Abstract
<title>Abstract</title> <p>Generative artificial intelligence (GenAI) research in higher education has focused primarily on tutoring, writing, feedback, and assessment, while comparatively limited attention has been paid to advising and pathway planning as consequential sites of human–GenAI interaction. This gap is especially important in community college contexts, where learners often navigate complex program requirements, transfer uncertainty, and evolving educational and workforce goals. This article reports a design-based case study of a pathway-planning prototype and advising interaction model developed around public curricular data, requirement-group representations, and privacy-preserving architectural separation from student-level records. Drawing on public catalog documents, system design artifacts, workflow analyses, and reflective practitioner documentation, the study asks what interaction-design, governance, and modeling principles are needed for GenAI-supported advising in community college pathway decision-making. The findings are presented as eight design principles: model requirements as flexible groups rather than flat checklists; position GenAI as an explanation and option-framing layer rather than the system of record; separate public program data from protected student records; preserve human advisor review for consequential recommendations; design for learner agency and self-regulated planning; use non-deficit advising language; make uncertainty visible; and align workforce relevance without reducing advising to labor-market demand. The article extends human–GenAI interaction research into advising and student-success operations and offers a privacy-conscious design framework for institutions seeking to use GenAI in educationally consequential but high-governance settings.</p>