The Use of LLMs in the Legal Field: Optimizing Contract Management with Generative Artificial Intelligence.

In recent years, Artificial Intelligence (AI), including the emergence of ChatGPT, has attracted significant attention due to its increasing prevalence in several aspects of business processes. AI involves the development of automated systems capable of executing tasks traditionally performed by humans, with the aim of speeding up processes and reducing wasted time within organisations. This technology has also opened significant opportunities for application in the legal sector, traditionally engaged in analysing large amounts of documentation. This Master's thesis explores the use of Large Language Models (LLM) to support legal staff and reduce document management time. The aim of this research is to study, design, develop a POC (proof of concept) to address these challenges by implementing a web application where lawyers can analyse contracts e generate contract. The application is based on Retrieval-Augmented Generation (RAG) capable of providing fast, effective and high-quality responses. To achieve this goal, an in-depth analysis was conducted on large language models and the prompts used to guide them. To achieve this, the analyses focused on the effectiveness of LLMs in interpreting legal language and their ability to integrate information to produce relevant and coherent output. Particular attention was paid to the configuration of prompts and their optimisation to improve the accuracy of responses. In conclusion, this thesis highlights the considerable potential of generative AI in the legal field. By integrating the advantages of semantic embeddings for information retrieval with those of generative AI for producing answers, lawyers can significantly reduce the time spent in drafting new contracts, taking into account previous clauses, and analysing new contracts. This approach enables effective optimisation of legal processes, making contract management more efficient and accurate.