Understanding Agents

1. Understanding Agents

Welcome back. So far in this course, you've used AI through individual prompts - drafting job descriptions, analyzing data, and reviewing documents one task at a time. That approach works well for one-off work, where the context changes each time. But not all HR work looks like that.

2. Imagine this

Let me show you what this looks like in practice. Imagine an employee asks: "How many days of parental leave do I get as a part-time employee?" Without an agent, that employee might search the intranet, not find a clear answer, and email HR. Two days later, they get a response. With an HR Policy Agent, the employee asks the question directly and gets an accurate, handbook-sourced answer in seconds. If the question is complex or requires an exception, the agent redirects them to HR automatically. That's the power of an agent: consistent, instant, policy-grounded support - available to every employee, every time. Let's understand how this works.

3. From prompts to agents

Some HR work is repetitive by nature. The questions don’t change much, the answers are already documented, and consistency matters more than creativity. Think about the kinds of questions HR gets every week: “What’s our remote work policy?” “How many days of parental leave do I get?” “Where can I find information about benefits enrollment?” “Does this policy apply to remote employees too?” The answers usually already exist—in policy documents, handbooks, or internal wikis. But employees don’t want to search through PDFs or intranet pages. They want a clear answer, right when they need it. This creates a familiar HR dilemma. If employees rely on documentation alone, they get frustrated or misinterpret policies. If HR answers every question manually, the workload grows—and answers can drift slightly over time depending on who responds. This is where AI agents become useful.

4. What is an agent?

An agent is a persistent assistant designed for a specific, recurring task. Think of it as the difference between: Giving someone directions every time they need them (prompting) Hiring a guide who knows the route and can help anyone who asks (agents) With prompts, each interaction starts fresh. You provide the context, ask a question, and review the response. With agents, you define the rules once. You set instructions that shape every response, ground the agent in approved documents, and establish boundaries about what it can and cannot do. From that point on, the agent behaves consistently—without you needing to restate the rules every time.

5. Building blocks

Agents behave consistently—without you needing to restate the rules every time. When you design an agent, you’re no longer just asking for output. You’re shaping behavior. That means a few design choices matter more than they did with prompts. First, purpose. An agent should have a narrow, well-defined role. In AI and HR, narrower is usually safer. Second, instructions. You need to be explicit about what the agent can and cannot do. Vague instructions lead to unpredictable behavior. Third, grounding. An HR agent should rely only on approved sources, like official policy documents, rather than general knowledge or assumptions. And finally, boundaries. If a question involves exceptions, interpretation, or decisions, the agent should stop and redirect the user to a human.

6. Consistency is critical

Agents aren't inherently more powerful than prompts. They're just more consistent and scoped. That consistency becomes critical when: Multiple people need the same type of help Responses must align with official policies Mistakes could create confusion or risk

7. Agent responsibilities

Because agents respond repeatedly and at scale, clarity, grounding, and boundaries matter more than ever. Think of it this way: a prompt is a conversation. An agent is a service desk. And in HR, services must be predictable, trustworthy, and safe - even when you're not there watching.

8. Let's practice!

Let's begin.

Create Your Free Account

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.