Google opens BigQuery conversational analytics to users
Wed, 1st Jul 2026 (Today)
Google has made Conversational Analytics in BigQuery generally available, adding natural language data analysis to its cloud data warehouse.
The product lets business and technical users query data, run multi-step analyses and generate visual reports in BigQuery using natural language. It is available without setup, and data teams can also configure it to create more specialised agents tied to specific data sources.
Those sources extend beyond native BigQuery tables. The service works with Lakehouse-managed Apache Iceberg tables and cross-cloud Lakehouse sources, including Databricks Unity, AWS Glue, SAP and Salesforce, allowing users to analyse data from multiple environments through a single interface.
Data practitioners can use the feature in BigQuery Studio and Data Canvas. They can also publish agents to Gemini Enterprise, Data Studio or their own applications through the Conversational Analytics API.
A customer example from the retail and finance comparison market was included as part of the launch. "At MoneySuperMarket, BigQuery Conversational Analytics has changed how our teams get to insight. Analysis that used to take weeks can now be done in minutes, saving our financial analysts around half a day each week. By making analysis more self-serve, we're helping teams create faster insight to support better product and commercial decision-making," said Suzie Millar, Head of Data, Mony Group.
Trust controls
As it expands generative AI tools into data analysis, Google is emphasising auditability and traceability. Each agent is grounded in business context drawn from sources such as Knowledge Catalog, BigQuery Graph, verified queries and custom instructions.
Teams can also feed internal wiki material into Knowledge Catalog through its Open Knowledge Format. At query time, the system uses existing embeddings of column values to match user questions to the right data, including when a plain-language term differs from the stored value.
Users can inspect how an answer was produced. The product shows step-by-step reasoning, exposes the SQL generated by the agent, cites source material such as tables and glossary terms, asks clarifying questions when prompts are vague and retains contextual memory so repeated disambiguation is not required.
Security model
Governance is another focus as Google seeks wider adoption across large organisations. Conversational Analytics inherits BigQuery's existing access controls, meaning users can query only the data they are authorised to see, while all queries are logged for audit purposes.
The service also supports Access Transparency, Customer-Managed Encryption Keys, Private IP and VPC Service Controls. Google added that data residency is guaranteed for data at rest and for machine learning processing within EU and US multi-region endpoints.
Administrators can set cost controls at the user or project level, cap an agent's maximum query size in bytes and track usage through BigQuery job labels. These controls are aimed at organisations trying to extend natural language querying to large groups of users without losing oversight of spending and access.
AI functions
Beyond standard query retrieval, the feature can call BigQuery AI functions in response to user questions. That allows people to ask for the likely drivers behind a metric change, forecast trends or detect anomalies without manually building models or writing SQL.
The service can also work across relational data and unstructured files through object tables, including PDFs, images, logs and video. The goal is to let users investigate structured and unstructured information within the same conversation.
From queries to workflows
The general availability release also introduces a broader analytical workflow. In a deep-dive mode, users can ask why a metric moved, and the agent will generate its own analytical plan, work through a multi-step investigation and produce a report for download and sharing.
Google is also pushing the product towards scheduled and autonomous use cases. Organisations can deploy agents to monitor data, reason over events, run multi-step workflows on a schedule and deliver findings directly into chat-based tools.
Examples included a Monday morning business report and daily anomaly detection for key metrics, each driven by custom directives. This positions the product as more than a conversational interface for dashboards, and closer to a managed analysis layer inside the data warehouse.
The launch reflects a wider contest among cloud providers to make enterprise data stores easier to use through generative AI while preserving access controls, auditability and cost discipline. Google's approach ties those tools closely to BigQuery, with natural language interaction, source inspection and workflow automation built into the existing platform.