DEVELOPMENT OF A CHATBOT WITH API FOR CUSTOMER SUPPORT OPTIMIZATION - NEXUSSUPPORT AI

Abstract

The growing demand for efficiency in customer support is driving the adoption of solutions based on Artificial Intelligence (AI). Given this scenario, this work presents the conception and implementation of NexusSupport IA, an automated support chatbot that utilizes the Google Gemini Large Language Model (LLM) API as its inference engine. The central thesis is that integrating a cutting-edge LLM can reduce the First Contact Resolution Time (FCR) and optimize ticket management, thus elevating the quality of the user experience. The system is structured to interact with users, offering detailed solutions, maintaining the historical context of the conversation (via session management), and, critically, performing Intelligent Handoff to specialized technicians when the problem's complexity exceeds the model's scope. The research encompasses Prompt Engineering, the development of its authentication interface (HTML/CSS) and the chat screen (frontend with React/JavaScript), communicating directly with the backend service that orchestrates calls to the Google API. The project is structured with modern web development tools like Vite for the build process and NPM for dependency management.