Once you've created an agent, you can customize it to perfectly match your needs. This guide covers all the configuration options available for your agents, from basic settings like prompts and models to advanced features like tools, knowledge bases, and memory.

Accessing Agent Configuration

To configure an agent:

Navigate to the Agents page from the main navigation
Click on the agent you want to configure
Click the Settings or Configure button (usually found in the top-right corner of the agent page) to open the configuration panel

The configuration panel is organized into several sections, each controlling different aspects of your agent's behavior. You can modify these settings at any time, and changes will apply to new conversations.

1. Customizing the System Prompt

The system prompt is the foundation of your agent's personality and behavior. It tells the agent who it is, what it should do, and how it should respond.

Editing the Prompt

In the agent configuration, find the System Prompt section
Click to edit the prompt text
Be specific about:
- The agent's role and expertise
- The tone and style of communication
- Any specific guidelines or constraints
- Expected output formats

Using the Enhance Feature

If you're not sure how to write an effective prompt, use the Enhance button. This feature uses AI to automatically refine your description into a more effective system prompt that will help your agent perform better.

[!TIP] Best Practices for Prompts

Be specific about the agent's role and domain expertise

Include examples of desired behavior

Set clear boundaries and constraints

Mention the expected output format (e.g., "Always provide step-by-step instructions")

2. Switching the AI Model

Different AI models have different strengths. You can switch models to find the best fit for your agent's tasks.

Available Models

GPT-4/5: Excellent for complex reasoning and following detailed instructions
Claude Opus/Sonnet: Great for nuanced understanding and creative tasks
Gemini: Strong performance for general tasks and multimodal capabilities

Changing the Model

In the configuration panel, find the Model Selection section
Choose from the dropdown of available models
The change takes effect immediately for new conversations

[!NOTE] Model Capabilities Different models may have varying capabilities with tools, memory, and knowledge bases. Some advanced features may work better with certain models.

3. Adding and Configuring Tools from Integrations

Tools allow your agent to interact with external services and perform actions beyond just conversation.

Enabling Integrations

Before an agent can use tools, you must first connect integrations at the platform level (see Adding Integrations).

Selecting Tools for Your Agent

In the Add Tools section, you'll see all available integrations that you've connected at the platform level
Expand an integration to see its available tools
Select the specific tools you want this agent to have access to
The agent's system prompt will automatically update to include instructions on when and how to use these tools
You can select multiple tools from multiple integrations for a single agent

Tool Configuration Options

Auto-approval: Tools execute automatically when the agent decides to use them
Manual approval: You'll be prompted to approve each tool call before execution
Tool-specific settings: Some tools may have additional configuration options

[!IMPORTANT] Security Consideration Only give agents access to tools they actually need. This follows the principle of least privilege and helps prevent unintended actions.

4. Adding Knowledge to Your Agent

Knowledge bases give your agent access to specific information, making it an expert on your content.

Document Upload

You can upload documents that your agent can reference:

In the Knowledge Base section, click Add Documents
Upload files in supported formats:
- PDF: Reports, manuals, documentation
- DOCX: Word documents, formatted text
- TXT: Plain text files
- CSV/XLSX: Spreadsheet data
Documents are automatically processed and indexed for semantic search

Indexed Websites

In addition to documents, you can add websites to your agent's knowledge base:

Click Add Website in the Knowledge Base section
Enter the website URL you want to index
The system will crawl and index the website content automatically
Once indexed, your agent can reference and cite information from these websites in its responses
You can add multiple websites to build a comprehensive knowledge base

How Knowledge is Used

When knowledge is enabled:

The agent automatically searches your knowledge base when relevant
Relevant documents or website content are included in the agent's context
The agent can cite sources from your knowledge base
This enables RAG (Retrieval Augmented Generation) capabilities

For more details, see Configuring Knowledge & Memory.

5. Configuring Memory

Memory allows your agent to remember information from previous conversations, creating a more personalized and context-aware experience.

