Skip to main content
Captain is Chatwoot’s AI agent system that enables intelligent, context-aware customer support automation. With Captain, you can create AI agents that understand your product, answer customer questions, and seamlessly hand off to human agents when needed.

​
Overview

Captain AI agents are powered by advanced language models and can be customized to match your brand voice and support needs. They work alongside your team to provide 24/7 support, reduce response times, and improve customer satisfaction.

AI Assistants

Create custom AI agents for your inboxes

Configuration

Configure models and features

Features

Editor, copilot, transcription, and more

Knowledge Base

Connect documentation and FAQs

​
Captain Features

Captain provides multiple AI-powered features to enhance your support workflow:

​
Available Features

AI-powered message editor that helps agents improve, expand, or simplify their responses.Default Model: GPT-4.1 MiniAvailable Models:
  • GPT-4.1 Mini, GPT-4.1 Nano
  • GPT-5 Mini, GPT-4.1, GPT-5.1, GPT-5.2
  • Claude Haiku 4.5
  • Gemini 3 Flash, Gemini 3 Pro
Full AI assistant that can engage in customer conversations, search documentation, and answer questions.Default Model: GPT-5.1Available Models:
  • GPT-5 Mini, GPT-4.1, GPT-5.1, GPT-5.2
  • Claude Haiku 4.5, Claude Sonnet 4.5
  • Gemini 3 Flash, Gemini 3 Pro
AI copilot that assists agents during conversations with suggestions and context.Default Model: GPT-5.1Available Models:
  • GPT-5 Mini, GPT-4.1, GPT-5.1, GPT-5.2
  • Claude Haiku 4.5, Claude Sonnet 4.5
  • Gemini 3 Flash, Gemini 3 Pro
Automatically suggests conversation labels based on content.Default Model: GPT-4.1 NanoAvailable Models:
  • GPT-4.1 Nano, GPT-4.1 Mini, GPT-5 Mini
  • Gemini 3 Flash, Claude Haiku 4.5
Transcribes voice messages and audio attachments.Model: Whisper-1 (OpenAI)

​
Creating AI Assistants

AI Assistants are Captain agents that can be assigned to inboxes to handle customer conversations.

​
Assistant Configuration

When creating an assistant, you can configure:
​
name
string
required
Display name for the assistant
​
description
string
required
Description of the assistant’s purpose and capabilities
​
product_name
string
The product or service name the assistant supports
​
temperature
number
Controls response creativity (0.0 to 1.0). Lower values are more focused and deterministic.
​
response_guidelines
array
Guidelines for how the assistant should respond to customers
​
guardrails
array
Rules to prevent certain behaviors or topics

​
Assistant Features

  • FAQ Integration: Connect to your knowledge base for accurate answers
  • Memory: Remember context from previous messages in the conversation
  • Document Search: Search through uploaded documentation
  • Handoff: Seamlessly transfer to human agents when needed
  • Custom Tools: Extend functionality with custom integrations

​
Configuration

​
API Configuration

Captain requires OpenAI API configuration. Set these in your installation config: 
CAPTAIN_OPEN_AI_API_KEY=your_api_key_here
CAPTAIN_OPEN_AI_ENDPOINT=https://api.openai.com/v1  # Optional, for custom endpoints

​
Account Preferences

Configure Captain features and models per account: 
curl -X GET \
  https://app.chatwoot.com/api/v1/accounts/{account_id}/captain/preferences \
  -H 'api_access_token: YOUR_ACCESS_TOKEN'

​
Enabling Features

Each Captain feature can be enabled or disabled at the account level: 
# Check if a feature is enabled
account.captain_editor_enabled?
account.captain_assistant_enabled?
account.captain_copilot_enabled?

# Get the selected model for a feature
account.captain_editor_model        # => "gpt-4.1-mini"
account.captain_assistant_model     # => "gpt-5.1"

​
Knowledge Integration

Captain Assistants can access your knowledge base to provide accurate answers:

​
Document Management

  • Upload documents (PDF, text files)
  • Connect to help center articles
  • Enable FAQ generation from conversations
  • Semantic search with embeddings

​
FAQ Features

Enable FAQ functionality in your assistant config: 
{
  "feature_faq": true,
  "feature_memory": true
}

​
AI Providers

Captain supports multiple AI providers:

OpenAI

GPT-4.1, GPT-5, Whisper, Embeddings

Anthropic

Claude Haiku, Claude Sonnet (coming soon)

Google

Gemini 3 Flash, Gemini 3 Pro (coming soon)

​
Model Selection

Choose models based on your needs:
  • Nano/Mini: Fast, cost-effective for simple tasks
  • Standard: Balanced performance for most use cases
  • Advanced: Premium models for complex reasoning
Model availability and credit multipliers vary. Premium models consume more credits per request.

​
Assigning to Inboxes

Connect Captain Assistants to your inboxes:
  1. Navigate to Settings > Inboxes
  2. Select an inbox
  3. Go to the Collaborators tab
  4. Choose a Captain Assistant from the dropdown
  5. Save changes
The assistant will now handle incoming conversations in that inbox.

​
Best Practices

  • Be specific about tone and style
  • Define boundaries clearly
  • Include brand voice guidelines
  • Specify when to hand off to humans
  • Set topic boundaries
  • Define prohibited actions
  • Establish escalation criteria
  • Protect sensitive information
  • Keep documentation up-to-date
  • Organize content logically
  • Use clear, concise language
  • Test with common questions
  • Use appropriate models for tasks
  • Balance cost vs. quality
  • Monitor performance metrics
  • Adjust temperature for creativity
Captain Assistants should complement, not replace, your human support team. Always provide clear handoff paths to human agents.

​
Monitoring and Analytics

Track Captain performance:
  • Response accuracy
  • Handoff rates
  • Customer satisfaction
  • Credit usage by feature and model

Agent Bots

Webhook-based automation

Canned Responses

Pre-written response templates

Build docs developers (and LLMs) love

Get started for freeTalk to us