Chatbot Rag no suporte técnico: eficiência no autoatendimento com inteligência artificial
Data
Autores
Título da Revista
ISSN da Revista
Título de Volume
Resumo
O uso de sistemas como ERPs (Enterprise Resource Planning) e ECMs (Enterprise Content Management) é essencial para a gestão de processos e a produtividade nas empresas modernas. Entretanto, a crescente complexidade desses sistemas e a dependência tecnológica, geram desafios para o suporte técnico, especialmente devido à frequência de incidentes repetitivos com soluções já documentadas, mas de difícil acesso pelos usuários. Essa situação sobrecarrega as equipes de suporte e reduz a eficiência. Este projeto explora a aplicação de chatbots integrados a Large Language Models (LLMs) combinados com a técnica de Retrieval-Augmented Generation (RAG) para reduzir a interação humana do suporte técnico em incidentes de TI com soluções já catalogadas, diminuindo assim a sobrecarga no setor. A solução buscou utilizar bases de conhecimento da empresa para proporcionar respostas precisas e contextuais a perguntas feitas ao chatbot. A pesquisa analisou a efetividade do chatbot, utilizando critérios estabelecidos, que foram avaliados pela equipe de suporte técnico da empresa, com foco em resolver os incidentes descritos. Ao combinar a flexibilidade dos LLMs com o acesso dinâmico aos dados proporcionados pelo RAG, o estudo demonstra como essa abordagem pode ser usada para reduzir a carga de trabalho do suporte humano, melhorar os tempos de resposta e diminuir custos operacionais, representando uma evolução no uso da inteligência artificial para atendimento corporativo.
The use of systems such as ERPs (Enterprise Resource Planning) and ECMs (Enterprise Content Management) is essential for process management and productivity in modern companies. However, the growing complexity of these systems and the increasing technological dependency pose challenges for technical support, especially due to the frequent recurrence of incidents with already documented solutions that are difficult for users to access. This situation overloads support teams and reduces overall efficiency. This project explores the application of chatbots integrated with Large Language Models (LLMs), combined with the Retrieval-Augmented Generation (RAG) technique, to reduce human involvement in IT support for incidents with previously cataloged solutions, thereby relieving pressure on the support sector. The solution aimed to leverage the company’s knowledge bases to provide accurate and contextual responses to questions posed to the chatbot. The research assessed the chatbot’s effectiveness using predefined criteria, which were evaluated by the company’s technical support team, focusing on resolving the reported incidents. By combining the flexibility of LLMs with the dynamic data access enabled by RAG, the study demonstrates how this approach can be used to reduce human support workload, improve response times, and lower operational costs, representing an evolution in the use of artificial intelligence for corporate support.
The use of systems such as ERPs (Enterprise Resource Planning) and ECMs (Enterprise Content Management) is essential for process management and productivity in modern companies. However, the growing complexity of these systems and the increasing technological dependency pose challenges for technical support, especially due to the frequent recurrence of incidents with already documented solutions that are difficult for users to access. This situation overloads support teams and reduces overall efficiency. This project explores the application of chatbots integrated with Large Language Models (LLMs), combined with the Retrieval-Augmented Generation (RAG) technique, to reduce human involvement in IT support for incidents with previously cataloged solutions, thereby relieving pressure on the support sector. The solution aimed to leverage the company’s knowledge bases to provide accurate and contextual responses to questions posed to the chatbot. The research assessed the chatbot’s effectiveness using predefined criteria, which were evaluated by the company’s technical support team, focusing on resolving the reported incidents. By combining the flexibility of LLMs with the dynamic data access enabled by RAG, the study demonstrates how this approach can be used to reduce human support workload, improve response times, and lower operational costs, representing an evolution in the use of artificial intelligence for corporate support.
Descrição
Citação
TRINDADE, João Vitor. Chatbot Rag no suporte técnico: eficiência no autoatendimento com inteligência artificial. 2025. Trabalho de Conclusão de Curso (Bacharelado em Sistemas de Informação) - Instituto Federal de Educação, Ciência e Tecnologia de Santa Catarina, Caçador, 2025.
