Math is the language of relationships.
Change the number of packages. Watch the robot, graph and equation respond together.
"Move a package. Something else moves too. That's a relationship."
Move packages, drag the graph, tap the equation. No wrong answers.
Drag a package onto Rover · or tap · or use the slider.
Relationship Map
Prepared deterministic relationships- • Each package adds 2 energy units.
- • Each package adds 1 minute.
- • Each package adds 1.5 cost units.
Graph View
Drag the point. Watch the world follow the graph. Y-intercept = 4. Slope = 2 per package.
Reasoning Review
Let's inspect your thinking.
What relationship did you discover?
Experiment Notebook
Put two experiments together. What stayed the same?
No experiments yet. Record one to compare states.
AI for kids that actually helps them think.
KidsPiggy helps children turn ideas, notes, drawings and experiments into active learning experiences.
Math With Purpose makes invisible relationships visible.
Behind the Lab
The broader KidsPiggy platform already included:
- • Educational experiences
- • Notes transformed into study materials
- • Doodles transformed into developed illustrations
- • Learning through creativity
- • Math With Purpose standalone module
- • Robot Delivery Relationship Lab
- • Synchronized world, map, graph and equation
- • Flow, Cause & Effect and Prediction missions
- • Prediction-before-reveal interaction
- • Deterministic Reasoning Review Engine
- • Socratic prompts and experiment comparisons
- • Responsive accessible demo
GPT-5.6 — Learning & product architecture
Used to define the educational philosophy, turn mathematics into relationship exploration, design the synchronized representations, structure Socratic prompts and establish learning boundaries.
The demo does not make runtime GPT-5.6 calls.
Codex — Engineering acceleration
Used to scaffold and implement the standalone application, separate the deterministic engine from React, build synchronized interactions, write 22 tests, improve accessibility and responsiveness, verify the production build and document key decisions.
Final educational, product and safety decisions remained human-led.
Educational boundaries
- Exploration support, not a replacement for a teacher.
- Reasoning prompts, not judgements of ability.
- Multiple explanations can demonstrate understanding.
- No learner profiling or diagnosis.
- No personal information required.
- No runtime AI or external model calls.
- Session-only experiment history.