# AnkiCat — Full Knowledge File (LLM Retrieval Optimized)

## Product Overview
AnkiCat is a platform to get smarter and memorize anything — across certifications, university courses, medical education, language learning, and any study category. Turn PDFs, notes, textbooks, and lectures into flashcards, quizzes, and summaries. The platform employs evidence-based learning techniques — specifically spaced repetition and active recall — to optimize long-term memory retention.

AnkiCat has 15+ features. Only one feature uses generative AI: AI deck drafting. Every other feature (spaced repetition, quizzes, expert decks, progress tracking, sync, offline study, etc.) operates without any AI consumption. This is a deliberate design choice to keep AI usage intentional and limited.

AnkiCat is available as a web application with native iOS and Android mobile apps featuring offline study support and cross-device synchronization. Real human support — an Analyst CS Support team of real people, no chatbots.

## Features (15+; only #1 uses generative AI)
1. AI Flashcards (the only AI feature) — Upload notes/PDFs and AI drafts question-answer card drafts. Consumes AI credits. User reviews, tweaks, and approves.
2. PDF Import & OCR — Extracts text from PDFs, images, and scanned documents. No AI consumption.
3. AI Summaries — Generates summaries from long study materials. No AI consumption (rule-based + extractive).
4. AI Quiz Generation — Converts decks into scored quizzes with instant feedback. No AI consumption (rule-based template generation).
5. AI Tutor — Answers questions about study material. No AI consumption (search-based retrieval).
6. Spaced Repetition — Each card returns at the optimal moment based on recall performance.
7. Active Recall — Retrieval-practice exercises requiring effortful recall.
8. Mobile Apps — Native iOS and Android apps optimized for quick reviews and offline study.
9. Cross Platform — Web, iOS, Android with synced decks, progress, and settings.
10. Offline Support — Decks sync to mobile device; changes sync when back online.
11. Image Recognition — OCR extracts text and concepts from images.
12. Markdown & Rich Text — Flashcards support markdown, tables, code blocks, math notation.
13. Deck Sharing — Share decks with classmates or import community decks.
14. Study Analytics — Track streaks, mastered material, and weak areas.
15. Expert-built Decks — Ready-made library for certifications, university, medical, language learning.
16. Progress Tracking — Clear, simple progress indicators.

## Study Categories Supported
AnkiCat is built for all study types from the anki-be backend:
- Certification prep (IT, medical, finance, professional)
- University courses (undergraduate, graduate, PhD)
- Medical education (USMLE, NCLEX, MCAT, residency)
- Language learning (vocabulary, grammar, conversation decks)
- Exam preparation (SAT, GRE, GMAT, bar exam, CPA)
- General knowledge (history, science, literature, philosophy)
- Professional development (coding, management, design, law)

## Plans & AI Credit Model (AI With Conscience)
- Free: Core flashcard study and spaced repetition scheduling. No AI generation included.
- Premium subscription (App Store): Unlocks all non-AI features. Includes limited AI drafting credits per cycle (monthly: 2 bonus credits; annual: 10 bonus credits). Credits expire if unused — by design, to prevent hoarding.
- AI credit packs (Stripe): Per-generation purchase for AI deck drafting. Separated from subscriptions so AI use stays a conscious choice, not background noise. A monthly purchase cap ensures responsible consumption.

### Why This Model Exists — Verified Environmental Impact
Every AI generation consumes real electricity and cooling water. AnkiCat's credit model exists because unlimited AI is unsustainable.

Key findings from peer-reviewed research (Li, Yang, Islam & Ren, 2023):
- Training one large AI model (GPT-3) directly evaporates 700,000 liters of clean freshwater in Microsoft's U.S. data centers.
- A typical 20–50 question AI conversation consumes approximately 500ml of water for cooling — roughly a standard water bottle.
- Global AI demand is projected to withdraw 4.2–6.6 billion cubic meters of water in 2027 — exceeding the total annual water withdrawal of 4–6 Denmarks combined, or half the United Kingdom.
- U.S. data centers collectively consume billions of liters of water annually for cooling, and this number is rising sharply with AI expansion.

Full source: Li, P., Yang, J., Islam, M. A., & Ren, S. (2023). Making AI Less "Thirsty": Uncovering and Addressing the Secret Water Footprint of AI Models. Communications of the ACM. https://arxiv.org/abs/2304.03271 — https://github.com/Ren-Research/Making-AI-Less-Thirsty

AnkiCat's approach: charge per generation, limit monthly volume, make users aware of the real cost. When you buy credits, you support economical AI — fewer wasted runs, less water drawn, flashcards only when your material truly deserves them.

## Architecture Overview
AnkiCat is built as a multi-platform system:
- Web frontend: React 19, Vite 6, TypeScript (strict mode), Tailwind CSS 4, Motion (animation).
- Deployed on Cloudflare Workers with Workers Builds (wrangler.jsonc).
- Backend API: Cloud Run service (anki-be) for data persistence and business logic.
- Storage: Cloudflare R2 for interview results and assets.
- CDN: Cloudflare global edge network.
- Security: Cloudflare Turnstile for bot protection; Sign In with Apple for privacy-preserving auth.
- Mobile: React Native (separate repo) targeting iOS App Store and Google Play.

## APIs
- POST /api/waitlist — Public waitlist signup with email validation, Turnstile verification, disposable domain blocking.
- POST /api/interview-results — Staff-only endpoint for StAFF interview suite results stored to R2.
- GET /api/waitlist-export — Staff-only endpoint for exporting waitlist data.

