- The reported assessment covers technical, application and service capabilities.
- A reported assessment rating does not establish universal safety or performance in every deployment.
A scoped assessment
Huawei said its AgentArts enterprise-agent platform passed an assessment organized by the China Academy of Information and Communications Technology, IT Home reported on November 4. The company described the result as the current highest rating, 4+, under the trusted-AI enterprise-agent-platform assessment.
According to the report, the evaluation followed enterprise-agent technical requirements and examined technical, application and service capabilities. Areas covered included cognition, planning, execution, business integration, knowledge management, operations and security compliance. The rating concerns that assessment's scope; it is not evidence that every possible customer deployment is safe or performs identically.
Platform operating options
IT Home also describes several ways to use AgentArts. Its managed-agent option hosts the runtime, session state and tool execution after users configure a model, prompts and tools. Customers can alternatively deploy their own agent code into its hosted runtime, including agents built with mainstream or custom frameworks.
The platform also offers observation and automated evaluation for agents, with third-party systems able to connect through OpenAPI or an SDK. These are reported platform capabilities, rather than independent benchmark results from this publication.
Sources & context
Go to the original material. Company claims remain attributed to their sources.
01
