Decentralized Artificial Intelligence Through Controlled Emergence Service

The issuing agency is seeking research and development services to establish decentralized artificial intelligence through controlled emergence. The agency aims to move beyond brittle, centralized orchestration and ad hoc compositions of AI agents by harnessing self-organization. The core problem is ensuring these self-organizing agent collectives remain predictable, doctrinally consistent, mission-aligned, and resilient to both benign failures and adversarial compromise. The scope of work requires combining artificial intelligence, control theory, formal methods, and game theory to develop decentralized multi-agentic AI collectives. The agency expects the research to investigate innovative approaches that enable revolutionary advances in science, devices, or systems. Ultimately, the project must develop the theory and algorithms for decentralized coordination and local inference control to enable a scalable, adaptive, and resilient collective of heterogeneous AI agents that can autonomously execute sustained, long-time-horizon missions in contested environments. Bidders must be onshore US organizations to be eligible for this opportunity. The performance of the requested work will take place offsite. The agency is currently looking for proposals to determine the budget and has not signaled a specific funding limit. Bidders must submit all questions regarding this opportunity no later than August 18, 2026.