lesson
25 minutes

AI Agents

AI agents don't just answer questions -- they take action. Learn what agents are, how they work, and why they're changing the way work gets done.

Up until now, AI has been mostly reactive. You ask a question, it gives an answer. You give a prompt, it generates text. It's a conversation -- you lead, AI follows.

AI agents are different. They don't just respond. They plan, decide, and act.

What Is an AI Agent?

An AI agent is an AI system that can:

  1. Understand a goal -- you tell it what you want to accomplish
  2. Break it into steps -- it figures out what needs to happen
  3. Take actions -- it actually does things (send emails, search the web, update databases)
  4. Adapt -- if something goes wrong, it adjusts its approach

Here's a simple example. If you tell ChatGPT: "What's the weather in San Antonio?" it gives you a text response about weather in general. But an AI agent connected to a weather service could actually check the current weather, see that it's 97 degrees, and then proactively suggest: "It's extremely hot today. Want me to reschedule your outdoor meeting to an indoor location?"

The difference: chatbots answer. Agents act.

How Agents Work

Under the hood, an AI agent has three key parts:

The brain (the AI model). This is the same kind of language model you've been using -- GPT, Gemini, Claude. It understands language and can reason through problems.

Tools. These are connections to external systems. A tool might let the agent search Google, read your email, update a spreadsheet, or post to social media. Without tools, an agent is just a chatbot. With tools, it can do real work.

Memory. Agents can remember context from earlier in a conversation -- and some can remember things across multiple conversations. This lets them learn your preferences and build on previous work.

Think of it like this: the AI model is the brain, tools are the hands, and memory is the notebook it carries around.

Real-World Examples

AI agents are already being used in workplaces around San Antonio and beyond:

Customer service agents handle support tickets from start to finish. They read the customer's issue, look up their account, find relevant help articles, draft a response, and escalate to a human only when necessary.

Scheduling agents coordinate meetings between multiple people. They check everyone's calendars, suggest times, send invitations, and handle rescheduling -- all without human intervention.

Research agents can take a question like "What grants are available for workforce development nonprofits in Texas?" and spend 15 minutes searching databases, reading grant requirements, and compiling a summary with deadlines and eligibility criteria.

Coding agents help developers by understanding a bug report, searching through code to find the problem, writing a fix, and testing it.

At Community Incubator, we use AI agents to help with outreach. Our agent can look up contacts, draft personalized emails aligned with our mission, and queue them for review -- work that used to take hours.

Why Agents Matter for Your Career

AI agents are about to change a huge number of jobs. Not by replacing people, but by changing what people spend their time doing.

Right now, a lot of work is process-driven: follow step 1, then step 2, then step 3. AI agents can handle those processes while humans focus on judgment, relationships, and creative problem-solving.

People who understand how agents work will be the ones who:

  • Design what agents should do
  • Supervise agents to make sure they're doing it right
  • Improve agents based on results
  • Decide where agents should and shouldn't be used

That last one is crucial -- and it connects directly to the judgment skills you built in Module 5.

The Human-Agent Partnership

The best way to think about AI agents isn't "AI replacing humans" or "humans doing everything manually." It's a partnership.

Imagine you're managing a community job fair. Without an agent, you'd spend hours on logistics -- emailing vendors, updating the event page, tracking RSVPs, posting on social media. With an AI agent, you could say: "Set up the spring job fair for March 20th at the Westside Community Center. Invite all employers from our database. Create social media posts. Send reminder emails one week before."

The agent handles the logistics. You handle the judgment calls: Which employers are the best fit? What should the event feel like? How do we make sure community members feel welcome?

That's the partnership. AI handles the process. You bring the purpose.

Building Simple Agents

You don't need to be a software engineer to create a basic AI agent. Tools like:

  • Zapier and Make.com let you create automated workflows with AI (no code needed)
  • CustomGPTs in ChatGPT let you create specialized agents with custom instructions and tools
  • Google's Agent Builder lets you create agents with access to Google Workspace tools
  • Microsoft Copilot Studio lets you build agents for business tasks

In the next lessons, you'll get hands-on experience with automation and building AI-powered tools.

Agents and Ethics

Remember the STAR framework from Module 5? It applies double for agents because agents take action in the real world:

  • Source: What data is the agent using to make decisions?
  • Test: How do you verify the agent is doing the right thing?
  • Audience: Who is affected by the agent's actions?
  • Reflect: Should an agent be doing this task at all, or does it require human judgment?

An agent that sends emails on your behalf, for example, is acting as you. If it sends something wrong, offensive, or inaccurate, that reflects on you, not the AI.


Try This

Go to ChatGPT and explore the "Explore GPTs" section (or search "GPTs" in the sidebar). Find three specialized agents (GPTs) that could be useful for job searching, learning, or community work. List what each one does and who it would help.


Talk About It

If you could have an AI agent handle one part of your daily routine, what would it be? What would you need to make sure the agent got right? What would you still want to do yourself?

Guardrails before you let an agent act

0 of 4 done