Natural language questions in open-source Metabase

Metabase’s open-source platform has included Metabot, an AI assistant for natural-language data questions and SQL generation, plus an MCP server and Agent API, since version 60, with the instructions reflecting version 64. Metabot can find existing questions or build new charts and queries, lets users inspect and modify generated query-builder or SQL logic, and works best when tables, fields, metrics, models, transforms, and glossary terms are clearly described. Self-hosted users can connect providers such as Anthropic, OpenAI, Bedrock, Gemini Enterprise, and others, while Metabase Cloud also offers a metered Metabase AI service; prompts, schema details, and sample values are sent to any connected external provider. Access is governed by each user’s existing Metabase permissions, including through OAuth-authenticated MCP connections for clients such as Claude, Cursor, Codex, and ChatGPT, which can search content, execute permitted queries or SQL, and, where allowed, create or modify Metabase objects. Pro and Enterprise add verified or curated content restrictions, group-based AI controls, token limits, system prompts, and usage analytics. Metabot can also operate in Slack after account linking, and developers can embed natural-language analytics in their own products through the permissions-aware REST Agent API.

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.