This Master’s Thesis presents the design and implementation of an intelligent automation system for telecommunication networks, based on large language models (LLMs) and specialized network analysis tools. The main objective is to reduce the operational and technical complexity associated with network management by providing an environment that enables natural language interaction with the infrastructure and automates diagnostic, analytical, and optimization tasks. The developed architecture integrates key components, including a modular backend built with FastAPI, a network analysis layer powered by Batfish, and a conversational interface driven by generative models (Google’s Gemini). These components enable the automation of complex tasks such as reachability analysis, configuration comparison between snapshots, BGP route evaluation, and the generation of technical responses in natural language—all orchestrated through a Streamlit-built graphical interface. The system is structured into three main layers: a REST API for managing and analyzing network snapshots, a logical layer with agents that generate technical responses and summaries using prompting strategies, and a visual interface that facilitates interaction through menus, graphs, and conversational assistants. Notable contributions include the seamless integration of traditional network analysis with natural language generation, automatic evaluation of generated outputs, and efficient execution of complex tasks in real time. This project demonstrates how LLMs can function as technical assistants, understanding network configurations, providing interpretable explanations, and significantly reducing operators’ cognitive load. This approach represents a step toward more autonomous, adaptable, and accessible networks and lays the groundwork for future extensions in 6G environments, security analysis, and domain-specific language model training for the telecommunications field.