Google Cloud's Agentic Data Cloud Enhances Cross-Cloud Data Access for AI Agents

Google Cloud has announced key updates to its Agentic Data Cloud, focusing on enhancing cross-cloud data access and providing richer business context for AI agents. A core component of this update is the "Borderless Lakehouse" feature, which allows AI agents to query remote Iceberg data residing in AWS and Azure without incurring egress fees. This is achieved through secure private links, with retrieved data blocks being cached within Google Cloud. Additionally, the Knowledge Catalog enrichment feature, currently in preview, scans unstructured data from sources like Google Workspace, Jira, SharePoint, and GitHub. This operational knowledge is then converted into context sets, providing agents with a deeper understanding of organizational information. Google reports that early adopters of this enrichment saw a fourfold increase in agent accuracy and up to a 75% reduction in token consumption. This development is significant for practitioners because it directly tackles some of the most persistent challenges in multi-cloud and AI deployments: data gravity and the cost of data egress. Historically, moving large datasets between cloud providers has been expensive and complex, creating silos that hinder the effectiveness of AI models that require comprehensive data access. By enabling agents to query data in place across AWS and Azure without egress fees, Google Cloud is lowering the barrier to entry for multi-cloud AI strategies and making it more feasible to build AI applications that leverage data distributed across different environments. This move aligns with the broader industry trend towards multi-cloud and hybrid cloud architectures, where organizations seek to leverage the best services from different providers while avoiding vendor lock-in. The increasing adoption of AI and agentic workflows further amplifies the need for seamless, cost-effective data access across these disparate environments. Other cloud providers have also been making strides in multi-cloud connectivity, such as AWS Interconnect – multicloud and Azure Multicloud Interconnect, emphasizing the industry-wide recognition of this critical need. In practice, this means that data engineers and AI developers can spend less time on complex data integration pipelines and more time on building and refining AI agents. The ability to access data across clouds without incurring significant egress costs can lead to more accurate and robust AI models, as agents can draw upon a wider and more diverse range of information. Practitioners should investigate how these new capabilities can be integrated into their existing multi-cloud data strategies, particularly for use cases involving large datasets spread across AWS, Azure, and Google Cloud. Furthermore, the Knowledge Catalog's ability to provide business context is crucial for grounding AI agents in reality, reducing hallucinations, and improving the trustworthiness of AI-driven insights and actions.