Meta Unveils Agent-Based Data Warehouse Access Solution at @Scale
Meta has unveiled a novel, agent-based solution for data warehouse access and security at the @Scale conference on August 13, 2025. This system transforms hierarchical data into a language model-friendly format, enabling efficient communication with large language models.
The solution introduces a multi-agent system comprising data user agents and data owner agents that negotiate access requests autonomously while upholding security protocols. The data user agent consists of three sub-agents: one suggesting alternatives when access restrictions occur, another enabling low-risk data exploration, and a third assisting with permission requests. Meanwhile, the data owner agent includes two sub-agents focused on security operations and access management.
Performance metrics show impressive results, with an overall recall rate of 90%. Acceptance and rejection recall rates stand at 73% and 100% respectively, indicating efficient user access and reduced data owner workload. The system exhibits intention management capabilities through explicit and implicit methods, modeling business needs and user activities. Context management differentiates between automatic, static, and dynamic contexts to enable more precise access decisions. Partial data preview functionality addresses data exploration by implementing context-driven decisions, fine-grained query-level permissions, data access budgets, and rule-based safeguarding.
Meta's agentic solution aims to simplify data access for billions of users and tens of thousands of engineers. By leveraging artificial intelligence agents, the system transforms hierarchical data into a text-based format compatible with large language models. While Meta has not disclosed specific security measures, the solution promises efficient user access and reduced data owner workload, with a high recall rate and intention management capabilities.
