AI-native guilt-free flashcard system: book OCR -> DeepSeek deck generation -> practice-date SRS with in-review AI editing. Self-hosted Docker + PWA.

Find a file

Justin Aquino 27d6912949 260818 initial plan: objectives, open questions, milestones, draft FRD		2026-08-18 20:10:04 +08:00
FRD.md	260818 initial plan: objectives, open questions, milestones, draft FRD	2026-08-18 20:10:04 +08:00
PLAN.md	260818 initial plan: objectives, open questions, milestones, draft FRD	2026-08-18 20:10:04 +08:00
README.md	260818 initial plan: objectives, open questions, milestones, draft FRD	2026-08-18 20:10:04 +08:00

Adaptive-Flashcards

An AI-native, guilt-free flashcard system. Self-hosted (Docker + database), multi-user, PWA client, powered by the user's own DeepSeek API credits.

The Problem It Solves

Fixes the two biggest pain points of Anki:

Review Hell — the guilt and anxiety of a massive accumulated backlog when you miss a few days.
Editing Friction — having to stop a mobile review session and sit at a desktop to fix a bad card.

Upload a book → OCR → text. Source material is user-provided, so AI generation stays grounded (no hallucinated content).
Prompt-driven deck creation. Pick a chapter/section, run a prompt, get a flashcard deck.
Guilt-free SRS. Tracks dates practiced instead of punishing chronological due dates; no card accumulation, sustainable daily pacing, optional deep-dive mode.
In-review AI editing. "Rewrite this to make it stickier", split overloaded cards into atomic concepts, add/delete/modify — all mid-review, no desktop required.
True ownership. Export decks, run on your own server, share decks with others.

File	Contents
`PLAN.md`	Objectives, goals to clarify, milestones (worked backwards), prerequisite tasks, next steps
`FRD.md`	Draft Functional Requirements Document

Both public. Keep them in sync; gi7bfj is the default push target per workspace convention.