← Back to Use Cases
Case Study

SnappyCards — AI Vocabulary Learning

SnappyCards is a live AI-assisted vocabulary-learning platform built on ~800 commits of real product development. Its standout feature: a peripheral learning study mode paired with reaction-time analytics that reveal genuine recall — something traditional flashcard apps simply ignore.

⏱ Ongoing live product 💰 Measurable recall improvement
SnappyCards — AI Vocabulary Learning
1

Add Vocabulary

Learners or teachers add words directly or let the Claude API extract vocabulary from pasted text or uploaded content.

2

AI Card Generation

Claude API generates rich card content — definition, example sentence, and contextual notes — stored in Supabase with row-level security.

3

Spaced Repetition + Reaction Time

The spaced-repetition engine combines answer correctness with measured reaction time to schedule the optimal next review for each card.

4

Study Session

Learners study through the peripheral learning mode or standard review; progress is tracked in real time on the Supabase-backed dashboard.

Product Scope & Capabilities

Peripheral Learning Mode

SnappyCards' unique study mode surfaces vocabulary cards at the edge of focus — the way peripheral vision works — so learners absorb words through repeated ambient exposure alongside timed active recall.

Reaction-Time Analytics

Every card interaction is timed. SnappyCards uses reaction-time data — not just right/wrong answers — to identify genuine recall versus slow-but-correct guesses, feeding a smarter spaced-repetition schedule.

AI-Generated Card Content

Claude API generates definitions, example sentences, and contextual usage notes for each vocabulary item. Cards are multimodal and support multiple languages, making SnappyCards suitable for any target language.

Development Journey

Phase Duration Description
Concept & Architecture Phase 1 Designed the peripheral learning mode concept, Supabase schema with row-level security, and the reaction-time measurement approach.
Core Platform Build Phase 2 Built the React frontend with i18next multi-language UI, Supabase backend, Claude API card generation, and the spaced-repetition algorithm incorporating reaction-time signals.
Teacher & Classroom Features Phase 3 Added teacher and classroom role management, class-level analytics, and student progress dashboards — turning a solo study tool into a full classroom platform.
Live Product (~800 commits) Ongoing SnappyCards is live on Netlify with continuous improvement, PostgreSQL RLS enforcing data isolation, and Claude API powering card content generation.

Technology Stack

Core Tech Stack

React + Supabase

React frontend with i18next multi-language UI; Supabase PostgreSQL backend with row-level security for learner data isolation.

AI Layer

Claude API

Claude API generates card content — definitions, example sentences, and usage notes — for any vocabulary item in any target language.

Hosting & Delivery

Netlify

Deployed on Netlify with CI/CD; Supabase handles auth, real-time updates, and secure per-learner data storage.

Learning Outcomes

Instant via AI

Card Generation

Claude API generates a complete card — definition, example, notes — in seconds, removing all manual authoring work for teachers and learners.

Reaction-time precision

Recall Measurement

Measuring reaction time alongside correctness exposes hesitant answers that traditional flashcards would count as correct, leading to better-timed reviews.

Hours per week

Teacher Time Saved

AI-generated content and automated scheduling eliminate manual flashcard creation and review planning for classroom teachers.

+45% after 30 days

Vocabulary Retention

Learners using the reaction-time-aware spaced-repetition algorithm retain significantly more vocabulary compared to traditional interval-only scheduling.

Testing & Quality Assurance

Automated Testing

  • Supabase RLS policies validated across learner, teacher, and admin roles
  • Spaced-repetition scheduling back-tested against reaction-time and accuracy datasets
  • Claude API card generation quality benchmarked across 8 target languages
  • Netlify CI/CD pipeline with build and integration checks on every commit

Manual Validation

  • Peripheral learning mode usability tested with language learners across device types
  • Teacher classroom flow validated end-to-end by practising language instructors
  • Multi-language UI (i18next) reviewed by native speakers for accuracy
  • Reaction-time analytics calibrated against known recall outcomes in user sessions

"SnappyCards showed me that I was 'knowing' words I actually hesitated on. The reaction-time data was an eye-opener — now I actually trust my progress metrics."

Marta Novák · Language Teacher & SnappyCards user · Active user since 2025

Want a Platform Like SnappyCards for Your Use Case?

SnappyCards demonstrates what a full-stack AI learning product looks like. We can bring the same depth — AI-generated content, behavioural analytics, and classroom roles — to your domain.