CASE STUDY · K12 LEARNING APP
We built and launched a K12 learning app for an edtech product team—parent/student accounts, textbook units, dictation tasks, dot-matrix pen writing, AI essay review, phonics assessment, and points on one learning data stack. The hard part was never any single feature; it was making every scenario share one identity, task, result, and operations model—not stacking features, but making a learning product that actually runs.
Phonics set
8.5 / 10
Background
The product had to cover unit dictation, pen writing, error retry, AI essay review, picture-book sharing, phonics assessment, points and streaks, plus admin-side word lists, tasks, and rules. None of that is hard alone—but they cannot each carry their own accounts, tasks, and data.
01 / Identity
Whose dictation record is this? Who owns the essay? Where do points post? What does a parent see after switching students? One identity model has to answer that—not a different hack in every module.
Without clear ownership, no learning result can be trusted downstream.
19 errors
02 / Tasks
Textbook unit → word list → audio → writing → submit → result → errors → retry. Only when those states connect does dictation become operable by unit, not a one-off demo.
Tasks stay linked so content can update by unit over time.
03 / Results
Results need save, search, share, rework, picture-book generation, and a work library. We structured AI output and folded it into one work model so each essay submission keeps living in the product.
AI is a step in the work pipeline—not a standalone feature.
04 / Incentives
Phonics, writing, streaks, and other behaviors all earn points. Rules locked in code mean every tweak needs a release—so configuration and maintenance sit in the admin console.
Incentives can evolve after launch; the product stays operable.
The first problem to solve was how every learning capability shares one product data chain.
Core architecture
The project uses five business layers: identity → tasks → results → incentives → governance. The app owns learning experience; the admin console owns content and operations—the key architectural decision behind the whole build.
Core loops
Dictation, writing, and phonics look like different products—but all sit in one task system. Pen strokes, AI review output, and phonics reports become data the rest of the product can keep using.
8.5
userId: st_042Coexisting loops
Dictation, writing, and phonics differ in task type, outcomes, and next steps—but all resolve to student identity → learning task → result → incentive → governance.
if (type==='phonics') +5Product experience
After launch, what changes is not just app screens—content, tasks, rules, and feedback keep moving. These views show how the learner side aggregates tasks and how results feed operations.
Implementation
Clients deliver learning UX; core services unify accounts, tasks, dictation, works, and points; external capabilities connect pen SDK, AI review, and speech assessment—supporting three loops on one data model.
Architecture
iOS / Android learning app + PC admin + pen SDK + AI essay review + speech assessment
iOS / Android app (home / dictation / writing / phonics / works / points) · PC admin (content / tasks / feedback / rules / export)
Unified accounts & student profiles · Learning task system (grade / term / unit) · Dictation (lists / strokes / errors) · Works (review / share / picture books) · Points (rules / ledger / reconciliation)
Pen SDK (stroke capture tied to tasks) · AI essay review (structured output) · Speech assessment (phonics analysis & grouped scores)
Errors / works / review / phonics reports · Points ledger & ops export
Keywords
Outcomes
From scattered modules to one operable learning app—these delivery outcomes show how the product runs in production and keeps room to operate after launch.
Reusable pattern
This pattern fits edtech products that need student profiles + learning tasks + AI / hardware + results + incentives—establish the data chain first, then add capabilities.
Unit-based practice, error books, reports, parent views
Structured review, work sharing, admin search and reply
Stroke submit, page validation, tasks linked to errors
Dictation / phonics / recording assessment + points
Student profiles, task assign, work showcase, back-office ops
Talk about your product
Planning a K12 app, AI homework product, smart-hardware learning system, or family-education product? What to decide first is usually not which screen to build—it is who learns, what they learn, how tasks form, where results live, what happens next, and how ops keeps running. We can start from your core scenario, shape an MVP, then set boundaries for accounts, tasks, data, and AI / hardware.
Related cases
Other projects delivered by Wavesteam — AI, IoT, platform builds and enterprise software.
A factory operations platform that unifies system collaboration, production workflow and shop-floor visualization for manufacturing teams.
Read the case study
An occupational health monitoring system with early-warning analytics and control-room dashboards for industrial safety teams.
Read the case study
A real-time AI conversation system for learners of Mandarin Chinese — combines LLM dialogue, smart content generation and edge inference hardware.
Read the case studyThese projects usually involve business systems, device integration, AI workflows or multi-role back-offices. We assess feasibility against real delivery constraints and give recommendations close to the implementation stage.