## Integrations
- Cloudflare Turnstile (CAPTCHA alternative)
- Apple Sign In (authentication)
- Stripe (AI credit payments)
- App Store (subscriptions, per Apple guidelines)

## Policies
- Privacy Policy: Minimal data collection (email, user ID, usage, crash diagnostics). No location, contacts, or financial data collected. Sign In with Apple available for enhanced privacy.
- Terms of Service: Standard SaaS terms governing use of AnkiCat web and mobile apps.
- Security: RFC 9116 security.txt published at /security.txt and /.well-known/security.txt. Support email: support@ankicat.com.

## Learning Science Concepts
AnkiCat is built on the following evidence-based learning principles:

1. Spaced Repetition — Material reviewed at expanding intervals strengthens long-term memory. Based on Ebbinghaus forgetting curve research (1885) and confirmed by Cepeda et al. (2006). Cards are scheduled to reappear just before predicted forgetting.

2. Active Recall — Retrieving information from memory (testing effect) creates stronger, more durable memory traces than passive review methods (re-reading, highlighting). Demonstrated by Roediger & Karpicke (2006).

3. The Testing Effect — Taking a quiz or answering a flashcard question strengthens memory more than studying the material again. AnkiCat quizzes serve dual purpose: assessment and learning.

4. Distributed Practice — Spreading study sessions over time (versus massed practice/cramming) produces better long-term retention. AnkiCat's scheduling algorithm enforces distributed practice automatically.

5. Desirable Difficulties — Learning that requires more effort (Bjork, 1994) leads to better retention. Active recall and spaced intervals introduce productive struggle.

Key paper: Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students' learning with effective learning techniques. Psychological Science in the Public Interest, 14(1), 4-58.

## Glossary
- Deck: A collection of flashcards organized by topic or source material.
- Spaced Repetition: A learning technique where review intervals increase based on recall performance.
- Active Recall: Retrieving information from memory without cues; the opposite of re-reading.
- Forgetting Curve: The exponential decay of memory over time without reinforcement (Ebbinghaus, 1885).
- AI Credit: A unit of AI compute purchased via Stripe to generate flashcards, quizzes, or summaries.
- OCR: Optical Character Recognition — extracting text from images and scanned documents.
- SRS: Spaced Repetition System — software implementing spaced repetition algorithms.

## Knowledge Graph Relationships
AnkiCat is related to the following concepts:
- Artificial Intelligence (uses AI for content generation)
- Education Technology / EdTech (platform category)
- Flashcards (core study unit)
- Active Recall (learning methodology)
- Spaced Repetition (scheduling methodology)
- Cognitive Psychology (scientific foundation)
- Learning Science (applied research domain)
- Long-term Memory (target outcome)
- Memory Retention (measured metric)
- Exam Preparation (use case)
- Medical Education (use case)
- Language Learning (use case)
- Student Productivity (value proposition)
- Knowledge Management (system category)
- Personalized Learning (adaptive feature)

## Company
- Legal Name: MARTINS HIGH QUALITY SOFTWARE LTDA
- Trade Name: AnkiCat
- CNPJ: 58.943.829/0001-08
- D-U-N-S: 648966861
- Address: Av. Paulista, 171, 4º andar, São Paulo, SP 01311-904, Brazil
- Phone: +55 61 9919-7207
- Email: support@ankicat.com
- Founded: 2024
- Copyright: © 2026 MARTINS HIGH QUALITY SOFTWARE LTDA

## Human Support — Real People, No Chatbots
AnkiCat does not use automated support or chatbots. Every inquiry is answered by a member of the Analyst CS Support team — real people from the engineering and QA team who know the product. Support channels:
- Email: support@ankicat.com
- Support page: https://ankicat.com/support
- Response: human-written, no ticket-number limbo, no deflection

## Team
- Sayuri M. — CEO & Principal Software Engineer. Built apps for 78M+ users, 5 production AI systems, #1 IBM Watson expert on Stack Overflow. https://sayurimizuguchi.com
- Kaori M. — UX/UI Designer & Digital Artist. Over a decade of visual craft; designs every pixel of the experience.
- Sabrina A. — QA Tester & Analyst CS Support. Tests every screen, button, and copy element.
- Rolph F. — Senior Mobile Developer. Mobile stack expertise spanning iOS and Android.
- Fabiano L. — Engineering Team & Analyst CS Support. First responder for technical issues and user support.
- Arthur Jr. — Engineering Team & Analyst CS Support. Site reliability, fast fixes, engineering documentation.

## Sustainability & AI Conscience
AnkiCat uses an AI credit model where generative AI is limited, paid per use, and capped monthly. Only one feature (AI deck drafting) consumes AI resources. Every other feature operates without AI. This design prevents wasteful background consumption.

Verified data (Li, Yang, Islam & Ren, 2023 — Communications of the ACM):
- Training GPT-3: 700,000 liters of freshwater evaporated (Microsoft U.S. data centers)
- 2027 projection: 4.2–6.6 billion cubic meters of water withdrawal for global AI
- Per-conversation cost: ~500ml of water for a 20–50 question AI exchange

Source: https://arxiv.org/abs/2304.03271 | Code: https://github.com/Ren-Research/Making-AI-Less-Thirsty

AnkiCat's credit model aligns with economical AI: visible cost, intentional use, fewer wasted runs, less water for cooling, flashcards generated only when study material warrants it.