SandBoxes for Agentforce and Data Cloud
Teams wishing to test Agentforce must do it in safe, segregated environments. Salesforce Sandboxes, which are mirror versions of your production org’s data and customizations, now supports Data Cloud and Agentforce.
By recreating the organization’s data and metadata in a risk-free environment, development teams can quickly create their unstructured data foundation and rigorously prototype Agentforce without affecting business operations.
Teams can now perform UAT (User Acceptance Testing) with an initial set of users to ensure that Salesforce Agentforce performs the tasks it is designed to do, and then migrate those changes to production using familiar tools like Change Sets, DevOps Center, and the Salesforce CLI, which now support Data Cloud and Agentforce.
Monitoring and Observability for Agentforce
With the wide availability of Data Cloud Sandboxes, the entire Einstein Trust Layer can be evaluated in a safe, pre-production environment, allowing for quick configuration of Salesforce Agentforce agents and Prompt Templates. Teams can leverage the Einstein Trust Layer’s audit trail and feedback storage in sandboxes to create a closed loop for AI testing, iterating on prompts and actions in response to user feedback.
AI-Generated Test for Agentforce
Teams working with Agentforce must appropriately test all of the various ways a customer may ask a question or engage with an agent. In addition to Agent Builder, which includes a Plan Tracer for evaluating an agent’s reasoning process, the new Agentforce Testing Center allows teams to test subject and action selection on a large scale.
Using natural language instructions, the Testing Center may construct hundreds of simulated interactions such as requests a customer might make while interacting with an Agentforce Service Agent, and test them in parallel to evaluate how frequently they produce the desired result.
Transparent usage monitoring in Digital Wallet
Data Cloud Sandbox and Salesforce Agentforce usage are metered in Digital Wallet, giving customers comprehensive visibility on their consumption throughout the AI development life cycle. New upgrades provide specific insights into which functionalities utilize credits, allowing teams to identify new consumption trends as they scale.