Salesforce Docs Portal

Create a Test Suite

A test suite is the starting point for evaluating your agent. Name your test suite, select the agent to test, and choose whether you're evaluating individual responses or full end-to-end conversations.

Required Editions

green checkmark

This article applies to:

 New Testing Center in Agentforce Studio (Beta)
red crossmark

This article doesn’t apply to:

 Legacy Agentforce Testing Center in Setup
Available in: Lightning Experience
Available in: Enterprise, Performance, Unlimited, and Developer Editions. Required add-on licenses vary by agent type.
User Permissions Needed
To create tests in the Testing Center:

Manage Agentforce Grids

AND

Manage Agentforce Testing
  1. To begin tests from Agentforce Studio, select Tests. You’ll see your existing test suites including the agent and version each suite is configured to test. To create a test suite, select New Suite.
  2. Enter a name. Optionally, add a description.
    A new test suite with the conversation level testing scope selected.
  3. Select the agent and version you want to test.
    Note
    Note Agent version and testing considerations:
    • The agent you select to test impacts the test conditions, type, and data available for testing. To generate test scenarios from your knowledge base, the agent must be linked to a data library containing that information. Similarly, to generate tests for subagents or actions, the agent must be assigned those subagents and actions.
    • The version selector defaults to the latest version of your agent. To test a different version, select it from the dropdown.
    • Switching versions clears the current results in the test worksheet, but previous results remain available in test run history.
    • Your agent version must be committed to appear in the Testing Center.
    • Draft agent versions and Agentforce Employee Agents aren't supported for conversation-level testing.
  4. Under Test Scope, select how much of the interaction you want to measure and score.
    • Turn-Level. Evaluates a single agent response in isolation. Use turn-level testing to validate prompt changes, verify reasoning quality at a specific step, or confirm accurate knowledge retrieval.
    • Conversation-Level. Evaluates a full multi-turn conversation from start to finish. Use conversation-level testing to assess whether your agent achieves the user's goal across an entire interaction.
  5. Click Next.