Google rebuilds its data system for AI agents, not humans — Agentic Data Cloud announced at Cloud Next

Google Cloud announced the Agentic Data Cloud on Wednesday, a new architecture designed for AI agents (self-directed AI tools) that work around the clock instead of humans running occasional data queries. The system has three components: Knowledge Catalog (automatically organizes data structure using past query patterns), Cross-cloud lakehouse (lets BigQuery access data stored on AWS without extra fees), and Data Agent Kit (lets engineers describe desired outcomes in VS Code instead of hand-coding data pipelines).

The key difference: traditional data systems wait for a human to ask a question and run a query; this one lets AI agents autonomously decide what data to fetch and what actions to take in response, 24/7. Data engineers now describe business goals instead of writing hundreds of lines of pipeline code — the AI figures out the steps.

For data teams, this means less time spent writing and maintaining code for moving data around, and more time on strategic work. For companies relying on real-time decisions (fraud detection, inventory management, pricing), autonomous agents can now act instantly without waiting for a human analyst to notice a problem and run a report.