🤖
← Owens AI Institute

AI Agents & Automation

Build, orchestrate, and supervise autonomous AI agents.

Elementary

What Is an AI Agent?

Discover what makes an AI agent special: it does not just talk, it takes action to reach a goal. Kids meet agents they already know and learn that every agent has a job and follows steps.

  1. 1Talkers and Doers
  2. 2Meet an AI Agent
  3. 3A Day With AI Helpers
  4. 4What Can an Agent Do?
  5. 5Agents You Already Know
  6. 6An Agent Has a Job
  7. 7Agents Follow Steps
  8. 8When Is an Agent Done?
  9. 9Agent Spotting Game
  10. 10You and Your AI Helpers

Sense and Act

Agents work in a loop: they sense the world, think about it, and act, then check what happened. This module teaches the sense-think-act cycle that powers every AI agent.

  1. 1Eyes, Ears, and Sensors
  2. 2What an Agent Notices
  3. 3From Noticing to Doing
  4. 4The Sense-Think-Act Loop
  5. 5Acting in the Real World
  6. 6Acting on a Screen
  7. 7Did It Work?
  8. 8Trying Again
  9. 9Sense-Think-Act Game
  10. 10The Loop Keeps Going
  11. 11Lab: Agent Helper

Giving an Agent a Goal

An agent is only as good as its goal. Kids learn to set clear goals, break big goals into small steps, and build simple plans an agent can follow.

  1. 1What Is a Goal?
  2. 2Big Goals, Small Steps
  3. 3Telling an Agent What You Want
  4. 4Clear Directions, Fuzzy Directions
  5. 5A Plan Is a List of Steps
  6. 6When a Step Does Not Work
  7. 7Picking the Best Next Step
  8. 8Reaching the Goal
  9. 9Plan-It Challenge
  10. 10You Are a Planner

Agents Working Together

One agent is helpful, but a team of agents is powerful. Kids learn how specialist agents pass work along, follow a team leader, and get more done together.

  1. 1One Helper or Many?
  2. 2Each Agent Has a Job
  3. 3Passing Work Along
  4. 4The Team Leader Agent
  5. 5Agents Talking to Agents
  6. 6A Team Makes Lunch
  7. 7When Teammates Disagree
  8. 8Faster Together
  9. 9Build-a-Team Game
  10. 10Teamwork Makes It Work

Being the Boss of Your Agents

Humans are always the boss of their AI agents. Kids learn to check an agent's work, use the stop button, set rules, and keep themselves safe.

  1. 1Who Is in Charge?
  2. 2Agents Make Mistakes
  3. 3Checking an Agent's Work
  4. 4The Stop Button
  5. 5What an Agent Should Not Do
  6. 6Ask First, Then Do
  7. 7Keeping Secrets Safe
  8. 8When to Ask a Grown-Up
  9. 9Agent Boss Training
  10. 10A Good Agent Boss

Capstone Project: Design a Helper Agent

Plan an AI agent that senses, thinks, and acts to help with a chore.

  1. 1Project: Design a Helper Agent

Middle

From Chatbot to Agent

The leap from chatbot to agent. Students learn what separates a system that just produces text from one that takes autonomous action with tools to reach a goal.

  1. 1The Limits of a Chatbot
  2. 2What Makes an Agent Different
  3. 3Autonomy: Acting Without a Human Each Step
  4. 4Tools: An Agent's Hands
  5. 5The Agent Mindset
  6. 6A Real Agent in Action
  7. 7Where Agents Shine
  8. 8Where Agents Struggle
  9. 9Agent or Not?
  10. 10Module Check: The Big Leap

The Agent Loop — Perceive, Think, Act

Every AI agent runs a loop: perceive, think, act, observe — and repeat. Students trace the core cycle and meet the ReAct pattern that powers modern agents.

  1. 1The Core Cycle
  2. 2Perceive: Gathering Information
  3. 3Think: Deciding What to Do
  4. 4Act: Taking the Action
  5. 5Observe: What Happened?
  6. 6Looping Until Done
  7. 7Reasoning Out Loud
  8. 8The ReAct Pattern
  9. 9Trace the Loop
  10. 10Module Check: Loop Mastery
  11. 11Lab: Agent Loop Game

Tools, APIs, and the Outside World

Tools are an agent's hands. Students learn what tools and APIs are, how agents choose and call them, and why powerful tools demand care.

