To act at the speed of business, AI agents must operate in fast and trusted reasoning loops. They need to “think” by reasoning across both your historical context and your live operational reality. Only by understanding this complete, real-time picture can they “do” — taking immediate action.
For decades, data architectures have been built with a structural wall that breaks this loop, separating the platforms that generate insights from the platforms that manage actions. This latency means insights are gleaned after the critical window for an agent to take action has closed. Achieving true AI transformation requires organizations to move from a passive system of record to a proactive System of Action, built on a closed-loop architecture that converges operational and analytical data.
At Google Cloud Next, we announced new unifying capabilities that drive our Agentic Data Cloud, eliminating silos and enabling 98% of our largest data cloud customers to run operational and analytical workloads in a unified data platform. By operating AlloyDB, BigQuery and Spanner together, we are delivering an AI-native architecture that unlocks the full potential of your data for real-time, agentic applications.
Flexible, real-time data agents
To act effectively, agents require both operational and historical signals for sound decision-making. Our Agentic Data Cloud bridges the gap between the operational “now” and analytical history by handling the complex plumbing for you. We provide diverse integration models across data federation, reverse ETL, and real-time ingestion to the lakehouse, empowering your agents to make high-stakes decisions with both live context and historical depth.
For example, sometimes an agent driving a live operational application needs to pull historical context on demand. Through Lakehouse federation for AlloyDB (Preview), agents can access Lakehouse data directly from AlloyDB itself. This allows frontline systems to instantly query extensive historical data without relying on brittle data movement pipelines.
In other scenarios, the challenge is reversed: deeply complex historical insights have already been calculated in the data warehouse, but an agent needs to deliver them to millions of users at conversational speeds. Reverse ETL for BigQuery (Preview) provides a one-click solution to push these heavy analytical insights back into AlloyDB, Bigtable, or Spanner, enabling agents to serve them with sub-millisecond latency.
One-click reverse ETL
Teams running real-time analytics on live operational data typically have to move that data into analytical systems — an error-prone process that introduces lag. With Spanner Columnar Engine (GA) users can perform analytical queries that run up to 200 times faster with zero impact on production transactional workloads.
Finally, the reasoning loop is not complete until an agent’s real-time action is captured for downstream analysis. To close this loop, Datastream for Lakehouse Apache Iceberg tables provides real-time Change Data Capture (CDC) from AlloyDB, Cloud SQL, Spanner, and Oracle directly into the open Lakehouse. This process streams every operational change as an append-only event into Lakehouse tables, making that data immediately available in BigQuery for ML model training, feature engineering, and real-time analytics.
“AlloyDB, along with other Google Cloud products like BigQuery, provides the agility and performance needed to continually enhance our platform’s capabilities and help us anticipate emerging trends rather than merely reacting.” – Javi Fernández, CTO, Loyal Guru
Grounding agents in a unified governance foundation
Inconsistent definitions and unclear data ownership across operational and analytical systems can cause agents to hallucinate. To address this, we are extending Knowledge Catalog (Preview), formerly Dataplex, with new integrations for AlloyDB, BigQuery, Bigtable, Cloud SQL, and Spanner to provide a unified map of your data landscape. Integrations with Oracle AI Database@Google Cloud and Firestore are coming soon. The Knowledge Catalog works by aggregating native context across your Google and partner data platforms, semantic models, and third-party catalogs, unifying them into a single, governed source of truth needed to build and scale reliable agents.
“Seven-Eleven Japan created “Seven Central,” a scalable data platform that uses Spanner and BigQuery to provide real-time insights and support the company’s digital innovation strategies. We collect data from all 21,000+ stores, and in anticipation of a future expansion in business operations, we have designed a system that can scale up and run without issue, even if we were to have 30,000 stores, with 1,000 customers per store per day.”
-Izuru Nishimura, Executive Officer and Head of ICT Department, Seven-Eleven Japan
Unified engines for deep reasoning
To move beyond simple Q&A chatbots to autonomous agents, AI must reason across every dimension of your data estate. Historically, combining keyword search, semantic understanding, and relationship mapping meant moving data out of operational databases and into specialized, siloed search engines — introducing latency and complexity.
Google’s Agentic Data Cloud eliminates these silos. By embedding native vector and full-text search directly into operational databases like AlloyDB, Bigtable, Cloud SQL, Firestore, and Spanner, agents can execute highly accurate hybrid searches combining keyword relevance and semantic intent.
We’re also bringing together graph and vector support across BigQuery and Spanner. With graph federation, an agent can match live user intent in Spanner and immediately trace that intent through historical graph relationships in BigQuery Graph — accelerating autonomous decision-making without moving the data. This multi-model approach powers advanced GraphRAG patterns, equipping agents with the rich, interconnected context required to accelerate autonomous decision-making.
“To deliver AI that actually works across HR, payroll, and workforce operations, you need a consistent, real-time data layer. With the power of Google’s Agentic Data Cloud, People Fabric is the backbone of UKG’s Workforce Operating Platform — turning fragmented systems into a single source of truth that powers intelligent, agent-driven experiences.”
-Radhi Chagarlamudi, Group Vice President, Product Engineering, UKG
Built for performance at agent scale
Our Agentic Data Cloud delivers the closed-loop architecture required for the AI era without compromising operational performance. Built on open standards like Iceberg and PostgreSQL, and governed by universal semantics, Google Cloud provides the speed, throughput, and trusted context needed to build the next generation of conversational and autonomous applications.
- Build: Explore the AlloyDB AI documentation to start grounding your agents.
- Connect: Visit the BigQuery Console to set up your first federated query to Spanner.
- Govern: Opt-in to the Knowledge Catalog for unified visibility.
