Overview
Roark provides multiple ways to execute simulation tests, from one-off manual runs to fully automated scheduled testing. Whether you’re validating a quick fix or running comprehensive regression tests, you can choose the execution method that best fits your workflow.Before running simulations, you need to have a configured run plan that defines your test matrix (scenarios, personas, agent endpoints, and evaluators). Learn how to create one in our Run Plans documentation.
Execution Methods
Manual Execution
Run simulations on-demand directly from your dashboard:- From Run Plans - Execute existing test configurations
- Create and Run - Build new test suites and run immediately
- Re-run Previous Tests - Repeat past simulations with one click
Scheduled Execution
Automate testing with recurring schedules:- Set up hourly, daily, weekly, or monthly runs
- Configure timezone-aware scheduling
- Monitor execution history and results
Running from Run Plans
The most common way to execute simulations is through run plans, which define your test configuration.Execute Existing Run Plan
- Navigate to Simulations → Run Plans
- Find your run plan in the list
- Click the Run button
- The system will:
- Generate all test combinations
- Queue simulations for execution
- Begin running based on your concurrency settings
- Track progress in real-time
Create and Run
When creating a new run plan, you can execute it immediately:- Click Create Run Plan
- Configure your test matrix:
- Select scenarios
- Choose personas
- Pick agent endpoints
- Add evaluators
- Choose Create and Run instead of just Create
- Tests begin executing automatically after creation
The “Create and Run” option saves time by combining setup and execution into a single workflow.
Re-running Simulations
Easily repeat previous test runs to verify fixes or track improvements:Re-run Options
-
From Run History
- Navigate to Simulations → Runs
- Find the previous run
- Click the Re-run button
- Executes with identical configuration
-
From Run Details
- Open any completed run
- Click Re-run in the header
- Automatically uses the same run plan configuration
Benefits of Re-running
- Verify bug fixes with identical test conditions
- Track performance improvements over time
- Quickly validate configuration changes
- Compare results between runs
Execution Configuration
Key Settings
When running simulations, you can configure:| Setting | Description | Range |
|---|---|---|
| Iterations | Times to repeat each test | 1-100 |
| Max Concurrent Jobs | Parallel execution limit | Based on quota |
| Direction | Inbound or Outbound | See below |
Inbound vs Outbound Considerations
Inbound Simulations:- Execute immediately when triggered
- Support multiple iterations
- Roark calls your agent
- Fully automated execution
- Require manual dialing action
- Limited to 1 iteration
- Your agent calls Roark
- Provides phone numbers to call
Monitoring Execution
Real-time Progress
Track your simulation runs as they execute:-
Status Indicators
Pending- Queued for executionCreating Snapshots- Preparing test configurationsRunning Simulations- Actively executing callsCompleted- Successfully finishedFailed- Execution encountered errors
-
Progress Metrics
- Number of simulations completed
- Current execution count
- Estimated time remaining
- Success/failure rates
Execution Dashboard
Navigate to Simulations → Runs to view:- All active executions
- Recent run history
- Status filtering options
- Detailed metrics for each run
Viewing Results
Run Summary
After execution completes, view comprehensive results:- Overall success rate
- Average evaluation scores
- Total duration
- Individual test case results
Detailed Analysis
Click into any run to see:- Test case breakdown
- Conversation transcripts
- Evaluation scores
- Error details (if any)
- Performance metrics
Comparing Runs
Use re-run functionality to:- Track improvements over time
- Identify regression issues
- Validate bug fixes
- Monitor performance trends
Execution Limits
Concurrency Limits
Your organization has maximum concurrent simulation limits:- Check your quota in organization settings
- System automatically optimizes based on your limits
- Upgrade your plan for higher concurrency
Best Practices for Performance
Start with Lower Concurrency
Start with Lower Concurrency
Begin with 5-10 concurrent jobs to test your agent’s capacity
Monitor Agent Load
Monitor Agent Load
Watch your agent’s response times during high-concurrency runs
Use Iterations Wisely
Use Iterations Wisely
Multiple iterations help identify intermittent issues
Schedule Off-Peak Runs
Schedule Off-Peak Runs
Run heavy test suites during low-traffic periods
Quick Start Examples
Running Your First Simulation
-
Create a Run Plan
- Navigate to Simulations → Run Plans
- Click Create Run Plan
- Select one scenario, one persona, one agent
- Choose Create and Run
-
Monitor Progress
- Watch status change from
PendingtoRunning - View real-time completion updates
- Watch status change from
-
Review Results
- Check overall success rate
- Examine individual test results
- Review evaluation scores
Setting Up Automated Testing
-
Build Your Run Plan
- Create comprehensive test matrix
- Include multiple scenarios and personas
- Add relevant evaluators
-
Create Schedule
- Navigate to Simulations → Schedules
- Select your run plan
- Configure timing (e.g., daily at 2 AM)
- Activate schedule
-
Monitor Automated Runs
- Check schedule dashboard for results
- Set up notifications for failures
- Review trends over time