Skip to main content

Agent Lifecycle Management

Agent Lifecycle Management and Canary Release Management are critical components for safely rolling out new versions of agents to specific channels or customer groups in a multi-agent system like LMOS.

Importance of Canary Release Management​

  • Risk Mitigation: Releasing an agent to a specific channel or group allows you to detect and resolve issues before they affect all users. For example, if a new billing agent version introduces a bug, limiting its release to a small group ensures that only a subset of users experiences the problem, making it easier to contain and address.

  • Customized Rollouts: Channels may have unique requirements or capabilities. For instance, a new sales agent feature might be relevant only to customers using the web channel. Testing in this specific context ensures the feature is functional and valuable for the intended audience.

  • Performance Validation: New features or updates may introduce performance impacts. Testing in a controlled environment, such as a subset of users interacting through an IVR channel, helps verify that the new agent performs with low latency.

  • User Feedback and Iteration: Early rollout to specific groups allows gathering real-world feedback that can guide further improvements. For instance, releasing a new contract agent to a small subset of beta users in a mobile app might reveal usability issues or feature gaps that weren't evident during development.

Example of Canary Release Use Cases​

  • Web vs. IVR: Suppose a tenant has both a web channel and an IVR channel. A new customer support agent feature may work well in text-based interactions but needs adjustments for voice-based interactions. By releasing the agent update only to the web channel first, developers can refine the agent's logic for IVR scenarios before enabling it there.