Enabling Memory

In the Memory section, toggle memory on or off
When enabled, the agent will:
- Extract key information from conversations
- Store important facts and preferences
- Recall past interactions in future conversations

Memory Behavior

Automatic extraction: The agent automatically identifies and stores important information
Persistent storage: Memories persist across conversations
Contextual recall: Relevant memories are automatically included when needed

[!TIP] Memory Best Practices Memory works best when you have ongoing conversations with the same agent. It's particularly useful for agents that need to remember user preferences, project details, or ongoing tasks.

6. Using Agents in Automated Workflows

Agents can be used as building blocks in automated workflows, allowing you to create complex automation chains.

Sequential Workflows

In sequential workflows, agents execute one after another:

Navigate to Workflows and create a new workflow
Add your agent as a Task node
Configure how context is passed between agents:
- Share context: Agents see previous agent outputs
- Isolated: Each agent starts fresh

Orchestrator Workflows

In orchestrator workflows, a manager agent delegates tasks to worker agents:

Create an Orchestrator workflow
Configure the manager agent with delegation logic
Add your agent as a worker agent
The manager will intelligently decide when to use your agent

Workflow Configuration

When adding an agent to a workflow, you can configure:

Context management: How the agent receives information
Tool approval: Whether tool calls need approval
Output format: How the agent's response is structured

7. Chatting One-on-One with Your Agent

The most direct way to interact with your agent is through the chat interface.

Starting a Conversation

Navigate to your agent's page
The chat interface opens automatically
Start typing your message
The agent responds using all its configured capabilities:
- System prompt for personality and behavior
- Tools for external actions
- Knowledge base for information retrieval
- Memory for context from past conversations

Chat Features

Streaming responses: See responses as they're generated in real-time
Tool execution: Watch tools execute in real-time with visual indicators
Memory indicators: See when the agent is using stored memories from past conversations
Knowledge citations: View sources from your knowledge base when the agent references them
Attachment support: Upload files and images for the agent to analyze
Multimodal capabilities: Agents can process text, images, and documents in conversations

Conversation Management

New conversation: Start fresh without previous context
Thread history: Access past conversations
Export: Save conversations for reference

8. Using the Playground for Testing

The Playground is a powerful testing environment where you can experiment with agents before using them in production workflows.

Agent Testing Mode

Test individual agents or switch between multiple agents:

Navigate to the Playground
Create a new session in Agent Testing Mode
Select an agent to test
Configure tools, knowledge, and other settings
Start chatting to see how the agent behaves

Key Playground Features

Switch agents mid-conversation: Test different agents with the same input to compare responses
Hand off context: See how agents pass information to each other, simulating workflow behavior
Tool testing: Test tool integrations without affecting production agents or workflows
Configuration experimentation: Try different settings, models, and prompts without saving changes
Session management: Save and load playground sessions to continue testing later

Orchestrator Mode

Test how agents work together in orchestrator patterns:

Create a session in Orchestrator Mode
Configure a manager agent with delegation logic
Add multiple worker agents to the team
Test the full orchestration:
- Run Mode: Execute the entire flow and see the final result
- Step-by-Step Mode: Watch each agent execute individually

Exporting from Playground

Once you've tested and refined your agent configuration in the Playground, you can:

Export to Workflow: Convert your playground session into a production workflow
Save as Template: Create a reusable template from your configuration
Save as Draft: Keep your work for later refinement

[!TIP] Playground Best Practices Use the Playground to test new agent configurations, tool combinations, and workflow patterns before deploying them. It's a safe space to experiment without affecting your production agents.

Configuration Best Practices

Start Simple

Begin with basic configuration (prompt and model) and gradually add complexity (tools, knowledge, memory) as needed.

Test Thoroughly

Use the Playground to test your agent configuration before deploying it in workflows or using it for important tasks.

Iterate and Refine

Agent configuration is an iterative process. Start with a basic setup, test it, gather feedback, and refine based on results.

Document Your Choices

Keep notes on why you made specific configuration choices. This helps when optimizing or troubleshooting later.

Next Steps

Now that you understand agent configuration, you might want to:

Learn more about Configuring Knowledge & Memory
Explore Adding Integrations
Discover how to Create Workflows

Your agent is now fully configured and ready to help you automate tasks and solve problems!

Configuring your Agent