This study addressed the lack of contextualized, efficient, and objective speaking assessments oriented toward sustainable tourism. At SMK Negeri 1 Medan, speaking activities were not sufficiently oriented toward sustainable tourism, while conventional teacher-led assessment was constrained by limited time, potential scoring subjectivity, and delayed individualized feedback. Therefore, this study aimed to develop and evaluate an AI-integrated speaking assessment using the Yoodli platform for vocational tourism students in Medan. Employing a Research and Development design adapted from the Borg and Gall model, the study involved needs analysis, product development, expert validation, and limited implementation with two English teachers and 60 Grade X–XI students. Data were collected through interviews, questionnaires, and validation instruments. Needs analysis revealed students’ difficulties in vocabulary, grammar, pronunciation, fluency, and confidence, while teachers reported time-consuming assessment and limited feedback. The developed assessment incorporated sustainable tourism tasks supported by Yoodli’s automated feedback. Expert validation yielded an overall score of 93.33% (highly valid), with content validity 95.25%, construct validity 95.00%, and practicality 90.25%. Student perceptions were highly positive (mean = 4.4/5). The findings indicate that the assessment is valid, usable, and contextually relevant for supporting speaking practice, immediate feedback, and independent learning in vocational tourism education.
Copyrights © 2026