Skip to main content
Simulations let you test your agent against realistic scenarios before those scenarios happen in production. They provide controlled, repeatable coverage for the customer behaviors you care about most.

What You’ll Learn

  • What a Simulation represents in Bluejay
  • How the simulation lifecycle works from creation to evaluation
  • How to choose the right simulation type for your use case

How Simulations Work

You use Simulations to run Digital Humans against an Agent, compare outcomes, and identify failure modes early. They are the core of pre-production testing and regression detection in Bluejay. Each Simulation is a container for a set of Digital Humans that will interact with your Agent. When you queue a run, Bluejay orchestrates each conversation, captures transcripts, and evaluates them against your Custom Metrics. Simulations are re-runnable, so you can track performance over time and catch regressions.

Key Capabilities

  • Re-runnable scenarios — execute the same simulation repeatedly to measure performance over time
  • Multi-channel support — run simulations over telephony, SIP, LiveKit, WebSocket, or HTTP webhooks
  • Batch execution — run an entire Community of Digital Humans in a single batch
  • Automated evaluation — every conversation is scored against your Custom Metrics automatically

Common Use Cases

  • Validate a new agent build against a standard set of customer scenarios before launch
  • Re-run a regression suite after prompt or knowledge base changes
  • Stress-test edge cases like angry customers, language switching, or long hold times

Next Steps

Simulations Deep Dive

Full reference for Simulation configuration and execution.

Simulation Types

Understand the different types of simulations available.

Queue a Simulation Run

Start a simulation run via the API.