Overview

Test AI agents and prompts through controlled simulations before deploying to production.

About

Simulation lets you test voice and chat agents against simulated customers before going live. You define agent definitions (how to connect to your agent), scenarios (what the customer wants), and personas (who the customer is). The platform runs the conversations, records transcripts, and scores them with evaluations.

How Simulation Connects to Other Features

  • Evaluation: Scores every simulated conversation automatically. Learn more
  • Observability: Simulation traces flow into Observe so you can replay and debug conversations. Learn more
  • Optimization: Use simulation results to improve prompts with Fix My Agent. Learn more
  • Datasets: Create scenarios from datasets, and export results back for further analysis. Learn more

Getting Started

Run Simulation

Run your agent against scenarios from the platform.

Chat Simulation Using SDK

Run chat simulations programmatically.

Prompt Simulation

Test prompts in multi-turn conversations.

Fix My Agent

Get AI-powered diagnostics and fixes.
Was this page helpful?

Questions & Discussion