01
Trace the mechanism
Read an original lesson, work through an example, and check your understanding. Keep your own explanation beside the code.
A WORKBENCH FOR RELIABLE AIBuild agents
Build agents
you can explain.
Understand the mechanism. Write the Python. Test what breaks. Leave with working code and evidence of the decisions you made.
Free beta · Basic programming required · No API key needed
to “here’s the evidence.”
82Python labs
96Original lessons
32Mini projects
8Capstone blueprints
01 / THE LEARNING LOOP
Less watching.
More working things out.
This is an independent study workspace for people who can already read basic code. Every module connects an explanation to practice and a project.
02
Make the checks pass
Run real Python in your browser. Inspect a failing check, change the code, and try again. Reveal a hint when you need one.
03
Build the evidence
Take a mini project beyond a passing test. Record your baseline, measurements, failure cases, and limitations in an exportable evidence brief.
YOUR FIRST FIVE-MINUTE EXPERIMENTOne boundary.
One boundary.
A real engineering decision.
An agent has used its last tool call. Should it get another? Fix a small budget check, run three boundary cases, and explain why zero matters.
Open the first exercise Allow extra time for the first Python download. Full course labs have their own time estimates. budget_guard.py FIRST EXERCISE
def allow_tool_call(remaining_calls):
return remaining_calls >= 0
# A budget of zero should block a call.
# What needs to change? Positive budget allows a call
× Zero budget must block a call
Negative budget blocks a call
Illustrative starter output. Run the real checks in the exercise.
02 / CHOOSE YOUR DEPTH
A path from Python
to systems you can defend.
The complete core curriculum is about 395 hours. Start with one module and choose a pace you can sustain.
010203
Foundations & agent patterns
Python, model interfaces, tools, memory, and bounded execution.
Evaluation & reliable systems
Retrieval, test cases, observability, safety boundaries, and failure handling.
Portfolio & applied practice
Mini projects, capstone milestones, and reproducible evidence briefs.
03 / BEFORE YOU STARTA clear agreement
A clear agreement
about what you get.
- A free, self-paced beta with executable Python practice.
- A full keyboard and recent desktop browser work best for coding.
- Sign in with ChatGPT for saved progress, notes, code, and evidence across devices.
This beta does not include a paid review service, a hosted model API, certification, placement, or a job guarantee. Practice checks are learning feedback, not proof of production readiness.
04 / USEFUL ANSWERSA few things
A few things
you may be wondering.
Do I need to pay or bring an API key?+
No. The current beta is free. Python exercises run on your device. Some project briefs discuss external services, but the site does not make paid model calls on your behalf.
Can I try it before signing in?+
Yes. Browse the curriculum and try the first exercise as a guest. Guest work stays temporarily in this browser tab. Sign in and choose to import it if you want to keep it across devices.
Is this for complete programming beginners?+
You should be comfortable reading functions, variables, and simple conditions. The course develops AI engineering skills; it is not a first introduction to programming.
What can I show after completing a project?+
An evidence brief: the problem, baseline, how to reproduce your work, checks, results, and limitations. You write and export it. The site does not independently review or certify your submission.
How do I get help or delete my data?+
The course guide explains each workspace. Signed-in learners can open a support request and read the reply in the site. Your learning data page includes export, optional analytics, and deletion controls.