Conversational analytics in BigQuery brings trusted agentic reasoning to everyone

Google launched Conversational Analytics in BigQuery to general availability, enabling natural language queries, multi-step analyses, and visual reports directly on governed data. The feature uses Gemini models and BigQuery's security framework to provide inspectable, grounded answers, aiming to democratize data analysis across organizations.

Businesses run on fast decisions, but the teams who hold the answers are often buried under a backlog of routine requests, leaving users waiting in line for insights they need now. Today, we are bringing Conversational Analytics in BigQuery to general availability, so both business and technical teams can query data, run multi-step analyses, and generate visual reports using natural language, right where the data lives. With this release, Conversational Analytics in BigQuery now delivers an agent that behaves like an analyst who knows your business, thinks before it answers, and stands behind its work. Built on Google’s latest Gemini models and BigQuery’s secure, governed foundation, it brings that trusted analyst to everyone in your organization. BigQuery’s conversational capabilities are built-in and available for use instantly, with no setup required. For deeper, more consistent insights, data professionals can author specialized agents grounded in the exact sources that matter, from projects, datasets, and tables to views, graphs, and user-defined functions. And because your data rarely lives in one place, Conversational Analytics reaches beyond native BigQuery tables to Lakehouse-managed Apache Iceberg tables and cross-cloud Lakehouse sources like Databricks Unity, AWS Glue, SAP and Salesforce, so you can break down data silos and analyze data across clouds from a single conversation. As a data practitioner, you work with Conversational Analytics right inside BigQuery Studio and Data Canvas, and publish the agents you build to Gemini Enterprise, Data Studio, or your own application through the Conversational Analytics API, putting them in the hands of business users wherever they work. “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.” - Suzie Millar, Head of Data, Mony Group Accuracy in Conversational Analytics is by design, not aspirational: every agent is grounded in your business context, not a model's assumptions. That context comes from the Knowledge Catalog https://cloud.google.com/blog/products/data-analytics/introducing-the-google-cloud-knowledge-catalog glossaries, profile scans, and context bundles , BigQuery Graph for multi-hop queries, and your own verified queries and custom agent instructions. With the new Open Knowledge Format https://cloud.google.com/blog/products/data-analytics/how-the-open-knowledge-format-can-improve-data-sharing , the wiki your team already maintains can feed straight into Knowledge Catalog. At query time, Conversational Analytics leverages existing embeddings of your column values, generated by AI.GENERATE EMBEDDINGS, to match your question to the right data, so asking about "Texas" finds rows stored as "TX." Grounding only earns trust if the user can see it. So every answer is inspectable, providing: Visible thinking steps: Review the agent's step-by-step reasoning and the exact SQL it generates before it returns an answer. Context citations: See the precise sources behind every response, including tables, schema definitions, verified queries, and glossary terms used to calculate it. Proactive disambiguation: When a prompt is vague, the agent asks targeted clarifying questions instead of guessing. Long-term memory: The agent remembers what your terms and questions mean, so you don't have to disambiguate the same thing twice. One common barrier to scaling AI is governance. Reaching tens of thousands of users requires rigorous security, governance, and transparent cost controls https://docs.cloud.google.com/gemini/data-agents/conversational-analytics-api/manage-costs . Conversational Analytics inherits BigQuery's governance model, so users only query data they are authorized to see and every query is logged for auditing within the BigQuery compliance framework. On top of that baseline, it supports Access Transparency AxT https://cloud.google.com/security/products/access-transparency?hl=en , Customer-Managed Encryption Keys CMEK https://docs.cloud.google.com/kms/docs/cmek , Private IP https://cloud.google.com/vpc/docs/private-google-access , and VPC Service Controls https://docs.cloud.google.com/vpc/docs , and now guarantees data residency https://docs.cloud.google.com/assured-workloads/docs/data-residency for data at rest and for ML processing within EU and US multi-region endpoints. For your most engaged users, we also deliver the operational controls that scale demands: Configure Google Cloud-native cost controls so no user or project exceeds its allotment, cap an agent's maximum query size in bytes, and track usage through BigQuery labels on jobs. The agent doesn't just retrieve rows, but calls BigQuery's AI functions for you, turning advanced analysis into a question you can ask in plain language. Find the "why," not just the "what": Ask what drove a change and the agent runs root-cause analysis with AI.KEY DRIVERS, surfacing the exact segments behind the move. See what's coming: Move past historical reporting by triggering AI.FORECAST and AI.DETECT ANOMALIES right in the chat to project trends and flag outliers, with no model to build or manage. Query your entire data estate: With object tables, the agent reasons over relational data and unstructured files together, PDFs, images, logs, and video, so a single conversation spans your whole estate. Conversational Analytics agents are moving from human-scale reactive analysis to agent-scale proactive action. You're no longer limited to asking a question and waiting for the answer. Deep-dive mode: If you ask ‘Why a metric moved?’ the agent will build its own analytical plan, mapping the critical questions, working through a full multi-step investigation with no manual SQL, and minimizing analytical blind spots. The result is a comprehensive report you can download and share. Agentic workflows: Deploy autonomous agents that monitor your data, reason over events, run multi-step workflows on a schedule, and deliver insights straight to your chat. You can set up a Monday-morning business report or daily anomaly detection across key metrics, each with a custom directive so they investigate only what you care about. General availability of Conversational Analytics in BigQuery marks an official exit from the static dashboard era. By embedding Gemini’s deep cognitive reasoning directly into the data warehouse, we are enabling a self-managing environment that transforms raw data into active, corporate knowledge. This delivery is a key component of the Agentic Data Cloud, providing a true system of action that moves past retrospective reporting, incorporates security and governance by design and is engineered for enterprise trust. If you are ready to get started, learn more from our documentation https://docs.cloud.google.com/bigquery/docs/conversational-analytics , reach out to your Google Cloud account representative, or get started in BigQuery Studio https://console.cloud.google.com/bigquery today to build and deploy your first agent.