Skip to content
Trace
Students working at computers in a classroom

What can you make AI do?

Change a prompt and compare the answers. Connect blocks into an agent of your own, then follow each step to see how it works. No coding needed.

Start learning Try a lesson

Written for high school. Join through a class code, or start on your own.

Compare two prompts

Same model, same question, one change

Prompt A

Explain photosynthesis.

Plants use light energy to turn water and carbon dioxide into food.

Prompt B

Explain photosynthesis to a 10-year-old using an analogy.

A plant is like a tiny kitchen: sunlight powers the recipe that makes its food.

Eight words changed the reading level, the structure, and the length.

One run of the prompting lab. Real models, so your answers will differ.

Read it, change it, build it, watch it run.

That order is the whole thing. Every chapter ends with something you made.

Read a lesson

Short articles about models, prompts, data, tools, and agents, with a question to check yourself after each idea.

Change one thing in a lab

Swap a prompt or a setting and read the two answers side by side.

Build an agent

Connect blocks into a workflow that does a real job, then run it on a real model.

See what it did

Follow each step the agent took, including where something went wrong.

Build a game helper

An agent is a model with a job, some tools, and a loop. This one answers a player's question by checking two tools first. Every part of it is yours to change.

  1. 1

    Give it a job

    Answer a player's question about the farm. That is the whole instruction.

  2. 2

    Give it tools

    A crop guide and a season clock, so it checks real numbers instead of guessing.

  3. 3

    Run it and read the trace

    See which tool it asked, what came back, and how it wrote the answer. Then change something and run it again.

Game helper

The canvas is empty.

Building…

Can I grow melons before winter?

Waiting for the build to finish.

Message your agent
The real builder. Press Run agent and the answer comes from a real model.

Who can see your work?

Teachers see what you submit or choose to share. Your name is removed before anything goes to the AI, and the credits for every run are included.

Real models, no bill

You run on the same models professionals use. Credits come with your class or free account, so you never sign up for or pay a model provider.

Your name stays out of it

Trace removes your name and other identifiers before anything reaches the AI, and asks the provider not to store the request.

Only tools Trace provides

Agents can use the tools Trace gives them and nothing else, and responses go through a moderation check.

You choose what to share

Teachers see what you submit or share. You can turn a share off.

Common questions

A few things students ask before they start.

No. You build agents by connecting blocks on a canvas. Each lesson explains the idea before any code appears, and a few advanced ones include short Python examples.

Join through your class, or start on your own.

Read a lesson, change something in a lab, then build an agent that runs.

A free account starts with 2,000 credits a month for running your agents.

Create your account