Skip to main content
This guide empowers you to leverage the power of OpenAI Assistant within Chatbot Builder AI. It allows your chatbot to personalize responses based on your business data (PDFs, docs), creating a seamless communication experience.

Getting Started

1

Create the Assistant

  1. Navigate to Settings > Integrations > OpenAI.
  2. Click the Manage button.
  3. Click Add.
  4. Name: Choose a name that reflects the purpose (e.g., “Product Information Assistant”).
2

Define Capabilities

Configure the following settings to customize your assistant:
  • Instructions: Provide clear instructions guiding the Assistant on how to interpret and respond to user queries. Be as specific as possible.
  • Model: * Default: GPT-3.5 Turbo 16k (included with standard accounts).
    • GPT-4: Required if you want to combine Functions and Files (requires your own API key).
  • Functions: Define pre-programmed actions for specific tasks (e.g., book_appointments, connect_user_to_human).
  • Files: Upload data files (PDFs, Word Docs, Text). Max 25MB each. The Assistant will intelligently draw information from these files.
Important: Combining Functions and Files is an advanced feature that requires selecting a GPT-4 model.

Integration Methods

There are two ways to integrate your Assistant: as the central response engine (Main System Prompt) or within specific conversation paths (Flows).
Use this method to make the Assistant the primary brain for your chatbot.
  1. Navigate to Integrations > OpenAI > Edit.
  2. Switch to the Assistant tab.
  3. Select your newly created Assistant from the dropdown menu.
  4. Click Continue.
Crucial Settings: * Ensure Default Agent (formerly Auto-GPT) is Enabled.
  • Go to Settings > General and set Default Flow (formerly Default Reply) to “None”.
This activates the Assistant for all user queries.

Additional Tips

  • Experiment: Test different Assistant configurations (instructions, file types) to find the optimal setup.
  • Organize Data: Ensure your uploaded files are organized for clear and efficient retrieval.