MCP for data analysis: How to connect AI assistants to governed business data
Blog post from Sigma
The Model Context Protocol (MCP) gives AI assistants such as Claude and ChatGPT a shared way to discover and query business data through servers, reducing the need for separate integrations for each assistant and system. For reliable analysis, however, the server must enforce user-specific permissions, connect to governed data models and metric definitions, and support query monitoring; MCP itself does not guarantee these safeguards. Ungoverned approaches, including pasted exports, shared service accounts, and raw tables without business context, can produce stale or inconsistent answers or expose data improperly. A cautious rollout begins with a small set of curated, clearly described sources, tests access using users with different roles, connects through per-user OAuth, and expands based on usage and unanswered questions. The article presents Sigma’s MCP Server as one implementation that applies users’ Sigma permissions, draws on its governed models and certified metrics, runs queries on live warehouse data, and can create governed workbooks, while noting that available actions depend on permissions and licensing.
| Trend | Post Mentions | Total Month Mentions | Posts | Companies | MoM |
|---|---|---|---|---|---|
| MCP | 39 | 9,757 | 959 | 203 | -18% |
| AI Agents | 1 | 6,233 | 1,230 | 245 | -11% |
| AI Coding Assistant | 1 | 1,849 | 451 | 145 | -6% |
Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.