The development of digital technology drives the need for efficient and responsive communication services that support multilingual. This study aims to develop a chatbot that facilitates communication and operational tasks for users of the DigiTeam application by integrating a multilingual dictionary into the Feed Forward Neural Network (FFNN) model and IndoBERT. The research method used is CRISP-DM, a systematic approach in data exploration, preparation, modeling, and implementation. The DigiTeam application was developed using the Agile methodology to gradually enhance the features and functionalities of the application. The dataset consists of 456 patterns and 106 tags containing common and operational work-related questions. This study utilizes a multilingual dictionary with 309 words to improve the chatbot's context understanding and response accuracy to user queries. The test results show that integrating the multilingual dictionary into the FFNN and IndoBERT models yields an accuracy of 95.45% with balanced precision and recall, demonstrating the chatbot's ability to understand and respond to multilingual queries in real-time, while also improving operational efficiency and information access in the workplace.