Speaking competence remains a critical challenge for EFL learners, particularly in achieving accurate intonation and fluent speech due to limited speaking exposure, insufficient feedback, and conventional classroom practices that prioritize form over communication. Although recent studies have emphasized the potential of artificial intelligence (AI) in language learning, limited research has specifically examined the role of AI-powered speech recognition tools in improving suprasegmental features such as intonation within reflective classroom contexts. Therefore, this study aimed to identify to what extent the integration of AI-powered speech recognition tools optimize EFL learners’ intonation accuracy, and to describe the integration of the tools influence learners to speak fluency. This study employed a Classroom Action Research (CAR) design using mix method data approach involving 20 undergraduate learners which selected by purposive sampling technique at a private university in Pekanbaru. The research was conducted in one cycle consisting of planning, action, observation, and reflection stages. The data were collected through intonation test and speaking fluency performance, observation checklists, reflective journals, and AI-generated performance reports, and analysed by using three data analysis theory. The findings revealed a noticable improvement in learners’ speaking performance, with mean scores increasing from 60.25 in the pre-test to 80.75 in the post-test. In conclusion, the integration of AI-powered tools such as Stimuler can serve as effective complementary resources in speaking instruction, and it is recommended that educators integrate such technologies alongside communicative approaches while future research should explore long-term impacts and broader aspects of communicative competence.