AI & Datasources

Training your AI Chatbot

AI Integration & Datasources

LooperChat's AI Assistant uses RAG (Retrieval-Augmented Generation) to answer visitor queries based on your specific data. To make the AI effective, you need to configure its "Knowledge Base" via Datasources.

1. URLs (Web Crawling)

You can train the AI on your existing website content.

  1. Navigate to Projects > Select Project > Data Source > URL.
  2. Click New URL.
  3. URL: Enter the web page address you want to index.
  4. Enable: Toggle to active status.
  5. The system will crawl the content and make it available to the AI.

Note: You can define Auto-Crawling rules in Project Settings to automatically include or exclude URL patterns.

2. FAQs

Add structured Question & Answer pairs for precise responses.

  1. Navigate to Projects > Select Project > Data Source > FAQ.
  2. Click Add FAQ.
  3. Question: Enter the common query (e.g., "What is your refund policy?").
  4. Answer: Enter the definitive answer.
  5. Enable: Toggle on.

3. System Prompts

Customize the behavior and personality of your AI Assistant.

  1. Navigate to Projects > Select Project > Data Source > System Prompts.

  2. Prompt: Enter instructions for the AI.

    Example: "You are a helpful support agent for Acme Corp. Be polite and concise. Only answer questions related to our products."

  3. Use System Prompt: You can also select from pre-configured system templates if available.

OpenAI Configuration

To power the AI, you must connect an AI provider (usually OpenAI) at the organization level.

  1. Go to Connect > OpenAI Key.
  2. Click Connect with your AI Keys.
  3. Project: Select which project this key applies to.
  4. Type: Select openai (or gemini, grok).
  5. Model: Choose the model (e.g., gpt-4o, gpt-4o-mini).
  6. API Key: Paste your secret key from the provider.

With datasources and AI configuration in place, LooperChat's AI Assistant is ready to respond to visitors using your own data in a reliable and controlled way.