Demystifying AI agents for higher ed: From buzzwords to building blocks

"AI agents" are everywhere: in vendor pitches, board conversations, team meetings and campus strategy plans. But the vocabulary moves fast, and the terms are often used loosely. What makes something an agent and not a chatbot? How does an agent decide to call a tool, look up information or hand a task off to another system? And which of these ideas actually matter for your institution?

This session explains how AI agents work, from the basics to the more technical building blocks. Every concept is tied to a university use case: answering prospective-student questions, pointing current students to the right resources, mapping courses to program outcomes, and taking routine requests off busy staff. Newcomers will leave with a solid foundation. Those already experimenting with AI will get a clearer picture of how the pieces fit together.

You will also get free access to the Webless Agent Builder so you can build your own agent after the session. If time permits, we'll build one together as a group, live.

What you will learn:

• Core concepts: LLMs, prompts, context windows, tokens and how an agent differs from a chatbot or assistant
• How agents act: tool calls (function calling), skills, memory, planning and reasoning loops, and multi-agent workflows
• How agents connect: retrieval-automated generation, embeddings and vector search, APIs and model context protocol
• How agents stay reliable: guardrails, hallucinations, grounding, human-in-the-loop review and evaluation
• Where to start: high-value, low-risk agent use cases in enrollment, student services and curriculum mapping, plus how to separate real capabilities from vendor hype