Use BigQuery with AI via MCP
A data warehouse only pays off once business teams can ask it questions – without SQL and without a detour through the BI team. Google Cloud provides an official, generally available MCP server for BigQuery. The connection to CompanyGPT runs through Cloud IAM or OAuth; no API key is involved.
What is BigQuery?
BigQuery is the serverless data warehouse of the Google Cloud Platform. It separates storage from compute, handles very large tables with SQL queries and is typically used as the central analytics layer in a company – fed from ERP, CRM, web tracking, production systems and other sources.
As a result, BigQuery often holds the most dependable view of the business. That makes it a particularly valuable source for AI agents: answers rest on figures actually queried from the warehouse rather than on estimates, and business questions can be answered without building a new dashboard every time.
Connection at a Glance
- MCP server
- Official (Google Cloud), remote – generally available
- Endpoint
- bigquery.googleapis.com/mcp
- Transport
- Streamable HTTP
- Authentication
- Cloud IAM / OAuth, no API key
- Operating model
- SaaS (Google Cloud, not Google Workspace)
- Permissions
- IAM roles at dataset and table level remain authoritative
How to Connect BigQuery to CompanyGPT
MCP access runs through bigquery.googleapis.com/mcp using streamable HTTP. Unlike the Google Workspace servers, this endpoint is generally available and therefore intended for production use. BigQuery is a Google Cloud service; a Workspace agreement is not required.
Authentication uses Cloud IAM or OAuth – no API key is used. In effect that is a security advantage: access is tied to a concrete principal, meaning a user account or a service account with defined IAM roles. Your existing permission model down to dataset, table and column level continues to apply, as does row-level security where configured.
For a first step we recommend a service account with read permissions on selected datasets plus a cost limit for queries. Inside CompanyGPT, roles and groups determine which departments may use the BigQuery integration – controlling and sales operations, for example, while other areas stay out of scope.
Typical Use Cases
How AI agents in CompanyGPT work with BigQuery
Answer business questions
Sales management asks how revenue for a product group developed by region – the agent formulates the query against the approved datasets and explains the result.
Investigate anomalies
Controlling traces an unexpected deviation in the monthly report down to the underlying records instead of parking it until the next review.
Understand the data model
New members of the analytics team have the structure and meaning of existing tables explained and find the right source for an analysis faster.
Frequently Asked Questions
BigQuery and AI – the most important answers
Does BigQuery have an official MCP server?
Yes. Google Cloud provides it at bigquery.googleapis.com/mcp via streamable HTTP. It is generally available and not in preview.
How is access secured?
Through Cloud IAM or OAuth. No API key is used. Access runs via a user or service account with defined IAM roles, so your existing permission model applies unchanged.
Can the AI run arbitrary queries and drive up cost?
Scope follows from the IAM permissions granted. We recommend read roles on selected datasets plus configured query and cost limits in Google Cloud so the boundaries are clear from the start.
Are the results traceable?
Yes. Executed queries are logged in BigQuery, and the agent can surface the query it used in the chat. That keeps it verifiable what a statement is based on.
Is BigQuery part of Google Workspace?
No. BigQuery belongs to the Google Cloud Platform. The integration works regardless of whether your company uses Google Workspace.
Related Integrations
Google Workspace & Google Cloud
BigQuery and Your AI – GDPR-Compliant with CompanyGPT
CompanyGPT is the enterprise AI assistant in your own cloud tenant: all relevant AI models, your company data and integrations like BigQuery – centrally managed, with roles and permissions. We set up the connection together with you.
BigQuery is a trademark of its respective owner (Google Cloud). Its mention describes compatibility and does not imply any partnership or endorsement. Availability and scope of the connection depend on the respective vendor.