Building an Intelligent Multi-Agent Chatbot for SAP Troubleshooting and Training

Project Overview

The SAP LLM Chatbot was developed as an AI-driven assistant designed to streamline SAP issue resolution, provide on-demand access to training documentation, and support IT infrastructure troubleshooting. The solution leveraged multiple Large Language Models (LLMs) and was integrated with external systems to create a reliable, scalable, and user-friendly enterprise assistant.

Goals & Use Cases

1. SAP Issue Resolution

   • Enable users to insert SAP error messages directly into the chatbot.

   • The chatbot queries a structured knowledge base (Excel sheet) for relevant solutions.

   • If the issue requires escalation, the agent automatically raises a log using an API connected to the SAP Helpdesk platform.

2. Training Document Access

   • Provide instant access to training resources (e.g., “How to Create a PO”).

   • Retrieve and serve PDF documents from a centralized repository on demand.

3. IT Infrastructure Support

   • Handle general IT queries such as “My laptop is overheating”.

   • Provide step-by-step troubleshooting recommendations.

Technical Approach

• Platform Used: Voiceflow (for multi-agent orchestration and conversation design).

• Models Integrated:

  • Claude 4 – Sonnet → SAP Log Resolution Agent & IT Infrastructure Agent.

  • GPT-4o Mini → Training Document Access Agent (configured with low temperature for deterministic responses).

• Knowledge Base: Structured goggle sheet containing categorized SAP errors, solutions, and escalation paths.

• Integration: External API connection to the SAP Helpdesk for automated ticket/log creation.

Outcome & Impact

• Efficiency: Automated SAP error resolution and log-raising process, reducing manual intervention by IT consultants.

• Accessibility: Simplified access to training documentation, ensuring faster onboarding and reduced dependency on human trainers.

• Scalability: Modular multi-agent setup allows for easy extension to other enterprise use cases.

• User-Centric: Provided a single interface for SAP support, IT troubleshooting, and documentation requests.

Key Takeaway:

This project demonstrates how multi-agent LLM orchestration can be applied to enterprise systems like SAP to reduce operational bottlenecks, improve knowledge access, and optimize IT support workflows.