• Generative AI
• Feature Extraction
• OpenAI
• Advanced Analytics
• Microsoft Azure
• GPT
• Large Language Models

The present work provides an overview of the use of Generative Large Language Models as tools for extracting features from customer service chats in a corporate setting, integrating models from the GPT family into an advanced analytics architecture. This study gains significance through the development of an Italian-language prompt for feature extraction, utilizing state-of-the-art prompt engineering techniques. The proposed experiments not only showcase the achieved advancements but also shed light on the primary challenges tied to integrating a model of this nature into an analytics system. Particularly intriguing in the context of this research is the incorporation of Language Models into a business environment, probing the delicate balance between expenses, efficiency, and consistency. The proposed experiments underscore the significant leap forward in modernizing and streamlining analytical procedures that come with integrating generative models into a business process, returning suitable outcomes even without the need for model specialization or the reliance on contextually labeled examples. Even though satisfactory results have already been achieved, the ongoing experimental configuration of the proposed system allows for further improvement in the future by exploiting cutting-edge methodologies. It is evident from the experiments that harnessing the great power of these tools must be coupled with thoughtful assessments from a human perspective. Validation of outputs generated by a model of this nature by experts becomes imperative, highlighting the essential need to blend the capabilities of these advanced tools with the nuanced insights provided by human expertise.