​
What are AI Agent Simulations?

AI agent simulations are controlled environments where AI agents can be tested, evaluated, and refined through various scenarios and interactions.

​
Scenarios

Scenarios are structured test definitions used for simulating voice AI and chatbots to unearth potential issues and edge cases. They define the specific conditions, inputs, and expected behaviors that your AI agents will encounter during testing. For simulating voice AI and chatbots, scenarios help identify:
  • Communication breakdowns
  • Unexpected user behaviors
  • Edge cases that may cause failures
  • Performance under various conditions

​
Example: Sales Agent Selling Insurance

Consider a sales agent designed to sell insurance policies. Here are possible scenarios it might encounter:
Scenario TypeDescriptionExample
Interested CustomerCustomer actively seeking insurance”I need car insurance for my new vehicle”
Price-Sensitive CustomerCustomer focused on cost”What’s your cheapest option?”
Skeptical CustomerCustomer questioning the need”Why do I need insurance?”
Comparison ShopperCustomer comparing with competitors”Your competitor offers the same for less”
Technical QuestionsCustomer asking detailed policy questions”What’s covered under comprehensive?”
Objection HandlingCustomer raising concerns”Insurance is too expensive”
Urgent NeedCustomer needs immediate coverage”I need coverage starting today”

​
Types of Scenarios

Scenarios can be configured in four different ways:
  1. Datasets: Pre-defined collections of test data and customer profiles
  2. Graph: A graph structure containing all possible conversation edges and paths
  3. Script: Specific edge cases or targeted scenarios to simulate particular situations
  4. Agent-Generated: Automatically generated scenarios based on the agent definition and capabilities
Learn more about creating and managing scenarios in our Scenarios section.

​
Agent Definition

This is your agent that you want to test - the AI voice agent or chatbot that will be evaluated through simulations.

​
Definition

Each agent on Future AGI represents your unique AI agent. These are conceptual entities used to organize and configure your Voice Agents with specific behaviors, capabilities, and constraints within the simulation environment. Agent definitions include:
  • Agent Configuration: Core settings and behavior parameters
  • Contact Information: Phone numbers and communication channels
  • Voice Settings: Voice provider, tone, and conversation parameters
  • Capabilities: What the agent can and cannot do
  • Business Logic: Rules and workflows the agent follows
Learn more about configuring your agents in our Agent Definition section.

​
Test Agent

The test agent is the simulated user that interacts with your main agent during testing. It acts as the “customer” or “user” side of the conversation, following predefined behaviors and scenarios to test how your agent responds.

​
How Test Agents Work

Test agents are configured with:
  • System Prompts: Instructions defining the test agent’s personality and behavior
  • Voice Configuration: Voice provider, speed, and conversation settings
  • Conversation Parameters: Interrupt sensitivity, speaking patterns, and response timing
  • Model Settings: LLM model and temperature settings for generating responses
The test agent simulates realistic customer interactions, allowing you to evaluate how well your main agent handles different types of users and situations.

​
Run Tests

Run Tests orchestrate the execution of multiple scenarios against your agents in controlled environments. They combine your agent definition, test scenarios, and simulator agents to create comprehensive testing sessions. Learn more about executing tests in our Run Test section.

​
Next Steps

Learn how to get started with simulations in our Getting Started guide.