The dramatic enhancement in the capability of AI as it moves from bots to agents will bring about efficiencies and have a far greater impact on the day-to-day operations, strategies and effectiveness of our institutions. We will become less expensive, more personalized and more responsive to students and employers. Those are big claims, so for this column, I turned to my personal assistant, Google Gemini 2.5 Pro on Nov. 1, 2025, to help me with identifying the pathway to those outcomes.

Caveats

In this article, I can only hit the high points along the way in developing an agentic AI–enhanced university. This is merely a broad-brush, incomplete view that each university should adjust and enhance in order to meet its structures and needs. I also encourage you to read the valuable McKinsey article “Deploying agentic AI with safety and security: A playbook for technology leaders,” published on Oct. 15, 2025, for important security and safety information.

Gemini Pro prefaced its recommendations with:

This will take significant thought and planning across the university. It will be important that we include all members of the university community to make this a coordinated, comprehensive change that will significantly advance the efficiency, effectiveness and relevance of the institution. Timing is important. We must begin immediately if we hope to have major changes in place before the end of 2027. Let’s begin!

First Half of 2026 Foundation and Vision

We will need an executive task force with the knowledge, resources and shared vision to accomplish this task. Gemini recommends we be sure to include:

• Chief information officer: To map the data and systems.
• Chief financial officer: To identify cost-saving opportunities and fund pilots.
• Provost: To champion the academic quality and student-facing initiatives.
• VP of enrollment: To represent the entire student life cycle (recruitment to alumni).
• VP of operations: To represent the “back office” (HR, grounds, facilities).

The executive task force will want to set up opportunities for input and support of the initiative. Perhaps the first step will be to seek ideas of whether the first order of priority should be quality improvement (hyperpersonalization of services to the learners) or cost efficiency (operational excellence). Both of these will be needed in the long run in order to survive the agent-enabled competition that will be both of higher quality and less expensive. In seeking input on this choice, universitywide awareness can be fostered. Perhaps a broad university forum could be scheduled on the topic with smaller, targeted follow-ups with faculty, staff, students, administrators and external stakeholder groups scheduled as the initiative proceeds.

One of the first steps of the executive task force will be to perform a universitywide Agent Readiness Audit. Since agents run on data and processes, we need to identify any data silos and process bottlenecks. These will be among our first priorities to ensure that agents can perform work smoothly and efficiently. Resolving these may also be among the most time-consuming changes. However, removing these data roadblocks can begin to show immediate progress in responsiveness and efficiency.

Second Half of 2026 Into Spring 2027 Pilot and Infrastructure

Gemini suggests that a good starting point in the summer of 2026 would be to set up two pilots:

• Cost-Saving Pilot: The Facilities Agent
• Goal: Reduce energy and maintenance costs.
• Action: An AI agent integrates with the campus event schedule, weather forecasts and the building HVAC/lighting systems. It autonomously adjusts climate control and lighting for actual use, not just a fixed timer. It also fields all maintenance requests, triages them and dispatches staff or robotic mowers/vacuums automatically.

• Quality-Improvement Pilot Example: The Proactive Adviser Agent
• Goal: Improve retention for at-risk students.
• Action: An agent monitors student data in real time (LMS engagement, attendance, early grade-book data). It doesn’t replace the human adviser. It acts as their assistant, flagging a student who is at risk before the midterm and autonomously executing a plan: sending a nudge, offering to schedule a tutoring session and summarizing the risk profile for the human adviser to review.

Our most significant centralized expense will be to set up a secure digital sandbox. The pilots cannot live on a faculty member’s laptop. The CIO must lead the creation of a central, secure platform. This sandbox is a secure environment where AI agents can be developed, tested and given access to the university’s core data APIs (e.g., SIS, LMS and ERP).

Gemini reminds me that, concurrently, we must set up a new entity. The generative AI rules were about plagiarism. The agentic AI rules must be about liability. The new entity is a kind of Agent Accountability Framework. It deals with policy questions such as:

• Who is responsible when an agent gives a student incorrect financial aid advice?
• What is the off-switch when an agent-driven workflow (like course wait lists) creates an inequitable outcome? Who has authority to flip the switch?
• By whom and how are an agent’s actions audited?

Implementation Across University Through Fall 2027

There will be many personnel and staffing topics to address. By the summer of 2027, we should be well on the way to refining roles and position descriptions of employees. The emphasis should be efficient, enhanced redesign of roles rather than staffing cuts. Some cuts will come from normal turnover as staff find more attractive opportunities or retire. In most cases, employees will become much more productive, handing off their redundant, lower-level work to agents. For example, Gemini Pro envisions:

• The admissions counselor who used to answer 500 identical emails now manages a team of AI agents that handle the routine questions, freeing the counselor to spend one-on-one time with high-priority applicants.
• The IT help desk technician no longer resets passwords. The technicians now train the AI agent on how to troubleshoot new software and directly handle only the most complex, level-three issues.
• The human adviser now manages a caseload of 500 students (not 150), because the AI assistant handles 90 percent of the administrative churn, allowing the adviser to focus on high-impact mentoring.

Gemini Pro suggests that this approach can result in a higher-quality, more efficient university that will be able to compete in the years ahead. The final step is the most critical and is the job of everyone, from the president and board on down. We must champion a culture where AI agents are seen as collaborators, not replacements. This is a human-AI “co-bot” workforce.

The institutions that win in 2027 will be those that successfully trained their managers to lead mixed teams of human and AI employees. This is the single greatest competitive advantage one can build.

This framework will position the university not just to survive the agentic AI wave but to lead it, creating an institution that is both more efficient and, critically, more human-centered.