Srovnání platforem pro vývoj vlastních AI chatbotů

Abstract

This thesis presents a comparison of platforms for developing custom AI chatbots based on large language models (LLMs). The theoretical part introduces the concepts of generative artificial intelligence, LLMs, and the Retrieval-Augmented Generation (RAG) principle, which enables chatbots to work with custom data sources. The analytical part evaluates selected platforms with a focus on their functionality and possibilities for integrating proprietary data. Based on this analysis, an experimental scenario is designed, targeting the creation and verification of industrial documentation. The experiment is subsequently carried out on a selected platform using the RAG principle in combination with large language models. The outputs are then evaluated in terms of quality, accuracy, and response speed. The thesis concludes with an evaluation of the achieved results and recommendations for further development.