The integration of artificial intelligence (AI) into language education presents new opportunities for enhancing vocabulary learning through interactive, student-centered approaches. This study investigates the effectiveness and challenges of using LexiBot, a Telegram-based AI chatbot, as a vocabulary learning tool in English language instruction. Addressing the limitations of rote memorization, LexiBot promotes contextualized, feedback-driven learning experiences. Using a mixed-method design, survey data were collected from 110 university students and focus group interviews were conducted with 15 participants. Findings revealed that students perceived LexiBot as helpful (M = 4.2, SD = 0.8), engaging (M = 4.0, SD = 0.9), and confidence-boosting, with a statistically significant correlation between engagement and perceived helpfulness (r = 0.55, p < 0.05). Students appreciated LexiBot’s interactive nature, accessibility, and instant feedback, though they also reported difficulties in understanding contrast clues, a lack of interactive exercises, and limited personalization. Notably, the mean score for willingness to recommend LexiBot was high (M = 4.3, SD = 0.7), reflecting an overall positive reception. Qualitative findings suggested the need for enhanced contextual explanations, gamified learning activities, and adaptive features. This study highlights the potential of AI-powered chatbots in fostering self-directed vocabulary learning and underscores the importance of learner-centered design in educational technology implementation.