Dell's New Gateway Puts Data Intelligence at the Center of Enterprise AI
Dell Technologies has introduced an Agentic Data Gateway to control how AI agents access enterprise data, addressing the challenge of providing AI systems with trusted context while maintaining governance and oversight.

Enterprise organizations are discovering that artificial intelligence systems require more than computational power to deliver results. The ability to feed AI agents accurate, governed data has emerged as the critical differentiator in building functional AI factories, as companies move beyond the initial wave of cloud-based large-scale AI systems to deploy similar capabilities within their own infrastructure.
Dell Technologies Inc. has responded to this shift by enhancing its AI Data Platform with semantic and knowledge graph capabilities. The core challenge, according to Srinath Alapati, senior director of product marketing for AI Data Platform at Dell, involves combining structured and unstructured data sources while preserving human oversight and security controls. "The bottom line is if you feed the wrong data, you get the wrong outcomes," Alapati said. "What we've found is in our own testing — by giving the right context — you can improve the accuracy from roughly 60% to like 94%."
Controlling data access through orchestration
Dell's newly announced Agentic Data Gateway operates as a centralized control point within the orchestration layer, preventing AI agents from accessing enterprise systems without appropriate safeguards. The gateway leverages a Model Context Protocol server and draws on storage intelligence capabilities that catalog what data exists within an organization, who manages it and its volume. Dell's data intelligence features track data lineage and provenance across hybrid environments spanning edge locations, cloud platforms and on-premises data centers.
A key distinction in Dell's approach involves handling data distribution differently than competing platforms. Rather than requiring customers to consolidate all data in a single location, Dell transmits only metadata while keeping data in place. "They want you to bring all the data to wherever the platform is. That's not always possible," Alapati said. "Number one, it's super expensive. Two is data is always being generated. It's a never-ending game. You can't always keep bringing in data, consolidating data. What we do is we only send metadata … That way your data stays where you want to be."
Infrastructure management and deployment speed
Dell has expanded its infrastructure management capabilities through Dell OpenManage Rack Scale, which extends operational oversight beyond individual compute systems to encompass network infrastructure, power distribution and cooling systems. According to Alison Biers, senior director of product marketing for compute, networking and global industries at Dell, this unified management approach enables organizations to move from initial deployment through production operations on a consistent operational plane.
The company has redesigned its fabric management software, replacing SmartFabric Manager with an intent-based Dell Fabric Manager. PowerRack systems now treat entire racks as single deployment units rather than managing individual components. The result is a dramatic acceleration in time to production. "It is literally six hours from the time it hits your loading dock to full production," Biers said. "Something that used to take maybe months and months [is] now cut in half, maybe even less."
Growing enterprise adoption and ecosystem expansion
Dell's AI Factory platform now serves more than 6,500 customers, with an increasing proportion being large enterprises seeking production deployment assistance. According to Adam Glick, senior director of AI solutions product marketing for compute at Dell, validated designs developed in partnership with Nvidia Corp. and other technology providers encompass both hardware and complete software stacks. Dell is expanding its laboratory facilities to accommodate independent software vendors joining the ecosystem.
https://www.youtube.com/embed/w5WCA_7dIks?feature=oembed
While compute infrastructure typically receives the most attention in AI factory discussions, Glick emphasized that processing power represents only one component of a complete system. "If you've got a factory, compute is your assembly line. It's the important part," Glick said. "It's what you think of when you think of 'factory', but that assembly line doesn't run if you don't have a warehouse that's feeding it the raw materials — that's the data that goes into it."