Morgan funds student-athlete support chatbot; benefits remain prospective

Ask athletics academic-support leaders, compliance staff, IT and student representatives where students struggle to find timely help. Which questions recur outside office hours, which require a specialist, and how long does a successful referral currently take? A bounded engagement could define the approved information set and test one referral workflow with volunteer users. The value hypothesis is easier access to the correct person or resource. Funding is evidence of an initiative, not a purchasing opportunity or available services budget. Do not promise higher retention, improved eligibility, savings or fewer staff. Establish the existing support baseline and include review effort before discussing expansion.

A read-only information assistant is a reasonable initial fit; autonomous record updates need separate justification. Prerequisites include authoritative content owners, document permissions, identity decisions and a working human handoff. Prototype source-linked responses in a sandbox and test outdated policies, conflicting dates, unsupported eligibility questions and wellness escalation. Use synthetic records to test cross-user disclosure and prompt injection. Compare approved cloud, local and hybrid designs on privacy, latency, maintainability and cost; the announcement selects none. Proposed proof of value should score correct referrals, severe answer errors, response delay and total staff handling time against the existing service. Developer copilots may help implementation but are not evaluated here.

Assign academic support as service owner, IT as operator and compliance as reviewer of consequential content. Inventory and version the approved knowledge base, prepare accessible orientation and rehearse handoffs before a limited pilot. Dependencies include counselor availability, content maintenance, privacy review and student feedback.

Proposed acceptance criteria

every critical eligibility test reaches an authorized reviewer; all sampled permission tests pass; users can complete key tasks with assistive technology; unresolved severe errors block release. These are proposed gates, not reported outcomes. Track referral completion, corrections and staff effort alongside adoption. Preserve a non-chat route and fund maintenance beyond the grant period.