Universities are moving beyond chat-based tools toward agentic AI—systems that act autonomously to perform tasks—and some institutions are framing the technology as a colleague rather than a tool. Leaders and instructional designers say agentic systems can automate advising, personalized tutoring and administrative workflows but raise governance, privacy and academic‑integrity concerns. Higher-education technologists emphasise that integrating agentic AI requires institutional workflows, retrieval systems and governance controls, not just model access. Campus IT and teaching-and-learning centers are piloting agentic agents for tasks like course scheduling, academic alerts and research assistance. Experts caution that agentic AI amplifies risks around data governance, purpose limitations and auditability; institutions must define permissible actions, kill‑switches and provenance standards. A one-sentence technical note: “agentic” AI refers to systems that can take autonomous, multi‑step actions on behalf of users. Stakeholders say 2026 will be a testing year—pilots will determine which administrative and pedagogical applications scale and which require stronger human oversight.