Google’s New Data Agents Supercharge AI Workflows

Google just dropped a whole swamp-load of updates to its Agentic Data Cloud, and if you’re building AI workflows, you’ll want to paddle over. The new data agents and tools aim to make AI agents smarter, more accurate, and way easier to build—without drowning in complexity.

Why Data Agents Matter

Generic AI platforms often lack context from enterprise databases, leading to inaccurate responses and security gaps. Google’s solution? Infuse AI across the entire stack—from custom silicon to Gemini models—so agents can ground their reasoning in real-time data with near-100% accuracy.

Conversational Analytics Gets a Boost

Google is expanding Conversational Analytics across its Data Cloud, letting you chat with your databases in natural language. Here’s what’s new:

• BigQuery Conversational Analytics (preview): Integrates an AI reasoning engine into BigQuery Studio, helping teams go beyond manual SQL with multimodal synthesis and deep-dive research.
• Lakehouse Conversational Analytics (preview): Query distributed data lakes across AWS, Azure, and Google Cloud using natural language—no data movement required.
• AlloyDB, Spanner, Cloud SQL (preview): Start natural-language conversations with your operational databases for real-time insights.
• Looker Embedded Conversational Analytics (GA): Embed agents into custom apps via low-code iframe, with a new API for multi-turn workflows.

New Data Agents for Every Role

Google announced a slew of agents to automate and accelerate data work:

• Data Engineering Agent (GA): Automates pipeline building and maintenance, transforming natural language into optimized SQL/Python code.
• Data Science Agent (preview): Suggests features, generates notebook code, and automates documentation.
• Database Observability Agent (preview): Proactively monitors performance and recommends fixes.
• Database Onboarding Agent (preview): Recommends the best database for your use case and guides provisioning.
• Looker Dashboard Agent (preview): Enables conversational interaction with dashboard data and AI-generated summaries.
• Conversational Analytics in Gemini Enterprise (preview): Brings governed intelligence to business leaders via a single interface.
• Data Insights Agent (preview): Queries structured and unstructured data across Workspace and third-party apps.
• Deep Research Agent (preview): Builds comprehensive research plans with verifiable citations.

Tools for Building Custom Agents

Google also released developer tools to make agent building smoother:

• Data Agent Kit (preview): Standardized skills and tools for IDE/CLI environments.
• Managed MCP Servers for Databases (GA): Fully managed infrastructure to connect AI models to AlloyDB, Spanner, Cloud SQL, Bigtable, and Firestore.
• Managed MCP Server for Looker (preview): Query Looker’s semantic models from any MCP client.
• MCP Toolbox for Databases 1.0 (GA): Stable, production-ready toolbox with improved docs.
• QueryData (preview): Near-100% accurate natural language to SQL for Cloud SQL, AlloyDB, and Spanner.
• UCP Analytics powered by BigQuery (preview): Stream real-time commerce events for agentic commerce observability.

These updates make Google’s Agentic Data Cloud a formidable player in the AI agent space. For more on how Google Cloud stacks up, check out our Google Cloud Review. And if you’re comparing AI model costs, see our Model Pricing Comparison.

Original announcement published on Google Cloud.