Three machine abilities, built by hand — not prompted. One human discipline that audits all three. Every claim grounded in a framework you can read yourself.
Three things AI does. Your child learns each by building it from the parts — glass-box, never a black box.
How AI perceives. Pixels, text, audio, and sensor data become signals it can act on. Students train classifiers, explore datasets, and learn where perception fails.
Graphics labs · classifier training · dataset exploration
A student points a webcam at ten hand-drawn symbols and watches the classifier guess. It fails on the messy ones. Before retraining, the Navigator asks which feature it is missing — the student names the failure, then fixes the data, not the model.
How AI processes. Search, logic, planning, learning. Students implement these algorithms themselves in Python — not by calling libraries, but by building them. This is where Python becomes the technical spine.
Python · search / logic / ML implemented from scratch
No import statements. The student implements breadth-first search by hand to move a robot through a maze, then races it against A*. When it is slow, they can say exactly why — and what a heuristic buys them.
How AI generates. From rule-based output to foundation-model language. Students build with Claude and Google Gemini, learn to prompt as a form of writing, and verify every output.
Claude · Google Gemini · prompting, retrieval, agent patterns
The model returns a confident, wrong citation. The student does not reprompt — they identify which claim to verify first, check it against a known source, and write the override rule that catches it next time.
Every output AI produces is met with the same three habits. Critical thinking is the discipline that gives machine fluency its value. Societal impact — the fifth domain of machine literacy — is where this discipline pays out.
01
When a model misclassifies, students name the failure mode before retraining.
02
When Claude or Gemini outputs a confident answer, students identify which claim to verify first.
03
In advanced cycles, students design override conditions — explicit criteria for when the human stops the agent.
Just as English language arts teaches how language is read, thought through, spoken about, and written — and how the critical reader judges whether what is read is trustworthy — DODO Coding teaches how machine language is read, thought through, and written, with the same critical-reader discipline applied to AI output.
DODO Coding teaches five things, in our own framing. We teach the literacy — how machines see, how machines reason, how machines learn, how machines speak, and what all of that means for the society it is entering. We teach the progression — a Python curriculum that climbs from graphics sandboxes to a credit-bearing advanced track. We teach an optional extension — code that moves in the physical world, for students who want it. We teach the engineering — modern foundation-model practice, prompting as composition, retrieval, agents, verification. And we work in the environment your child will use professionally: Anthropic's Claude, Google's Gemini, Python in a real engineering shell, from the first session. Read each of these in detail on the curriculum page.
What we teach
Five domains of machine intelligence — perception, reasoning, learning, natural interaction, societal impact — taught by building each one. The literacy rubric measures every student, baseline to exit.
Read the literacy ↗A Python curriculum that climbs from graphics sandboxes to advanced algorithms. Four tracks; the Advanced track is credit-bearing — college credit that transfers.
Read the progression ↗Modern AI engineering practice. Prompting as composition, retrieval as memory, agents as architecture, verification as discipline. The post-2023 stack, taught the way working engineers use it.
Read the engineering ↗Anthropic's Claude and Google's Gemini, Python in a real engineering shell. Not toys. Not a sandboxed simulator. The same tools your child will use professionally.
Read the environment ↗Critical thinking is built into every session specifically so fluency never becomes dependence.
Curious about DODO Coding?
30 minutes, live, with a Navigator credentialed in the same standards top schools use. No pressure.