Campus AI vs general assistants: a decision framework

General assistants are extraordinary at general knowledge. They are structurally incapable of knowing your institution, because your institution is not in their training data and your documents are not in their retrieval path.

Ask a general assistant when the internal assessment retest window closes, which elective sequence a third-year student should choose under your regulation, or what your NAAC criterion three evidence file is missing. It will produce a confident, generic answer. That answer is worse than no answer, because staff act on it.

Evaluate on four axes. Groundedness: can the system answer from your syllabus, circulars and past papers with citations? Ownership: who holds the index? Administration: does the role model match how an institution actually works — chairman, principal, HoD, faculty, student? Cost trajectory: does the bill grow with every new user, or is it institution-wide?

Per-seat suites lose on all four for institutional work, and win on general capability. That is why the sensible architecture is not either-or: an institution-owned platform that handles institutional questions, with general capability available where it genuinely helps.