What Are AI Agents?

1. What Are AI Agents?

5.1 What Are AI Agents?
LOs: [5a]
Many automation systems follow fixed steps, but some problems need more flexibility. When the work changes from case to case, an AI agent can help because it can use a goal, make choices, and carry out more than one step before it returns an answer. Understanding this difference matters because it helps you choose the right tool for the task instead of forcing every problem into one workflow.
As you saw in Unit 4 when you explored generative AI in automation, AI can create content or respond to prompts inside a workflow. This subunit builds on that idea by showing a stronger form of AI use: instead of producing one output and stopping, an agent can act over several steps and adapt its behavior to the situation. That makes it different from a traditional workflow, which usually follows a fixed path from trigger to action.
A traditional workflow is like a set sequence. If a form is submitted, then an email is sent, then a record is updated. An AI agent is more flexible. It can receive a task, decide what information it needs, choose an action, and continue until it reaches a result. In this reading, the key point is not that agents replace workflows, but that they are useful when the task cannot be fully described in advance.
The source explains this difference in a practical way by comparing an agent to a digital assistant that can plan and act across multiple steps. For example, if you ask an agent to help organize incoming customer messages, it may sort the messages, identify the urgent ones, and prepare a response draft rather than only following one fixed rule. That kind of behavior shows why agents are useful for tasks that require judgment, not just repetition.
Building on the process mapping work from Unit 2, you can now ask a new question: which parts of a task need a fixed workflow, and which parts need flexible decision-making? As you read, consider how an AI agent changes the way you think about automation when the path from start to finish is not always the same.
Attribution
Title | Author | Source | License
PRDV000 Unit Readings | Saylor Academy | https://learn.saylor.org/ | CC BY 4.0

OUTPUT LICENSE: CC BY 4.0

5.1: What Are AI Agents?
LOs: [5a]
AI agents are software tools that can take input, decide on a next step, and act toward a goal with limited human direction. Traditional workflows follow fixed rules, but agents can use language understanding to choose actions when the path is not fully predictable.
As you saw in Unit 4 when you worked with AI-enhanced workflows, generative AI can help with tasks such as writing, summarizing, and classifying text. An AI agent goes one step further because it does not only produce content; it also selects actions inside a process.
A traditional workflow usually moves in a set order: trigger, action, and condition. An agent can support a more flexible pattern by reading a request, checking context, and deciding whether to answer, ask for more detail, or hand off the task. For example, in a customer support setting, an agent might sort a message, draft a reply, and decide whether the issue needs human review.
The key difference is decision-making. In a rule-based workflow, you define the path in advance. In an agent-based approach, you define the goal, the available tools, and the limits, then let the agent choose among possible steps. This makes agents useful when a task changes from case to case, but it also means you must design clear boundaries.
Job-ready Tip: Start with one narrow goal
If you want to use an agent in real work, begin with one task that has a clear outcome, such as classifying support tickets or drafting first responses. Small goals make it easier to test whether the agent is reliable before you connect it to a larger workflow.
As you read this section, consider which tasks in your own work need fixed rules and which tasks need flexible decisions, and explain why an agent would help in one case but not the other.
SOURCE: [5.1 What Are AI Agents?]
ATTRIBUTION:
AI Agents and Hybrid Automation | Saylor Academy | Saylor Academy Unit Readings | CC BY 4.0
OUTPUT LICENSE: CC BY 4.0
5.1: What Are AI Agents?
This reading supports Unit Learning Outcome 5a.
As you saw in Unit 4 when you studied AI-enhanced workflows, automation can use generative AI to help with a task inside a larger process. AI agents go a step further because they can take a goal, choose actions, and adjust their next step based on what they observe.
A traditional workflow follows a fixed path. You define the trigger, the actions, and the conditions ahead of time, and the system carries out that path in the same way each time. This works well when the steps are clear and predictable. By contrast, an AI agent works toward a goal with more flexibility. It can decide what to do next, use information it gathers along the way, and continue until it reaches the target or stops for a reason you define.
Traditional workflow
Example: A form submission creates a record, sends a confirmation email, and then alerts a team member.
Best Use: Repetitive tasks with stable steps.
Limitation: It does not adapt well when the situation changes.
AI agent
Example: A support assistant reads a user request, decides whether it needs a summary, a draft reply, or a handoff, and then takes the next action.
Best Use: Tasks that need judgment, step-by-step reasoning, or flexible follow-up.
Limitation: It can make unexpected choices, so you need clear limits and oversight.
The main difference is control. In a workflow, you control each step before the process runs. In an agent, you define the goal and the boundaries, but the agent may decide how to reach the goal. That means agents can handle more open-ended work, but they also need stronger checks because they do not always follow one fixed route.
A simple real-world example is customer support. A traditional workflow may send every message with a certain keyword to the same email inbox. An AI agent can do more than route the message: it can read the request, identify the likely issue, gather context, and prepare a proposed response before a person reviews it. The workflow is rule based; the agent is goal driven.
Job-ready Tip: Choosing the right tool
If you are deciding between a workflow and an agent, ask one question: "Do I need a fixed sequence, or do I need flexible decision-making?" Use a workflow when the steps are stable. Use an agent when the task needs adaptation, but keep human review in place when the outcome matters.
As you read, think about one process you already mapped in this unit: which parts are better handled by a traditional workflow, and which parts would benefit from an AI agent that can make choices?
Sources / Further reading
(Authored by Saylor Academy, CC BY 4.0)
Estimated reading time: 6 minutes
5.1: What Are AI Agents? (covers 5a, 5c)
Intelligent Agents (Intro CS Textbook) — Kansas State CS Open Textbook — cc-by-nc-sa