Learning & Family Apps
Status: in development. Child, teacher, and parent apps share one data spine; two child activities built, the rest have research-anchored stubs. Parent app is the current milestone.
Three apps on one record. The child sees a game with no scores. The teacher sees who needs support today and why. The parent sees the real data, in plain language, with every claim tied to a named piece of early-childhood research.
Three views, one record
| App | Who | What they see |
|---|---|---|
| Child | 6 months to 5 years, tablet | Tap, drag, listen. Warm, tactile, voice-first. Never a number, a percentage, or a failure state. Wrong answers bounce back; right ones lock in with a pop |
| Teacher | Classroom of about twelve, tablet or desktop | Who needs support today, a per-child skill profile they can defend to a coach, one-tap activity assignment, and a "why" paragraph with every suggestion |
| Parent | Phone | Weekly summary, milestones as they happen, the full data (mastery, trajectory rung, minutes, attendance), and the research behind every line |
The problem it is built against
Early-learning tools tend to fail one of two ways: they gamify with streaks and scores that pressure toddlers, or they hand adults an opaque "AI recommendation" nobody can tie to a recognised framework. Teachers and parents cannot defend what the child is doing to a coach, coordinator, or skeptic.
This is built to do neither. The child never sees a score. Every suggestion an adult sees names its construct and its zone.
How a suggestion is made
- The child plays an activity. The app records hesitation, pronunciation, and correctness, inside the school's environment.
- The result is mapped to a construct on a named framework: a rung on the sound ladder, a stage on a counting trajectory, a strand of the reading rope.
- The AI proposes the next activity, labelled Consolidate / Target / Stretch, and writes the teacher a one-paragraph "why."
- The teacher accepts, changes, or ignores it. The child never sees the suggestion.
- The parent sees the outcome in plain language, with the framework and the source linked.
Design rules
- The child never sees a number. Adults, including parents, see the full data.
- Voice-first. Child interactions are tap, drag, and listen. Speech is used to gauge confidence, not to grade.
- No open chat. The AI never talks to the child. See what the AI does and never does.
- Not used to train models. Data stays inside the school's environment. The legal frame explains why that is architectural.
- Accessible by default. Built to WCAG 2.2; status is never conveyed by colour alone.
The frameworks are established; the apps' effect on outcomes is not yet measured. The parent app is read-only in its first version (no messaging, single child). Multi-child, two-way messaging, and the remaining child activities are later stages, not promises.
Sources
See Learning frameworks for the full source list.