Home / Connectors / Google BigQuery
Data warehouse · Direct adapter

Google BigQuery × ServAI

Connect Google BigQuery to the ServAI Data Intelligence Role Pack with explicit data boundaries, read-only execution, human approval and traceable evidence.

Start a connection assessmentAll connectors
CategoryData warehouse
Connection routeDirect adapter
Data modeRead-only
Publish authoritySeparate human release
What can be connected

The Google BigQuery data you approve—not the whole account.

ServAI can read approved analytical datasets, modeled facts, dimensions and aggregate tables. An administrator defines the allowed datasets, fields, query windows and sensitivity ceiling. Exact availability depends on the customer’s Google BigQuery plan, permissions and source configuration.

Boundary ruleThe agent cannot select an arbitrary account, table or resource from its prompt. Scope must be registered and approved first.
What Data Intelligence can do
  • Query governed analytical models
  • Reconcile approved measures across dimensions
  • Analyse trends, cohorts, funnels and contribution
  • Draft dashboards with query and lineage evidence
How the connection works

Configured, admitted and health-gated.

Implementation differs by connector state. These are the actual boundaries for this ServAI catalogue entry.

01

Tenant configuration

An administrator supplies the endpoint, vault-held read-only credential and explicit dataset allow-list.

02

Fingerprint admission

A person reviews the driver, host, database, TLS options and dataset scope, then admits the exact fingerprint.

03

Health-gated reads

Bounded plans run only after health passes, using bound parameters against approved tables or views.

Current implementation detailsNative BigQuery adapterProject-pinned schema and named parametersBounded SELECT plans with LIMIT
Governed path

From source to a verifiable dashboard.

Connecting a source does not grant authority to change data, define metrics or publish.

  1. 01Configure connection
  2. 02Review & admit
  3. 03Approve meaning
  4. 04Read-only query
  5. 05Dashboard & evidence
  6. 06Different-person verify
  7. 07Separate release
What you need

Deployment requirements

  • Reachable network endpoint with TLS
  • Read-only service account held in the vault
  • Explicit approved table or view allow-list
  • Human owner for admission and health
What this connector cannot do

Fixed boundaries

  • ServAI does not write to Google BigQuery or change source records.
  • No self-admission, self-approval of metrics or self-verification of dashboards.
  • The agent cannot submit raw SQL or arbitrary resource names.
  • A catalogue listing never means a customer connection is operating.
Status truth

Operational after admission & health

The customer control surface shows actual per-tenant state: operational, needs attention, or catalogued—not live. ServAI never turns a logo into a connection claim.

OperationalAdmitted and health passing
Needs attentionConfiguration, health or entitlement needs action
Catalogued—not liveNo admitted customer connection
Google BigQuery + Data Intelligence Role Pack

Start with data scope—not a connector logo.

We will establish route state, requirements, human ownership and data boundaries before treating the connection as ready.

Start a connection assessmentExplore the Role Pack