AI-Powered SAP Error Resolution Assistant — Summary
This project delivers an intelligent assistant designed to resolve SAP system errors using advanced AI techniques, with a focus on Type E (hard stop) errors. The solution integrates both traditional AI libraries and modern Large Language Models (LLMs) to provide real-time, context-aware resolutions for SAP users and consultants.
Key Features
Technology Used
Python in Google Colab
core development environment
FAISS + Sentence Transformers
semantic search for matching error messages
Python in Google Colab
LLM backbone (used Mistral-7B-Instruct)
Pandas / JSON
for handling and persisting the error library
LLM Integration
- Replaced basic keyword search with a semantic assistant powered by Mistral-7B-Instruct, running on GPU-enabled Colab.
- Enables the assistant to generate human-like, step-by-step SAP error resolutions even if
the error isn't present in the training data.
Real-Time Learning Workflow
- User submits an SAP error (e.g., “F5808”).
- Assistant checks local library (via FAISS).
- If not found, Mistral LLM generates a detailed resolution.
- The solution is appended to the library for future use.
- Current library is updated and displayed for verification.
