Article · AI translation
IT Home reported on November 5 that Kimi announced the launch of Kimi Hosted Agents today, offering the Harness that supports Agent operation as a cloud service (Harness as a Service).
Enterprises and developers can obtain a 24/7 hosted environment without building and maintaining the runtime framework themselves, with support for long-running tasks and efficient collaboration among multiple agents.
Enterprises can configure Agents as needed to quickly build two types of business applications:
For internal business: Integrate them into internal enterprise systems to handle tasks such as business monitoring and operations response.
For product users: Embed them in products offered externally by enterprises to provide Agent services such as in-app investment research assistants and customer service assistants.
To ensure Agents operate safely and under control, Kimi Hosted Agents provides isolated sandboxes and network access controls. Enterprises can set operational boundaries and terminate sessions at any time when necessary. Execution records can be inspected and exported for auditing. Clear billing rules and usage reports also allow enterprises to track resource consumption and costs for each Agent project.
Kimi said that, with the Harness framework and hosted environment provided by Kimi Hosted Agents, enterprises can quickly define tasks and connect tools, shortening the development cycle from prototype validation to application launch. Kimi is responsible for maintaining and upgrading the underlying runtime environment, so teams do not need to repeatedly invest in infrastructure adaptation and operations, and can respond to business needs and iterate on application features more quickly.
Kimi Hosted Agents currently offers 140+ Skills and 30+ plugins covering tasks such as online search, specialized data queries, office document processing, and content generation. Developers can call on them as needed, reducing the development work required to integrate tools one by one. Enterprises can also connect their own tools and data sources to adapt agents to specific business processes.
Kimi Hosted Agents can be used with Kimi Inference Hook, a separate security component of the open platform, to connect to security services built or purchased by enterprises. Before each model inference, it checks task inputs and content returned by tools, intercepting private and sensitive data before inference.
Enterprises define the security check rules, and Kimi Inference Hook enforces whether content is allowed or rejected: only content that passes an enterprise's security checks enters model inference.
Kimi's open platform documentation site also provides supporting Agent architecture design guidance, configuration tutorials, and runtime cost estimates. Enterprises can copy Agent templates into their own project spaces, connect resources such as internal business systems and private databases, adjust task workflows as needed, and quickly build agents suited to their business.
The first batch includes 8 financial Agent templates covering tasks such as investment research modeling, in-depth research, morning report generation, IPO tracking, and position monitoring, with support for continuous operation in the hosted environment. Enterprises can copy the Agent templates into their own project spaces, connect resources such as internal research reports and private databases, and adapt them to actual business processes.
IT Home lists Kimi Hosted Agents pricing as follows:
Total cost = sandbox runtime (hours) × ¥0.072/hour + model inference costs + Σ (number of plugin calls × price per plugin call)