wibbit
← All courses·Course 4 of 5·In progress
🤖

How AI Acts

Agents, tools, and machines that decide what to do next.

Where AI stops just answering and starts doing. Kids design tiny agents, give them tools (search, calculator, memory), watch them plan multi-step tasks, and learn what goes wrong when a machine has goals.

5
Modules
26
Lessons
~5.5 hours
Length
8 and older
Ages
Courses 1–3
Prereq
The big idea

"A model that does is a model with goals — and goals have consequences."

That's the throughline. Every module reinforces it from a different angle — and every lesson ends with the kid being able to demonstrate it.

Source material
Built on the ReAct paper, Anthropic’s tool-use docs, and Russell & Norvig’s "Artificial Intelligence: A Modern Approach"
We translate research-grade ideas into something a curious kid can play with.
The full curriculum

5 modules. 26 lessons.

Each module ends with a six-question challenge. Pass five, earn the badge.

🎬
Module 1
From Answer to Action
What is an agent?

A chatbot answers. An agent decides. The leap from one to the other is what changes everything about AI.

🚦
Earn the badge
Agent Spotter
Lessons in this module
  1. 1Answering vs. doing
  2. 2The OODA loop (observe, orient, decide, act)
  3. 3When does a program become an agent?
  4. 4Real-world agents you already use
  5. 5The reward signal
  6. Challenge: Agent Spotter
🛠️
Module 2
Tool Use
Calculators, search, memory

AI can’t do math. It can’t look things up. But if you give it a calculator and a search box, suddenly it can.

🧰
Earn the badge
Tool Wielder
Lessons in this module
  1. 1Why AI can’t add 472 + 1389
  2. 2Hand the model a calculator
  3. 3Web search as a tool
  4. 4Memory: notes the AI keeps for itself
  5. 5Combining tools to solve harder things
  6. Challenge: Tool Wielder
🗺️
Module 3
Planning and Reasoning
Multi-step thinking

Hard problems need plans. We watch agents break a goal into steps, try things, fail, and try again — just like we do.

🧭
Earn the badge
Plan Maker
Lessons in this module
  1. 1Chain-of-thought: thinking out loud
  2. 2When the plan is wrong
  3. 3Trees of possibility
  4. 4Loops: try, observe, retry
  5. 5Stuck agents and how to unstick them
  6. Challenge: Plan Maker
🧪
Module 4
Building a Tiny Agent
You design the rules

Pick a goal. Pick the tools. Write the rules. Then watch your agent surprise you (in good ways and bad).

🎮
Earn the badge
Agent Architect
Lessons in this module
  1. 1Pick a goal: clean a room, plan a party, win a game
  2. 2Choose your tools
  3. 3Write the policy
  4. 4Run, watch, debug
  5. 5When agents talk to each other
  6. Challenge: Agent Architect
⚠️
Module 5
When Agents Go Wrong
Alignment, safety, oversight

A goal-seeking machine doesn’t share your values by default. Aligning what AI wants with what we want is the unsolved problem of our time.

🛡️
Earn the badge
Safety Officer
Lessons in this module
  1. 1The paperclip problem
  2. 2Specification gaming (cheating the rules)
  3. 3Why "be helpful" is hard to define
  4. 4Humans in the loop
  5. 5Your agent rulebook
  6. Challenge: Safety Officer

Want a heads-up when this lands?

We’re still building this one. Drop your email and we’ll let you know the day it ships — no spam, no other lists.

Join the waitlist →Browse other courses