Overview

Understand how simulations are used for testing in Bluejay

Simulations are controlled evaluation runs that let you test your agent against realistic customer scenarios before those scenarios happen in production. They are designed to be repeatable, measurable, and easy to compare over time.

What You’ll Learn

• How to create and configure a simulation
• How the simulation lifecycle works end to end
• How to review results and iterate on your agent

How Simulations Work

Use Simulations to exercise an Agent with Digital Humans, edge cases, and targeted goals so your team can catch failures early and ship with more confidence.

Each Simulation contains a set of Digital Humans that interact with your Agent through the integration channel you configure (telephony, SIP, LiveKit, WebSocket, or HTTP). When a run completes, every conversation is evaluated against your Custom Metrics and the results are available in your dashboard and via the API.

Running a Simulation

Dashboard

1. Navigate to Agents & Simulations in the sidebar
2. Select your Agent and click Create new simulation
3. Add Digital Humans with scenarios and success criteria
4. Click Create and start to queue the run
5. Review results in the simulation detail view

API

1. Add your agent to Bluejay if you haven’t already
2. Create a simulation to group your Digital Humans so you can run them in parallel
3. Generate or Create Digital Humans — make sure to attach the Simulation ID so they are associated with the Simulation you created in Step 2
4. Queue a simulation run to trigger the simulation — think of this as the Run button

Next Steps

Simulation Types

Understand voice, text, and regression simulation categories.

Simulation Runs

Learn about run lifecycle, statuses, and result retrieval.

Simulation Results

Interpret results, metrics, and compare across runs.

Tool Calls

Enrich evaluations with tool call tracking and metadata.