Creating Agents

Learn how to create and configure AI agents in ArcanFlows step by step.

Creating a New Agent

Step 1: Navigate to Agents

1. Log in to your ArcanFlows dashboard
2. Click Agents in the left sidebar
3. Click the Create Agent button in the top right

Step 2: Basic Information

Fill in the basic details:

Step 3: Choose a Template (Optional)

You can start from:

• Blank: Start from scratch
• Customer Support: Pre-configured for support scenarios
• Sales Assistant: Optimized for sales conversations
• Knowledge Base Q&A: Answers questions from documents

Configuring Your Agent

Model Settings

Choose your AI model and provider:

┌─────────────────────────────────────────────┐
│  Model Configuration                         │
├─────────────────────────────────────────────┤
│  Provider:    [OpenAI          ▼]           │
│  Model:       [GPT-4 Turbo     ▼]           │
│  Temperature: [0.7        ] ────○────        │
│  Max Tokens:  [2048       ]                  │
└─────────────────────────────────────────────┘

Provider Options

Parameter Guide

System Prompt

The system prompt defines your agent's personality, knowledge, and behavior:

markdownCopy
You are a helpful customer support assistant for TechCorp.

Your responsibilities:
- Answer questions about our products and services
- Help troubleshoot common issues
- Guide users through our documentation
- Escalate complex issues to human support

Guidelines:
- Be friendly, professional, and concise
- Always verify user identity before sharing account details
- If unsure, say so and offer to connect with a human agent
- Never share internal processes or confidential information

Available actions:
- Look up order status
- Check account information
- Create support tickets
- Schedule callbacks

Welcome Message

Set the first message users see:

Hi! I'm the TechCorp support assistant. How can I help you today?

I can help you with:
• Product questions
• Order status
• Technical support
• Account issues

What would you like to know?

Adding Knowledge

Upload Documents

Give your agent access to information:

1. Go to the Knowledge Base tab
2. Click Upload Documents
3. Select files (PDF, TXT, MD, DOCX)
4. Wait for processing to complete

Configure Retrieval

Optimize how your agent finds information:

Adding Tools

Enable your agent to take actions:

Built-in Tools

Custom Tools

Define custom functions:

jsonCopy
{
  "name": "get_order_status",
  "description": "Look up the status of a customer order",
  "parameters": {
    "type": "object",
    "properties": {
      "order_id": {
        "type": "string",
        "description": "The order ID to look up"
      }
    },
    "required": ["order_id"]
  }
}

Testing Your Agent

Test Chat

Use the built-in test interface:

1. Click the Test button
2. Send messages to your agent
3. Review responses
4. Iterate on your configuration

Test Scenarios

Try these scenarios:

✓ Basic greeting
✓ Product question
✓ Edge case handling
✓ Unknown question response
✓ Tool usage
✓ Knowledge base retrieval

Publishing

When you're satisfied with testing:

1. Click Publish in the top right
2. Confirm the publication
3. Your agent is now live!

Publication Checklist

Before publishing, verify:

• System prompt is complete
• Welcome message is set
• Knowledge base is uploaded
• Tools are configured
• Testing is complete
• Error handling is in place

Agent Settings

Access Control

Configure who can use your agent:

Rate Limiting

Protect your agent from abuse:

Messages per minute: 10
Messages per hour: 100
Messages per day: 1000

Conversation Settings

Next Steps

• System Prompts - Write effective prompts
• Model Selection - Choose the right model
• Knowledge Base - Add documents
• Tools - Enable agent actions