Teacher preparation

Try it on the classroom device.

Prepare a complete lesson, check the actual network, and gather evidence about what helps pupils learn.

This edition is a private owner preview. The learning activities do not require separate AI accounts, but the private site’s access restriction still applies. Prepare downloaded materials for an approved classroom pilot.

Before the lesson

1. Choose the lesson using readiness and preparation filters. Read its teacher pack, answer notes and curriculum scope. The 45- and 60-minute routes are focused investigations; longer projects have session boundaries.
2. Download the classroom ZIP, Word or printable journal and case cards. Keep sealed final cases separate. Open every needed file before class.
3. Use the exact managed device and student network. Open the core lab, change a control with the keyboard, download a project, reload it, and clear local work.
4. For Marimo, open the notebook and wait for its Python controls. Save a prediction, run a calculation, export a project and reopen it. First use downloads a Python runtime and packages. Allow the site, cdn.jsdelivr.net, wasm.marimo.app, files.pythonhosted.org and pypi.org as required by the school’s network policy; the observed requests in your browser are the final reference.
5. For native hardware, have IT install dependencies before class. Use the provided preflight and verify the actual device name. H100 is optional and user-provided; stop the provider’s billed runtime after saving work.

Plan for different ways of participating

Foundation–Year 2 learners can point, draw and dictate. Use the picture and arrow controls with an adult reading the prompt. Older learners can give an oral defence alongside their evidence table. Pair learners by complementary roles and rotate operator, predictor and evidence checker; retain individual explanations for assessment.

Use the Animations control or the device’s reduced-motion setting. Check 200% zoom, keyboard focus and the measurement tables. Wide evidence tables can scroll independently. A printed case pack and a teacher-operated projection are the fallback when individual devices or notebook downloads are unavailable.

Run a small teaching pilot

1. Before teaching, ask each pupil to predict one outcome and explain the rule in their own words. Accept pointing or an oral response.
2. Observe whether the pupil can start, change one thing, identify a result and recover their work. Record anonymous counts, time and the exact confusing instruction.
3. Use a new case after the lesson. Ask for a prediction, evidence and a limitation. Compare the explanation with the starting response; do not treat enjoyment or task completion alone as learning.
4. Moderate a few anonymous work samples against the supplied illustrative rubric responses. Note disagreements and revise the lesson or criteria before wider use.

What still needs external validation

Automated browser and numerical checks support engineering claims. They do not establish learning gains, universal accessibility or compatibility with every school device. Teacher and pupil pilots, screen-reader user testing, managed Chromebook/iPad/Windows network trials, and physical H100/DGX Spark/Mac Studio runs need their own recorded evidence. The current host’s Apple M3 results are not Mac Studio benchmarks.