Metadata Type: GenAiPlannerBundle
GenAiPlannerBundle is a metadata type in Salesforce (API version 64.0 and later) that represents an AI Planner for Agentforce (Einstein/GenAI). It acts as a container for all the topics (GenAiPlugin) and actions (GenAiFunction) used by an AI Agent to interact with a large language model (LLM).
Overview
The GenAiPlannerBundle metadata type was introduced in API version 64.0 as part of Salesforce’s GenAI and Agentforce enhancements. It allows administrators and developers to retrieve, deploy, create, update, and delete AI Planner bundles via the Metadata API. This ensures planners can be version controlled and deployed in CI/CD processes :contentReference[oaicite:0]{index=0}.
Key Features
- Container for topics (GenAiPlugin) and actions (GenAiFunction)
- Defines agent behavior and conversational flows
- Supported via Metadata API for automated deployment and versioning :contentReference[oaicite:1]{index=1}
- Integrated with Einstein Bot or Agentforce framework
Structure and Components
A GenAiPlannerBundle instance typically includes:
- Plugins: GenAiPlugin entries defining topics or conversational areas
- Functions: GenAiFunction entries defining actionable behaviors invoked by the planner
- Bundling: A single bundle groups a planner’s full set of topics and actions
Deployment Considerations
When deploying GenAiPlannerBundle metadata, administrators should consider the following:
1. Dependency Management
Ensure all associated GenAiFunction and GenAiPlugin components are included in the deployment; missing dependencies will cause failures :contentReference[oaicite:2]{index=2}.
2. API Version Support
Use Metadata API version 64.0 or higher; earlier versions only support GenAiPlanner (deprecated in 64.0) :contentReference[oaicite:3]{index=3}.
3. Related Bot Configuration
If deploying a new planner, include the Einstein Bot and BotVersion metadata to ensure the agent is visible and functional post-deployment :contentReference[oaicite:4]{index=4}.
4. Permissions
The deployment user must have access to GenAI/Agentforce, Einstein Bot, and Metadata API functionality.
Best Practices for Salesforce Administrators
1. Version Control
Store planner bundle XML in Git to enable change traceability and rollback capabilities.
2. Bundle Dependencies Together
Deploy GenAiPlannerBundle alongside its Plugin and Function dependencies, as well as related Bot and BotVersion metadata.
3. Sandbox Testing
Test deployments in sandbox environments first to validate conversational flows, plugin invocation, and function execution.
4. Document Planner Contents
Keep clear documentation of included plugins, functions, and their roles in agent conversations.
5. Collaborate With AI/Dev Teams
Involve AI architects and developers to ensure plugin/function logic aligns with intended agent behavior and user experience.
6. Audit and Monitor
Periodically review planner bundles to remove obsolete topics/actions and to align with evolving business objectives.
7. Log and Handle Errors
Capture deployment or runtime errors related to missing dependencies or function failures, and integrate logging to aid debugging.
8. Optimize Planner Scope
Keep the bundle focused on essential topics and actions to simplify deployment complexity and improve performance.
9. Manage Planner Versions
Use meaningful planner bundle names and version numbers to distinguish between major updates in agent capabilities.
10. Provide Training & Docs
Ensure team members understand the GenAiPlannerBundle structure and deployment process, emphasizing dependency inclusion and versioning practices.
Conclusion
The GenAiPlannerBundle metadata type enables Salesforce teams to programmatically manage AI Planner configurations for GenAI agents. By following best practices—such as dependency bundling, version control, testing, and documentation—administrators can deploy reliable, scalable, and maintainable planner bundles that power intelligent conversational agents within Salesforce.