  1. 1Why Agents Need Tools
  2. 2What Is a Tool?
  3. 3Search, Calculator, Calendar
  4. 4APIs: How Software Talks to Software
  5. 5Reading Tools vs. Doing Tools
  6. 6Choosing the Right Tool
  7. 7When a Tool Fails
  8. 8Tool Safety
  9. 9Tool Box Challenge
  10. 10Module Check: Tools Extend the Mind

Planning and Breaking Down Big Tasks

Big tasks need a plan. Students learn to decompose goals into subgoals, sequence steps, and build plans that adapt when things go wrong.

  1. 1Big Tasks Are Hard
  2. 2Decomposition: Splitting the Work
  3. 3Subgoals and Milestones
  4. 4Order Matters
  5. 5Plans That Adapt
  6. 6Dead Ends and Backtracking
  7. 7Estimating Effort
  8. 8The Planning Agent
  9. 9Plan a Project
  10. 10Module Check: Think Before You Act

Multi-Agent Systems and Automation

When one agent is not enough, agents team up. Students explore supervisor agents, handoffs, workflows, and automation that runs itself.

  1. 1One Agent vs. Many
  2. 2Specialist Agents
  3. 3The Supervisor Agent
  4. 4Handoffs and Messages
  5. 5Workflows and Pipelines
  6. 6Automation: Work That Runs Itself
  7. 7Triggers and Schedules
  8. 8When Automation Goes Wrong
  9. 9Design a Workflow
  10. 10Module Check: Systems That Run Themselves

Capstone Project: Build an Automation

Design a multi-step agent that gets a task done on its own.

  1. 1Project: Build an Automation

High School

The Architecture of an Agent

An AI agent is an architecture, not just a model. Students dissect the modern agent stack — reasoning core, planner, tool layer, memory, and orchestration loop.

  1. 1The Agent as a System
  2. 2The LLM Reasoning Core
  3. 3The Planner
  4. 4The Tool Layer
  5. 5The Memory System
  6. 6The Orchestration Loop
  7. 7Agent Frameworks
  8. 8Anatomy of a Real Agent
  9. 9Diagram an Agent
  10. 10Module Check: Architecture Matters

Tool Use and Function Calling

Tools turn a language model into an agent. Students learn function calling, schemas, structured outputs, the tool-call round trip, and tool security.

  1. 1How Agents Call Tools
  2. 2Function Schemas
  3. 3Structured Outputs
  4. 4The Tool-Call Round Trip
  5. 5Designing Good Tools
  6. 6Error Handling in Tool Calls
  7. 7Security and Permissions
  8. 8Tool Use in Practice
  9. 9Write a Tool Schema
  10. 10Module Check: Tools, Done Right
  11. 11Lab: Tool Router

Memory, State, and Context

Models are stateless; agents need memory. Students explore context windows, short- and long-term memory, retrieval-augmented generation, and memory design.

  1. 1Why Agents Need Memory
  2. 2The Context Window
  3. 3Short-Term vs. Long-Term Memory
  4. 4Retrieval-Augmented Generation
  5. 5State Across Steps
  6. 6Forgetting and Summarizing
  7. 7Memory Failures
  8. 8Designing a Memory System
  9. 9Memory Design Challenge
  10. 10Module Check: Context Is Everything

Multi-Agent Orchestration

Hard problems demand teams of agents. Students study supervisor-worker orchestration, handoffs, communication protocols, and the risks of emergent behavior.

  1. 1Why Multiple Agents
  2. 2The Supervisor-Worker Pattern
  3. 3Agent Handoffs
  4. 4Communication Protocols
  5. 5Parallel vs. Sequential Agents
  6. 6Shared State and Coordination
  7. 7Emergent Behavior and Risks
  8. 8Orchestration Frameworks
  9. 9Orchestrate a Team
  10. 10Module Check: Coordinated Intelligence

Reliability, Evaluation, and Guardrails

Powerful agents must be trustworthy. Students learn failure modes, evaluation, test suites, guardrails, human-in-the-loop, and responsible deployment.

  1. 1Why Agents Fail
  2. 2Failure Modes
  3. 3Evaluation: Measuring Agent Quality
  4. 4Building Test Suites for Agents
  5. 5Guardrails and Constraints
  6. 6Human-in-the-Loop
  7. 7Monitoring and Observability
  8. 8Responsible Agent Deployment
  9. 9Audit an Agent
  10. 10Module Check: Trustworthy Agents

Capstone Project: Build a Tool-Using Agent

Design an agent that routes tasks to the right tools, with guardrails.

  1. 1Project: Build a Tool-Using Agent