Product Requirements Document — StudyFlashcards v1

TL;DR

A web app where students create decks, add flashcards, and study with spaced repetition. After each card, the user rates recall (again, hard, good, easy) and the app schedules the next review using a simplified SM-2 algorithm. Target: 40 WAU within 60 days. Differentiator from Anki: web-first, clean UI, no sync setup; differentiator from Quizlet: real spaced repetition (not Leitner).

Goals (business):

• 40 WAU within 60 days
• 60% of weekly users complete 3+ study sessions per week
• Self-funded; no paid tier in v1

Goals (user):

• Create a deck of 20 cards in under 10 minutes
• Daily 15-minute study session that surfaces exactly the right cards
• See retention curve per deck after 2 weeks of use

Personas:

1. Priya, medical student: memorizing pharmacology; tried Anki but found the UI off-putting
2. Kenji, self-learner: learning Japanese vocabulary; uses Quizlet but wants better scheduling

Functional requirements (P0):

1. P0-1. Auth: email + magic link. Acceptance: signup-to-first-card under 3 minutes.
2. P0-2. Create / edit / delete decks. Each deck has a name and an optional description. Acceptance: a user can name a deck, save it, and add cards in under 30 seconds.
3. P0-3. Create cards with front + back (plain text, no images in v1). Acceptance: 20-card deck creation in under 10 minutes.
4. P0-4. Study session that surfaces today's due cards (any card with next_review_at <= now). Acceptance: session loads cards in correct order (most-overdue first).
5. P0-5. Simplified SM-2 scheduling: after each card, user rates recall (4 options); the algorithm updates interval_days, ease_factor, and next_review_at. Acceptance: scheduling math matches SM-2 reference behavior for the 4 rating paths.
6. P0-6. Daily streak: count consecutive days with at least one completed session. Acceptance: streak handles timezone correctly.

Functional requirements (P1):

1. P1-1. Retention chart per deck (% recalled correctly over time).
2. P1-2. Cram mode: study a deck regardless of due dates.

UX flow: 4 screens.

1. Sign-in
2. Deck list (name, card count, due-today count)
3. Deck editor (add / edit / delete cards)
4. Study session (one card at a time, keyboard shortcuts: space to flip, 1-4 to rate)

Success metrics:

• 60% of new users complete a study session on day 1
• 40% of week-2 users return for week-3
• Streak math has zero off-by-one errors across 100 test sessions
• SM-2 algorithm output matches reference values within 1 day on a 100-card test

Technical considerations:

• Stack: React + Supabase (Lovable defaults) or Next.js + Drizzle + Neon if going the Cursor route
• Deploy: Vercel
• SM-2 algorithm: implement the simplified variant inline; ~20 lines of TypeScript
• Timezone: store user IANA tz; calculate "today" server-side

Lovable appendix — Supabase schema:

• profiles (id uuid → users.id, timezone text)
• decks (id uuid, user_id uuid, name text, description text)
• cards (id uuid, deck_id uuid, user_id uuid, front text, back text, created_at timestamptz)
• reviews (id uuid, card_id uuid, user_id uuid, rating int, interval_days numeric, ease_factor numeric, reviewed_at timestamptz, next_review_at timestamptz)
• card_state view: latest review per card with next_review_at for the "today's due cards" query
• RLS: user_id = auth.uid() everywhere

SM-2 simplified algorithm (TypeScript signature):

function nextReview(rating: 1|2|3|4, prev: { interval: number; ease: number }):
  { interval: number; ease: number; nextReviewAt: Date }

Behavior:

• rating 1 (again): reset interval to 1 day, decrease ease by 0.2 (min 1.3)
• rating 2 (hard): interval × 1.2, decrease ease by 0.15
• rating 3 (good): interval × ease
• rating 4 (easy): interval × ease × 1.3, increase ease by 0.15

Out of scope (v1):

• Shared decks or community decks
• Image / audio cards (text-only)
• Mobile native app (web is PWA-installable)
• Cloze deletion or LaTeX rendering
• Deck import from Anki .apkg files