The US Army issued a request for information on June 15 seeking agentic AI building and management platforms that can be deployed on classified military networks, according to ExecutiveGov. The RFI, posted on SAM.gov by Army Materiel Command, targets platforms capable of operating on the Joint Worldwide Intelligence Communications System (JWICS) and SIPRNet at Impact Level 6. Responses are due July 3.
What the Army Wants
The platform must enable intelligence analysts and personnel to develop, deploy, monitor, and oversee AI agents that can retrieve information and conduct tasks in classified environments. A critical requirement is support for “edge and disconnected, intermittent, or limited bandwidth environments,” per the SAM.gov listing referenced in the ExecutiveGov report.
That edge requirement separates this procurement from typical enterprise AI deployments. Commercial agentic AI platforms assume persistent cloud connectivity, high-bandwidth data access, and real-time model inference. Military intelligence operations in forward-deployed or contested environments have none of those guarantees. An agent that cannot function when the network drops is not useful to a signals analyst operating from a tactical operations center.
The Broader Procurement Context
The agentic AI RFI is one of ten contract opportunities the Army published in June 2026, according to ExecutiveGov. The others span command-and-control interfaces for corps-level commanders, a Sovereign Defense Cloud for high-performance computing, synthetic training environments, robotic construction equipment, and an uncrewed aircraft system marketplace.
The agentic AI and Sovereign Defense Cloud RFIs are connected. The cloud initiative, run by the Engineer Research and Development Center’s Information Technology Laboratory, seeks hybrid cloud models integrating government-owned and commercial systems, with “AI-powered orchestration to automate workload and data management” and “vendor lock-in prevention that prioritizes open standards and modular architectures.” An agent governance layer would sit on top of that infrastructure.
The Constraint Problem
Building agentic AI for classified environments introduces constraints that commercial vendors rarely face. IL6 certification requires that all data, including metadata and telemetry, remain within DoD-controlled infrastructure. Models cannot call external APIs. Agent tool-use must be auditable to a degree that exceeds even EU AI Act requirements. And the “disconnected, intermittent, or limited bandwidth” requirement means agents must be capable of local inference and decision-making without phoning home.
The Army’s 2026 Army Summit, scheduled for June 18, will feature keynotes from Katie Thompson of the Army Contracting Command and Lt. Gen. Robert Collins of the Office of the Assistant Secretary of the Army for Acquisition, Logistics and Technology, per ExecutiveGov. The summit is expected to address how the service plans to integrate agentic AI into its intelligence operations workflow.