> ## Documentation Index > Fetch the complete documentation index at: https://docs.getstrada.com/llms.txt > Use this file to discover all available pages before exploring further. # Simulations > Test your agents with different personas and scenarios Simulations allow you to test your agents comprehensively by running conversations with different personas and scenarios. Unlike [test cases](/building/testing), simulations enable you to test full conversation flows with realistic user personas, helping you validate agent behavior across various interaction patterns and use cases. *** ## Creating a Simulation To create a new simulation, navigate to the **Simulations** section and click **Create Simulation**. You'll configure the following components: Create New Simulation ### Simulation Details **Simulation Name** - Give your simulation a descriptive name that clearly identifies its purpose (e.g., "Customer Support - Billing Issues"). **Intent Goal** - Define what the conversation should accomplish in this simulation. Be specific about the desired outcome and what success looks like. **Success Criteria** - Specify how you'll measure success, with one criterion per line. These criteria help evaluate whether the simulation achieved its intended goal. **Expected Turns** - Set the expected number of conversation turns. This helps the simulation understand the typical length of conversations for this scenario. You can also click **Import from Template** to start from a pre-built security template (e.g., Prompt Injection Defense, Self-Harm Prevention) or an organization-specific template. ### Configure Personas Personas represent different user types and communication styles that will interact with your agent during the simulation. You can: * **Import from Template** - Use pre-configured personas from existing templates * **Create New** - Build custom personas that match your target audience Multiple personas can be added to test how your agent handles different user types, communication styles, and interaction patterns. ### Tool Mocks (Optional) Tool mocks allow you to configure mock responses for agent tools during simulation runs. This is useful for: * Testing agent behavior when tools return specific responses * Simulating external system interactions * Controlling tool outputs to test different scenarios Add mocks to control how tools respond during simulation execution, enabling you to test various edge cases and system states. *** ## Running Simulations Once your simulation is configured, you can run it to see how your agent performs across different personas and scenarios. ### Simulation Results Page The simulation results page displays: * **Simulation Details** - Shows your intent goal, success criteria, and key metrics including: * Number of personas configured * Personas passing rate (indicates how many personas met the success criteria) * Last run time and total number of runs * Expected turns * **Simulation Results** - Results grouped by persona, with each persona card showing: * Run status (Passed/Failed) with visual indicators * Run timing information (when it ran, duration) * Version information (agent version used for the run) * Total runs for that persona You can **Run All Personas** to test all configured personas at once, or re-run individual personas. Use the **View** button on any run to see detailed results. ### Viewing Runs When viewing a specific run, you'll see: * **Summary** - An AI-generated summary of how the conversation went * **Success Criteria & Completion** - Whether the success criteria were met, with a criteria score percentage and detailed explanations * **Key Metrics** - Total turns, duration, and start time * **Conversation Transcript** - The complete conversation between the persona and your agent, including tool calls This detailed view helps you understand exactly how your agent performed and identify specific areas for improvement. View Simulation *** ## Best Practices * **Be specific with intent goals** - Clearly define what the simulation should accomplish * **Use measurable success criteria** - Each criterion should be a separate, testable metric * **Test multiple personas** - Include different user types to validate agent behavior across various communication styles * **Use tool mocks strategically** - Mock tool responses to test how your agent handles different system states * **Set realistic expected turns** - Base this on actual conversation patterns you've observed