Skip to main content

​
What are AI Agents?

AI Agents are intelligent assistants that can answer customer questions using your knowledge base. They use Retrieval-Augmented Generation (RAG) to provide accurate, context-aware responses.

​
How Agents Work

  1. Customer asks a question via widget, SMS, Slack, or Zendesk
  2. Agent searches knowledge base using vector similarity search
  3. Relevant context is retrieved from your data sources
  4. AI generates response using OpenAI with the retrieved context
  5. Response is sent to the customer

​
Agent Components

​
Knowledge Base

Agents use data sources to build their knowledge:
  • Files: PDFs, Word docs, text files (processed via Unstructured.io)
  • Q&A Pairs: Question-answer pairs for common queries
  • Text Content: Custom text snippets
  • Websites: Scraped website content

​
Configuration

Each agent has configuration settings:
  • AI Mode: rag (with knowledge base) or chat (simple chat)
  • Timeout: Response timeout in seconds
  • Temperature: AI creativity level (0-1)
  • Max Tokens: Maximum response length

​
Procedures

Agents can use procedures for structured workflows:
  • Conditional logic: {{if:condition}} blocks
  • Actions: {{action:action_name}} for data fetching
  • Escalations: {{escalate:path_id}} for human handoff

​
Creating an Agent

​
Step 1: Create Agent

  1. Navigate to Agents in the main menu
  2. Click “Create Agent”
  3. Fill in the agent details:
    • Name: Customer Support Agent
    • Description: Helps customers with common questions
  4. Click “Save”

​
Step 2: Add Data Sources

  1. Navigate to your agent’s Data Sources tab
  2. Click “Add Data Source”
  3. Choose the type of data source:
    • File Upload: Upload PDFs, documents, or other files
    • Q&A Pairs: Add frequently asked questions and answers
    • Text: Add custom text content
    • Website: Import content from URLs
  4. Configure and save each data source

​
Step 3: Configure Agent Settings

  1. Navigate to your agent’s Settings tab
  2. Configure the following:
    • AI Mode: Choose RAG (Retrieval-Augmented Generation)
    • Timeout: Set response timeout (default: 30 seconds)
    • System Prompt: Customize agent behavior (optional)
  3. Click “Save Changes”
  1. Navigate to your agent’s Channels tab
  2. Create a new channel (Live Chat, SMS, or WhatsApp)
  3. Configure channel settings and link to your agent
  4. Activate the channel

​
Agent Features

Combines information from files, Q&As, texts, and websites.
Execute structured workflows with conditional logic and actions.
Seamlessly hand off to human agents when needed.
Tracks AI usage for billing purposes.

​
Next Steps

Data Sources

Learn how to build your knowledge base

Procedures

Create structured workflows

Workflows

Build automation workflows