This study aims to explore how artificial intelligence (AI) technology impacts the improvement of students' oral communication competence in English. The ELSA (English Language Speech Assistant) application, powered by AI, provided feedback to users to enhance their English speaking skills. An Action Research approach was employed, integrating both quantitative and qualitative methods to gain a holistic understanding of AI's effectiveness in English language learning. The research was conducted in three phases: pre-observation, Cycle 1, and Cycle 2. Each cycle consisted of several stages: planning, implementation, action, observation, and reflection. To assess student achievement, quantitative analysis was used to evaluate test results. Additionally, a survey was conducted to gather students’ perceptions and experiences regarding the use of AI in learning. The results showed that after Cycle 1, 11 students (44%) achieved the minimum required score of 75, while 14 students (56%) scored below the success criteria (below 72). Based on the success criteria of 75% of students meeting the minimum score, the study proceeded to Cycle 2. In Cycle 2, 76% of students met the success criteria, showing significant improvement over Cycle 1. The survey also indicated positive perceptions from students regarding the ELSA app’s design, interface, variety of topics, and feedback feature. Students expressed satisfaction with the app, as it allowed them to practice English conversation anytime and anywhere.