Dell Technologies has announced updates to its AI Data Platform, introducing a Unified Semantic Layer and an Enterprise Knowledge Graph. These tools are designed to provide AI agents with structured context from existing enterprise data. The new features aim to improve the accuracy of AI responses by standardizing business definitions and mapping data relationships across various systems.
The platform utilizes Nvidia Nemotron Retriever models for reasoning and visual understanding. To ensure data security, these components operate within the customer's own data center. Knowledge Agents are restricted to specific data slices, with customers maintaining control over access permissions and resource allocation.
Performance testing on a PowerEdge R770 server equipped with Nvidia RTX PRO 4500 Blackwell GPUs showed that Apache Spark jobs ran 3.9 times faster on average compared to CPU-only processing. In specific batch data mining tasks, the company recorded a 20.4-times speedup. The system uses Apache Arrow for efficient data movement between storage and the processing engine.
Dell is also enhancing its PowerScale storage, which will support up to 500 tenants per cluster with improved role-based access control and mutual TLS encryption for file traffic. Additionally, a new Storage Performance Tool is now available to help enterprises benchmark S3-compatible object storage for AI workloads.
The Unified Semantic Layer, Enterprise Knowledge Graph, and Knowledge Agents are scheduled for release in the first half of 2027. PowerScale security updates are expected in November, followed by the accelerated Data Processing Engine in December. Further Apache Arrow optimizations are planned for the first half of next year.
