BigQuery Adds Six Analytics Functions for AI Agent Skills
The built-in functions bring metric diagnostics, trend analysis and pattern discovery closer to governed data while leaving availability and permission details unspecified.
Edited by Tyronne Panaino
Google Cloud says BigQuery introduced six built-in augmented-analytics table-valued functions during its October 5–9 product window. The functions cover metric diagnostics, trend analysis and pattern discovery, and Google says they can integrate as skills for AI agents.
The change matters to data teams building agents because it places repeatable analytical operations inside BigQuery rather than describing them only as model-generated steps. Analysts, developers and platform owners can evaluate a smaller, defined function surface when deciding which data investigations an agent should be allowed to invoke.
What changed in BigQuery
Google's October 10 data-cloud roundup identifies six new table-valued functions, or TVFs, under the augmented-analytics label. The official summary groups their work into three jobs: diagnosing metrics, analyzing trends and discovering patterns directly where the data lives. It also says the functions integrate as AI-agent skills.
That description marks a practical product boundary. A table-valued function returns a table-shaped result that downstream queries and tools can consume, so packaging analytical operations this way can give an agent a more explicit callable capability than an open-ended request to interpret a dataset. The source does not describe how each function is named or configured, so those implementation details should not be inferred from the announcement.
Why the agent-skill framing matters
The important delta is not that an AI system can discuss a trend. It is that Google is presenting built-in BigQuery analytics as reusable skills for agents operating near enterprise data. In practice, that framing can help teams separate the agent's planning layer from the analytical operation it requests and the governed data on which that operation runs.
The official page does not establish that the functions improve answer quality, reduce analysis time or prevent unsafe queries. It also does not explain the permission checks applied when an agent invokes a skill. Those questions remain deployment and governance work for customers rather than proven outcomes of the release.
Evidence limits and next checkpoint
The retained Google Cloud roundup does not name the six functions, state their availability tier, list supported regions, publish pricing or performance results, or document agent-specific authorization behavior. It also supplies no adoption figures or independent evaluation. The next verifiable checkpoint is detailed product documentation that identifies the function names, invocation contract, availability and access controls.
Status
Confirmed. Google Cloud's official product roundup supports the existence and stated purpose of the six functions. Internal confidence is medium because the evidence is a single first-party summary without detailed documentation or independent testing.
Sources
Update note: Last reviewed 2026-10-10. We will revise this post if Google publishes the function list, availability details or agent-access documentation.
Sources
Drafted with AI assistance from source briefs; reviewed for citation completeness and label accuracy.