RAG Chatbot / Knowledge Management
IntranetGPT – RAG Chatbot for SharePoint Knowledge
RAG chatbot for SharePoint knowledge: IntranetGPT answers employee questions in natural language. Internally reported: up to 70% shorter search times. No changes to existing infrastructure.
Starting point
At a mid-sized IT services company, company knowledge was scattered across SharePoint, HR documents and internal policies. In theory everything was there; in practice hardly anyone found anything without asking colleagues. The goal was a scalable, privacy-compliant solution that works without rebuilding the existing infrastructure.
Solution
An AI chatbot based on RAG (retrieval-augmented generation) that makes internal company knowledge from SharePoint and the intranet accessible in natural language. It answers questions about processes, documents, contacts and company policies with the right context.
The principle: the language model is not trained on company data. For each question, the system retrieves the relevant document passages, and the model answers based on those alone.
- Connection to SharePoint via the Microsoft Graph API, with no changes to the existing setup
- Python pipeline for extracting, converting and cleaning the documents
- Chat interface in React, hosted on Azure
- Built solo as lead developer in a client project at generic.de
Whether a RAG system answers correctly can only be checked with good test data. Our approach is described in Ground truth data in RAG systems.
Technology stack
- Azure AI Foundry (LLM orchestration)
- Azure App Service (hosting)
- MS Graph API (SharePoint integration / data retrieval)
- Python with OpenAI SDK (data extraction, conversion, cleaning)
- ReactJS (frontend / chat interface)
Result & Impact
- Up to 70% shorter search times for employees (internally reported)
- Ready to use, with no changes to the existing SharePoint setup
- Scalable to further knowledge areas such as HR, IT and compliance