How to build a Support Desk AI Assistant

Workflow Overview

This workflow powers a Support Desk AI Assistant that helps users quickly find answers to their questions. It pulls relevant content from multiple knowledge bases (Notion and StackAI Manifest) and uses an AI model to generate a concise, human-friendly response.

What This Workflow Does

• Purpose:

  Acts as a smart support desk assistant, combining multiple knowledge bases with AI to answer questions quickly and reliably.

• Ideal Use Case:

  
  

  • Customer support teams

  • Internal IT or HR help desks

  • Any company looking to centralize knowledge from different sources and provide instant answers

Check Out This Short Video Tutorial!

Step-by-Step Breakdown

1. Question (Input Node)

• Purpose: Entry point where the support representative or user types their question.

• Example input: “Main use cases of StackAI”

2. Knowledge Retrieval

The workflow looks up information in two places simultaneously:

• StackAI Manifest (Knowledge Base)

  
  

  • Searches through uploaded documentation (e.g., product docs, support material).

  • Finds the most relevant results using semantic search.

• Notion 1 (Knowledge Base)

  
  

  • Searches directly in a Notion workspace.

  • Retrieves meeting notes or documentation stored in Notion.

3. OpenAI (LLM Node)

• Role: Synthesizes the results into a clear answer.

• How it works:

  
  

  • Receives the original user question.

  • Pulls in relevant results from both StackAI Manifest and Notion.

  • Generates a brief, helpful answer.

  • If the question is unclear, it can ask for clarification or propose follow-up queries.

4. Answer (Output Node)

• Purpose: Shows the AI’s response to the user.

• Format: Text, code, or PDF download.

Summary Table