Agents are the core AI assistants you create and configure in Chatmode. They handle conversations with your users, provide information, and perform actions based on their configuration.

​
Creating an Agent

To create a new agent:

  1. Navigate to the Agents section from the sidebar
  2. Click the Create Agent button in the top-right corner
  3. Fill in the required information:
    • Name: A descriptive name for your agent
    • System Prompt: Define the agent’s personality, knowledge, and behavior (see Prompt Engineering below)
    • Features: Select optional features to enhance agent capabilities (see Features below)
    • Training Data: Select pre-existing Training Data to ground the agent’s knowledge
  4. Expand Advanced Settings and configure:
    • API Key: Choose the API key (e.g., OpenAI, Anthropic) that provides the model and will be used for requests
    • Model: Select the specific AI model (e.g., GPT-4o, Claude 3.5 Sonnet) to power your agent. Available models depend on the selected API Key provider
    • Temperature: Control the creativity/randomness level (0.0-1.0)
    • Max Tokens: Set the maximum length of a single response from the agent
  5. Click Create

The System Prompt is crucial for your agent’s performance. Take time to craft a prompt that clearly defines the agent’s role, tone, and constraints.

​
Agent Configuration Options

​
Prompt Engineering

The System Prompt is the foundation of your agent’s behavior. It should include:

  • Role Definition: What the agent is supposed to be (e.g., customer support specialist)
  • Personality: How the agent should communicate (e.g., professional, friendly)
  • Knowledge Scope: What information the agent should provide or avoid
  • Workflow Instructions: How the agent should handle different scenarios

Example prompt structure:

You are a customer support agent for TechGadgets, an electronics retailer.

Your personality is:
- Professional but friendly
- Helpful and patient
- Knowledgeable about tech products

Your goal is to help customers with:
- Product information and recommendations
- Order status inquiries
- Technical support for products
- Return and exchange policies

Always introduce yourself as "TechGadgets Support" at the beginning of conversations.

​
Model Selection

Chatmode supports multiple AI models with different capabilities:

ModelBest ForConsiderations
GPT-4Complex reasoning, detailed responsesHigher cost, slightly slower
ClaudeNuanced conversations, following instructionsStrong for content policies
GPT-3.5General-purpose, faster responsesCost-effective, less complex reasoning

​
API Key Selection

Ensure you have added the necessary API Keys for the AI providers (like OpenAI or Anthropic) you intend to use. The selected API key determines which models are available.

​
Temperature Setting

The temperature parameter controls response randomness:

  • 0.0-0.3: Highly deterministic, consistent responses (good for factual information)
  • 0.4-0.7: Balanced creativity and consistency (recommended for most uses)
  • 0.8-1.0: Maximum creativity and variability (good for creative content)

​
Max Tokens

This setting limits the number of tokens (roughly words or parts of words) the agent can use in a single response. Adjust this based on the desired length and complexity of answers.

​
Features

Enable specific features to extend your agent’s capabilities:

FeatureDescription
Lead CaptureAutomatically collect user information (like name, email) during the chat.
Auto AssignAutomatically assign conversations to available team members.
Auto ResolveAutomatically mark conversations as resolved after a period of inactivity.

Note: Available features may change.

​
Training Data Selection

You can link one or more completed Training Data sources to your agent. This provides the agent with specific knowledge from your documents or websites, improving its accuracy and relevance for your specific domain.

​
Managing Agents

​
Agent List View

The main Agents page displays all your agents with key information:

  • Agent name and status (active/inactive)
  • Associated environments
  • Creation date
  • Usage statistics

​
Editing Agents

To modify an existing agent:

  1. Select the agent from the list
  2. Click Edit in the agent details page
  3. Make your changes to any configuration options
  4. Save your changes

Agents currently linked to active Environments cannot be deleted directly. You must first remove the agent from all environments.

​
Agent Analytics

Monitor your agent’s performance with built-in analytics:

  • Conversation volume over time
  • Average response time
  • User feedback metrics
  • Common topics and questions

​
Best Practices

  1. Start Simple: Begin with a focused use case and expand as you learn
  2. Test Thoroughly: Try different scenarios to ensure your agent handles them correctly
  3. Iterate Based on Feedback: Regularly review conversations and refine your agent
  4. Use Training Data: Enhance your agent with domain-specific knowledge
  5. Monitor and Maintain: Regularly check agent performance and update as needed

​
Next Steps

After creating your agent, you’ll need to:

  1. Set up an environment to deploy your agent
  2. Add training data to improve agent knowledge
  3. Monitor conversations to ensure quality