RFI for Agentic AI Platforms

The issuing agency is seeking information on agentic Artificial Intelligence building and management platforms that can be deployed on accredited classified networks. The agency needs to enable intelligence analysts and staff to build, deploy, monitor, and govern AI agents capable of reasoning, retrieving information, and taking actions via tool and function calling. The goal is to solve the problem of managing and securing autonomous AI agents operating within highly sensitive, classified environments. The requested scope of work covers detailed platform capabilities, including agent authoring approaches such as no-code, low-code, and code-first, as well as supported agent patterns like single-agent, multi-agent, and graph-based systems. The agency requires information on supported foundation models, specifically identifying the country of origin, developer, and ownership status for each model. The platform must support tool and function calling with allow-listing and human-in-the-loop controls for sensitive actions, alongside retrieval-augmented generation over enterprise data that incorporates classification-aware access control. Security and integration requirements include identity integration with Department of Defense Public Key Infrastructure and Identity, Credential, and Access Management, detailed audit logging, and guardrails such as prompt injection defenses, content filtering, and policy enforcement. Eligibility is strictly limited to United States organizations. The agency requires the performance of the work to be conducted offsite. The agency is currently looking for proposals to understand market capabilities and has not disclosed a specific budget or key dates in the initial request.