November 5, 2026 – Huawei Cloud’s AgentArts intelligent-agent platform has become the first product to pass the Trusted AI enterprise-grade agent platform assessment organized by the China Academy of Information and Communications Technology (CAICT), earning the current highest rating of 4+.
The evaluation was carried out under the Enterprise-grade Agent Technical Requirements standard and examined the platform across three dimensions: technical capability, application capability and service capability. It covered agent cognition and planning, task execution, business integration, knowledge management, operations and maintenance, and security compliance. The certificate was issued on November 2, 2026, and the rating applies to the product status as of October 30, 2026.
AgentArts is built around four core capabilities. For production-grade long-running tasks, it uses a Serverless elastic architecture and the Harness runtime framework to support continuous agent operation, automatic recovery from exceptions, and uninterrupted execution in complex multi-step scenarios. On the security side, it provides session isolation, end-to-end permission control and sandboxed execution, while maintaining high concurrency and resource efficiency for large-scale deployment.
The platform also supports agent engineering self-optimization, using reinforcement learning to improve task success rates and reduce token consumption. It has accumulated knowledge assets and scenario solutions in finance, automotive, energy and other sectors. For operations, it offers unified observability across logs, metrics and call traces, enabling alerting and full-link troubleshooting to lower enterprise maintenance costs.
Huawei Cloud is pursuing a dual-track strategy with its commercial AgentArts platform and the open-source openJiuwen version, which share the same architecture and core. The platform has already been adopted by customers in finance, coal mining, transportation, manufacturing and power.
The result arrives as Chinese regulators and standards bodies accelerate work on agent evaluation frameworks. CAICT has been expanding its Trusted AI agent assessment system to cover platforms, technical capabilities, applications and industry scenarios, making such ratings one of the early reference points for enterprise agent procurement.
