This paper reports the second cycle of the LLM Council with Double Delphi methodology applied to faculty workload policy reform at an adult-serving Open University (OU) within a State University System (SUS). Cycle 1 (Chukhlomin, 2025a, 2025b) established the methodology, conducted the first AI round with five frontier large language models, and produced a working paper proposing a 704 Student Credit Hour (SCH) baseline and a two-track workload structure. That paper was then submitted to human domain experts for verification—the human round specified but not executed in Cycle 1. This paper opens with what happened when human experts engaged with the Cycle 1 output: their feedback reshaped the research agenda, challenged framing assumptions, demanded methodological transparency, and redirected attention to advisor caseload benchmarking as the empirical key to the mentoring conversion factor. The revised research questions were then returned to an expanded six-model LLM Council for Cycle 2, producing a new evidence base that includes tiered advisor-to-student ratios, annotated institutional sources, and job-description analysis for advisors performing individualized degree planning. The principal empirical finding is that OU’s own professional advisor caseload data (120 students in individualized programs, 240–300 in registered programs) provides a direct anchor for the previously unvalidated mentoring credit conversion factor, converging independently with the Western Governors University comparator surfaced by the AI layer in Cycle 1. The paper adopts 800 SCH as the teaching-focused ceiling, derives 260 mentoring-equivalent credits as the available space within the teaching-and-mentoring track, and presents an advisor-anchored sensitivity analysis yielding fair caseload estimates of 37–43 mentees at blended rates of 6–7 credits—substantially below the current 51-mentee assignment.
Valeri Chukhlomin (Sat,) studied this question.