This explanatory sequential mixed-methods study investigated the impact of voice-based AI chatbots on the communicative competence of Iraqi EFL learners as transnational learners in Iran. At the quantitative phase, 60 intermediate-level Iraqi EFL learners were randomly assigned to an experimental group, which engaged in conversational practices with a voice-based AI chatbot (Replika), and a control group, which received the traditional instruction. Communicative competence was measured using the Test of Spoken English (TSE) in pre-test and post-test assessments. At the qualitative phase, semi-structured interviews were conducted with a sample from the experimental group. The quantitative data was analyzed using ANCOVA, while the qualitative data was analyzed using inductive thematic analysis. The quantitative results demonstrated a statistically significant improvement in the TSE scores of the experimental group compared to the control group. The analysis of the qualitative data led to the extraction of six themes: increased fluency, the enhancement of confidence in speaking, strategic communication adjustment, increased cultural and sociolinguistic awareness, perceived linguistic limitations of the chatbot, and an objective assessment of the chatbot’s value in pedagogical matters as an easily-accessible and low-anxiety partner for speaking practice. Voice-based chatbots were regarded as spaces for non-judgmental, extensive speaking practice, which is often limited in traditional classroom contexts. They were considered as new possibilities for enhancing EFL learners’ strategic and sociolinguistic competence by creating a supportive environment in which learners develop their communicative strategies.