualizations as part of a broader conversation or workflow. Luzmo becomes the governed analytics layer inside that agent, rather than the entire agent itself.

In this article, you will learn:

What the Luzmo MCP server does and when to reach for it
The two common architectures for connecting an agent to Luzmo
Worked examples of MCP in a customer-facing product
Why MCP is a different choice from calling the Luzmo API directly

What problem MCP solves

A product assistant might need a documentation tool to answer a product question, a CRM tool to retrieve account context, and Luzmo to analyze that account's usage or performance data. The agent picks the right tool for each part of a user's request; MCP is what lets it pick Luzmo for the analytics step, without you having to hand-build that integration yourself.

Example: combine CRM context with product analytics

Suppose a sales or customer-success agent has access to both a CRM tool and Luzmo. A user asks:

Which of my accounts renewing in the next 60 days have declining product adoption? Show their eight-week usage trend and prepare a short briefing for each one.

A possible workflow:

The agent uses the CRM tool to identify accounts owned by the current user that renew within 60 days.
It uses Luzmo through MCP to analyze the approved usage metrics for those accounts.
Luzmo returns the data-backed comparison or visualization.
The agent combines the CRM context and analytics into a briefing.
If the user wants to email or save the result, the agent asks for confirmation before taking that action.

Luzmo handles the analytics step; the surrounding agent coordinates the full workflow.

Example: continue an analysis outside your application

A user may start with an embedded dashboard in your product, then open an AI tool and ask:

Show revenue by region for this quarter and compare it with the same quarter last year.

Through a proxied Luzmo MCP server, the user authenticates through your product and the request is scoped to their access. Where MCP Apps are supported, the response can include an interactive Luzmo visualization instead of a static text summary, and a save-back action can be offered if your product implements and authorizes that workflow.

Two common architectures

Add analytics to an agent you own. Connect your agent to Luzmo's hosted MCP server and let it call Luzmo when a question requires data analysis.
Expose analytics through your own MCP server. Proxy Luzmo's MCP server behind your product's MCP experience, so you retain your own authentication flow, branding, and user context.

Why MCP instead of calling the API directly?

The API remains the right choice for deterministic application logic and for developers building or managing Luzmo resources directly. MCP is the better fit when an AI agent needs to decide when and how to use analytics on behalf of an end user — it carries the user's identity, access, and product context into the request, so end users never need to manage Luzmo API keys themselves.