Overview
The project aims to design and implement a proof-of-concept (PoC) for an AI-powered enterprise search engine that enables the use of heterogeneous and predominantly unstructured corporate data. The starting point is the fact that the company’s existing data is distributed across multiple operational systems, including file servers, MailStore, and ERP systems, and has so far only been searchable and analyzable to a limited extent. At the same time, there is a strategic interest in making greater use of artificial intelligence in the future for knowledge management, decision support, and process automation.
The planned PoC therefore has two main objectives: First, it aims to establish a technical foundation for structured data organization by developing an automated ingestion and indexing pipeline that virtually integrates existing data sources without migrating the original data. Second, it aims to implement an AI-powered search function that enables employees to find information through natural language queries. The search interface will combine hybrid search methods—that is, both traditional keyword-based and semantic search techniques—and provide answers with traceable source references. User permissions are particularly central to security in this context.
The project comprises several work packages. First, a structured preliminary analysis of suitable standard software solutions for enterprise search will be conducted, taking into account criteria such as security mechanisms, integration capabilities, API openness, and AI and RAG capabilities. Building on this, a data pipeline will be implemented to access selected data sources, extract metadata, and prepare content for semantic search processes. Subsequently, a prototype chatbot interface will be developed that is based on Retrieval-Augmented Generation (RAG) and generates responses using local language models.
