Beijing Releases Measures to Accelerate AI Agent Development: Nine Key Directions

Beijing's municipal government has published a series of measures to accelerate the leading development of AI agents (intelligent systems with autonomous perception, memory, decision-making, interaction, and execution capabilities). The policy, drafted by multiple city departments including the Development and Reform Commission, outlines nine key directions: improving foundational model capabilities (e.g., agentic AI, world models), strengthening agent-level common technologies (e.g., middleware, frameworks, security), accelerating native applications and benchmark scenarios (e.g., AI-native software, scientific intelligence, intelligent terminals), promoting the integration of intelligent terminals with agents, supporting new entrepreneurship models like OPC (one-person company), fostering Token economy (including Token-as-a-Service), enhancing safety governance (e.g., classification-based regulation, safety infrastructure), ensuring key factor guarantees (e.g., computing power, finance, data, talent), and encouraging open-source development. The measures also establish a powerful leadership group, increase financial support, and strengthen promotion. The policy is intended to drive the evolution of production paradigms and lifestyles, and to build Beijing into a global AI innovation highland.