The Agentic Turn in Higher Education: Ambition, Cost, and the Reality of Enterprise AI at UCC

Across higher education, the “agentic turn” has quickly moved from idea to expectation. At University College Cork (UCC), staff are asking for AI agents to help automate tasks, support decisions, and improve everyday services. From administration to teaching, interest in these tools is growing across the University.

This presentation offers a grounded, practical perspective on what happens when ambition meets enterprise reality. Drawing on UCC’s experience supporting enterprise AI adoption and emerging agent use cases, this session explores three key tensions shaping the agentic landscape:

1. Expectation vs Reality

   Colleagues often imagine highly independent AI agents that understand a specific area of work, can reason through complex issues, and connect smoothly with enterprise systems. In practice, delivering these requires significant design effort, governance, and technical constraints that are often underestimated.

2. Innovation vs Cost

   While AI agents offer significant potential, the costs involved in building, running, and expanding them can become difficult to manage at scale. This creates a challenge: the more interest and demand grow, the harder it can be to sustain the work financially.

3. Empowerment vs Control

   Enabling staff to build and use AI agents supports UCC’s Securing Our Future: Strategic Plan 2023–2028, particularly Goal 5.3 to “enhance our digitally connected campus to enable greater innovation, agility and excellence, transforming how we work and learn. However, it also creates new challenges around managing agents safely, including oversight, security, data protection, and keeping agents up to date.

The session will highlight real examples from UCC’s Digital Advisory Centre and Enterprise Solutions Group, including:

• Patterns in staff demand for agents

• Misalignment between user expectations and current platform capability

• Early approaches to controlling cost and managing access

• Lessons learned from initial agent-related service requests and advisory engagements