Google Builds a Borderless Data Layer for Autonomous AI Agents

Google is reshaping its Agentic Data Cloud around a clear idea: autonomous AI workers need direct, trusted access to business data wherever that data lives. The company’s latest updates add a borderless Lakehouse, self-learning enrichment for the Knowledge Catalog, new Gemini machine learning capabilities, and a developer toolkit called the Data Agent Kit.

Google launched the Agentic Data Cloud earlier this year at Google Cloud Next. The platform is meant to serve as the architectural “system of action” for AI agents, replacing brittle extract, transform and load-based data pipelines, or ETL, that feed legacy software applications.

In a blog post, Sirish Chandrasekaran, Google’s vice president of product management for data analytics, outlined the company’s view that autonomous agents will soon become the primary consumers of data instead of humans. “That is why we built the Agentic Data Cloud — a system of action designed to activate all your data by breaking down walled gardens, deliver high accuracy and trust, and operate proactively at low cost,” Chandrasekaran said.

Data access without cloud boundaries

The main advantage of the Agentic Data Cloud is that it meets customers where their data is. AI agents can access data wherever it lives, while the platform focuses on low latency, zero data movement, and automated governance.

Its flagship new feature is borderless Lakehouse, a special version of Google’s “cross-border Lakehouse” product that also debuted at Cloud Next. Built on the open-source Apache Iceberg table format, borderless Lakehouse aims to remove multicloud friction through new caching capabilities for Amazon Web Services and Microsoft Azure.

Companies can query remote data held in those cloud platforms over a secure private link without paying egress fees. When data blocks are retrieved, borderless Lakehouse caches them within Google Cloud, which speeds up later queries against the same information.

A new AlloyDB integration extends that approach to database workloads. Users can query external Iceberg tables with standard PostgreSQL queries without copying the data first, keeping the information in its existing location while making it available through familiar tools.

Teaching agents what business data means

Access alone does not make an AI agent useful. The Knowledge Catalog acts as a map of business meaning, giving agents the organizational context they need to understand how a company defines and uses its data.

The Knowledge Catalog is getting a self-learning enrichment capability in preview. It scans unstructured data in Google Workspace, Jira, SharePoint, and GitHub, then translates operational knowledge into Open Knowledge Format context sets. Those context sets can ground AI agents in the organization’s own knowledge.

Chandrasekaran said early adopters found that the capability can make the average agent four-times more accurate and reduce token consumption by up to 75%. The Knowledge Catalog is also being enabled by default in the Gemini application starting November 04 2026.

Because the Knowledge Catalog is tightly integrated, it can address business metric definitions and access controls before translating a user’s prompt into Structured Query Language queries. Companies can ground agents in semantic models for specific data platforms, including Google BigQuery, Snowflake, SAP, Looker, and dbt.

That context is meant to keep agents aligned with the formulas and rules that business teams already use. “Agents are only as accurate as the business context they’re grounded in,” Chandrasekaran said. He added, “Whether an agent is calculating net revenue for finance or customer churn for operations, it applies the exact same governed formulas your business teams rely on.”

Gemini takes on data engineering and reporting

Gemini is also gaining autonomous machine learning capabilities designed to streamline everyday work for technical teams and business workers. Data engineers can describe a specific business outcome in natural language, and Gemini will generate the PySpark code, run the notebooks, and fix data pipeline errors as it works toward that goal.

Business users will be able to request reliable, real-time operational reports in Gemini by asking for them. Gemini uses operational reporting skills integrated with BigQuery and the Knowledge Catalog to construct and save the query.

“Gemini uses operational reporting skills integrated with BigQuery and the Knowledge Catalog to construct and save the query,” Chandrasekaran said. Once the query is saved, teams can run reports on demand without incurring token costs, “guaranteeing consistent and verified answers every time.”

Google is also making the new Data Agent Kit generally available from today. The kit brings Model Context Protocol tooling into integrated development environments such as VS Cod, giving developers another part of the Agentic Data Cloud alongside its data access, governance, context, and Gemini capabilities.

Taken together, the updates position the Agentic Data Cloud as a foundation for autonomous workers that can reach distributed data, understand an organization’s business language, build data workflows, and produce repeatable reports. Google’s approach centers on keeping data where it already lives while giving agents the context and controls needed to act on it.

Stay connected via Google News

Follow us for the latest AI News