Google Cloud's Gemini Enterprise can now query a Supabase project in plain language, fetching live answers without storing or indexing the data.
Supabase is now a prebuilt connector inside Google Cloud’s Gemini Enterprise, letting anyone with access to the platform ask a plain-language question about a Supabase project and get an answer pulled live from the database, the company said in a post on its blog (Supabase blog, Sept. 9, 2026).
What shipped #
Gemini Enterprise is Google Cloud’s search-and-agent platform for internal company data, already wired into GitHub, Linear, Notion, Slack, Salesforce and Jira. Supabase joins that connector list: a prompt like “list my tables from Supabase for [project]” now routes to the Supabase API and returns a live result inside Gemini Enterprise, alongside whatever else those other connectors surface in the same answer (Supabase blog, Sept. 9, 2026).
Supabase is explicit that this isn’t a data pipeline: Gemini Enterprise doesn’t store or index a project’s data, and every response is fetched in real time from the customer’s own Supabase organization (Supabase blog, Sept. 9, 2026). Google’s own connector documentation confirms the mechanism. It’s a federated data store that sends the search query directly to the Supabase API rather than importing anything, with each tool call annotated read-only or destructive so an admin can see what an agent is allowed to touch (Google Cloud, Gemini Enterprise Supabase connector docs).
Setup differs by plan. Business-edition customers get Supabase in the connector menu with just a sign-in; Standard, Plus and Frontline editions need an admin to register it with a client ID and secret. Authentication runs through Google-managed OAuth, so an organization doesn’t stand up its own OAuth app, and the connection is authorized once at the org level rather than per user (Supabase blog; Google Cloud connector docs). Google’s Rayn Veerubhotla, managing director for partner technical architects, framed the pitch: “Connecting Supabase’s MCP Server to Gemini Enterprise means your backend is part of that conversation, grounded in live data from your actual projects” (Supabase blog, Sept. 9, 2026).
The connector runs on Supabase’s own MCP server, the same server Supabase hardened with org-managed auth three weeks earlier (Supabase blog, “Enterprise-managed auth for the Supabase MCP server,” Aug. 24, 2026). Gemini Enterprise is the first outside platform to plug into that hardened MCP endpoint instead of a one-off integration built just for it.
The take #
For a small backend team, this mostly saves a Slack message to whoever has dashboard access. The bigger audience is IT and ops staff who never touch Supabase directly today. They can now ask “how many rows are in the orders table” or check a schema without a login, because the query, not the credential, moved into the chat window. Routing it through the already-audited MCP server, instead of building Gemini a separate path into Supabase’s API, means one set of auth and read/write annotations to secure instead of two. The next test is what happens when another AI platform wants the same access. Gemini Enterprise is the first to lean on Supabase’s MCP server this way, and it likely won’t be the last, which means a flaw in that one endpoint stops being Supabase’s problem alone.
Key Takeaways #
- Supabase became a prebuilt Gemini Enterprise connector on Sept. 9, 2026, alongside GitHub, Linear, Notion, Slack, Salesforce and Jira.
- Queries run live against the Supabase API; Gemini Enterprise doesn’t store or index the underlying data.
- Business-edition customers can enable it with a sign-in; Standard, Plus and Frontline editions need an admin to configure a client ID and secret.
- The connector runs through Supabase’s own MCP server, the same server Supabase added enterprise-managed auth to on Aug. 24, 2026.