Google has added Conversational Analytics to its BigQuery data warehouse, which it says will allow enterprise data teams and business users to ask questions about data in natural language, in turn speeding up data analytics for AI use cases.

The agent, currently in preview, can be found under the Conversations tab in the new Agents Hub inside BigQuery and activated by pointing to data tables. It expands the data warehouse’s current text-to-SQL capabilities, analysts say.

“BigQuery already offers features like data canvas to make query generation and visual exploration easier. What changes with the agent is not the ability to ask questions in simple language, but the ability to carry a contextual conversation with the data over multiple steps, which can be defined as conversational analytics,” said Abhisekh Satapathy, principal analyst at Avasant.

“Instead of treating each prompt as a one-off request, the new agent remembers what was asked earlier, including datasets, filters, time ranges, and assumptions, and uses that context when answering follow-up questions. This lets users refine an analysis progressively rather than starting from scratch each time,” Satapathy added.

Satapathy pointed out that this eases the pressure on developers to prebuild dashboards or predefined business logic for every possible question that a data analyst or business user could ask.

“Rather than encoding every scenario upfront, teams can let the agent interpret user intent dynamically, while still enforcing access controls, metric definitions, and governance rules already defined in BigQuery,” he said.

Ability to build and deploy custom agents via API endpoints

In addition to the agent, Google has also added tools to build, deploy, and manage custom agents across applications and operational workflows via API endpoints to the Agent Hub.

These tools, according to Satapathy, address three practical enterprise needs: “It reduces duplication of analytics logic across tools, ensures consistent definitions and policies across all analytics users, and centralizes access control and auditing rather than implementing them separately in each application.”

The reduction in duplication also frees up developers as they no longer have to rebuild logic to interpret user questions, map them to datasets, apply security rules, or explain results, Satapathy added.

Custom agents can also be deployed via Looker, which has a built-in conversational analytics feature.

Continued Text-to-SQL improvements

Over the past few months, Google has been adding natural language and SQL abilities to BigQuery to help developers and data analysts with SQL querying.

Earlier this month, it previewed a Comments to SQL feature that is aimed at enabling developers and data analysts to write natural-language instructions in SQL comments and have them translated into executable queries inside BigQuery Studio via Gemini.

Last November, Google added three new managed AI-based SQL functions — AI.IF, AI.CLASSIFY, and AI.SCORE — to help enterprise users reduce the complexity of running large-scale analytics, especially on unstructured data. In August, Google made incremental updates to the data engineering and data science agents in BigQuery.

Rivals, such as Snowflake and Databricks, have also been prepping natural language to SQL capabilities for their offerings.

While Databricks already offers AI Functions that can be used to apply generative-AI or LLM inference directly from SQL or Python, Snowflake provides AI_PARSE_DOCUMENT, AISQL, and Cortex functions for document parsing, semantic search, and AI-driven analytics. Other warehouses, such as Oracle’s Autonomous Data Warehouse, also support AI workflows alongside SQL.