What are agents?

Agents were a major topic of global discussion throughout 2024, but what exactly are they? In short, agents are automations. In the context of AI and Large Language Models (LLMs), agents refer to AI tools designed to autonomously perform tasks, interact with environments, and make decisions based on user instructions or predefined goals.

Here’s a simple example: Imagine you have a highly efficient research assistant. You give them a task:

"Find the most recent articles on climate change and summarize them for me."

A regular search engine, like Google, would return a list of links, requiring you to read everything yourself. But an AI agent goes further:

1. Searches for articles from academic databases and news sources.
2. Summarizes key points from multiple sources.
3. Formats the information into an easy-to-read summary.
4. Properly cites the sources in MLA or APA format.

This is what an AI agent does—it understands your request, takes action, and delivers results without requiring you to complete each step manually.

Developing Agents at the AI Learning Centre

Now that we’ve established what agents are, here’s what we at the AI Learning Centre have been focusing on during the first two months of 2025:

We have been researching, creating, and testing over 50 generic agents. By "generic agents," we mean agents designed to be flexible enough to work across different cases, individuals, and organizations. While they may not be as powerful as an agent specifically tailored for an institution like the University of Nicosia, they serve as a valuable foundation for developing more complex workflows.

Here are some interesting examples of generic-level agents:

• Interactive Activity Brainstorm Agent
  Purpose: Provides creative, student-centered classroom activities (e.g., debates, simulations, role-plays) aligned with learning outcomes and fostering personal engagement.
• Rubric-Creation Agent
  Purpose: Automatically generates or refines detailed grading rubrics for assignments, ensuring clarity, fairness, and alignment with higher-order thinking skills.
• Research Project Guidance Agent
  Purpose: Assists students in designing and refining research projects by integrating AI-driven and non-AI tasks while emphasizing ethical considerations and personal reflection.
• Peer Review Framework Agent
  Purpose: Develops structured peer review processes that encourage meaningful critique, higher-order thinking, and genuine student engagement, reducing over-reliance on AI.
• Student Enrollment Data Summarizer Agent
  Purpose: Processes raw enrollment data (from CSV or Excel) to produce clear, concise summaries for institutional planning, accreditation reports, or departmental insights.
• Campus Event Logistics Agent
  Purpose: Creates comprehensive, step-by-step logistical plans for campus events (e.g., conferences, open houses, workshops), covering timelines, budgets, resources, and contingency measures.
• Staff Training Module Creator Agent
  Purpose: Designs concise, targeted training modules for university staff on topics such as new software, HR policies, and data protection, incorporating learning objectives, interactive tasks, reflection points, and assessments.
• Document Translation & Localization Agent
  Purpose: Assists staff in translating official university documents into multiple languages while ensuring accuracy, cultural sensitivity, and adherence to institutional standards.

How Do These Agents Work?

One might wonder how these agents operate. The answer lies in the use of variables. Variables have many definitions, but within PowerFlow, they allow us to designate fields that request user input.

For example, when creating an agent, I might write:

"Here is the existing assignment idea I have: <<var:Copy paste your assignment idea here!>>"

This notation tells the agent that the assignment idea will be provided by the user. When the agent is executed, the user will see a text field labeled "Copy paste your assignment idea here!" where they can enter their content.

When a lecturer runs the agent, they won’t see its internal workings—only the input fields they need to fill (if any) and the final results.

The Future of AI Agents

In summary, AI agents will be a powerful tool for everyone once implemented. The AI Learning Centre is committed to developing highly customized agents that will help maintain UNIC’s position as a cutting-edge university in Europe.