Dell links supply chain security to the challenges of agentic AI

Dell is pushing supply chain security to the front of the AI security debate. The company says that as enterprises move AI into core business operations, they have to protect not just data and infrastructure, but the systems that can now act on their own.

That shift matters because agentic AI changes the threat model. A chatbot answers a prompt. An agent takes a goal, works through multiple steps, reaches for tools, touches systems of record and talks to other agents. That kind of autonomy makes control harder, and it moves questions about identity and permission into the center of the stack.

Kamlesh Shah, director of product management for Dell AI Factory Solutions, said Dell starts with supply chain security before a product is even envisioned, then keeps that protection going through the full life cycle of a deployment. He said Dell has been doing this through its AI Factory and inside its own IT operations, giving customers a model they can copy instead of building everything from scratch.

The permissions piece is especially tight. Shah said an agent should have identity tied to a specific workload, not broad access across applications, and its actions should be limited to narrowly scoped, time-based permissions. Around that, Dell recommends policy checks, runtime sandboxing and telemetry that shows who authorized what.

Ash Lakshmanan, director of product management for Dell Services, said the company’s work spans hardware, runtime, data and applications, but customers often underestimate observability across all four layers. He said older security tools were built for human identities, and even machine identities, not agent identities. Dell’s testing has already found chatbot deployments that were not ready and could have been exploited. The company is also working with Nvidia to move agentic AI from the desk to the data center.