Connect AI assistants to governed analytics through MCP
Give compatible assistants access to published business meaning and read-only dataset queries.
What does the Vizz MCP server do?
Model Context Protocol, or MCP, provides a way for compatible AI applications to call tools. The Vizz MCP server exposes tools to discover published definitions, inspect their fields, query datasets, and return version and health context.
Example: compare revenue and margin by channel
An assistant first finds a published dataset and inspects its available fields. It then requests a grouped query using the published revenue, gross profit, margin, and sales-channel definitions. Vizz resolves the definitions and executes a bounded, read-only warehouse query. The assistant can explain the returned result alongside its definition and version.
Keep access scoped
The MCP client authenticates with a scoped Vizz API token. Warehouse passwords are not passed to the assistant. MCP access is read-only; authoring and publication follow the platform’s review workflow. Access remains subject to the configured workspace and permissions.
Reduce guesswork without promising certainty
Human-reviewed definitions reduce the need for an assistant to invent formulas or infer business rules from column names. The assistant still interprets the question and writes the explanation, so users should examine the definitions, results, and any health warnings. The integration has been verified with Codex. Other MCP clients require compatible tooling and configuration.