Abstract

This thesis investigates the integration of generative AI in industry, specifically how small companies can integrate confidential, domain-specific data into AI systems using Retrieval Augmentation Generation (RAG). Theoretical foundations essential to understanding the RAG architecture are presented in a structured manner. Through a practical implementation, this thesis evaluates the usability of the generative AI framework LangChain for the development of a locally hosted commercial RAG system suitable for smaller companies. The findings indicate that LangChain's first stable version v0.1.0 provides the necessary tools for developing basic RAG applications. This thesis highlights the challenges of evaluating and optimising the system due to its various interdependent components and its dependence on the integrated data. Recent advances in generative AI platforms promise to assist the RAG developer in the necessary optimisation process. Overall, this research provides insight into the development of RAG applications and highlights the evolving landscape of generative AI technologies.

Results/Implementation/Project Description

Conclusion

[ PDF (optional) ] 

[ Slides Kickoff/Final (optional)]