Snap Words

Vocabulary apps often force one rigid study path or bury you in configuration. Snap Words starts from a simple idea: flash cards should be flexible - deck structure, card faces, and review order should adapt to how you learn, not the other way around.

AI fills in what static dictionaries miss: natural examples, short explanations, and context so a word sticks beyond a single translation line. The goal is a calm, repeatable habit - open a deck, snap through cards, and leave with clearer recall than another abandoned word list.

Client - A focused mobile-first experience around decks and sessions; flash cards are the primary surface, with room to add spaced repetition and alternate drills without rewriting the core model.

Stack - React Native for the mobile client (mobile-first), with a Next.js web app planned on the same core model. A lightweight API backed by MongoDB via Mongoose keeps deck/card schemas flexible as the learning model evolves.

AI - Prompted generation of examples and explanations per word, with guardrails so output stays concise and appropriate for study. The pipeline is built so providers can evolve as models and pricing change.

Translations - DeepL is used for reliable, high-quality translation between your working languages so you spend time learning, not copy-pasting into separate tools.

Principles - flexible card layouts first, minimal friction between 'add word' and 'review', and transparent handling of third-party services (translation and AI) in product copy and settings.

Core -

• Create and edit decks; add words with source language, target language, and optional notes.
• Flash card session with configurable fronts and backs (term, translation, example, explanation).
• AI-generated example sentences and short explanations per word, user-triggered or on save.

Integrations -

• DeepL for translation payloads the learner expects in a language app.

Quality -

• Graceful offline or failure states when AI or translation APIs are unavailable; never lose in-progress edits.

Non-goals (for now) - full classroom product, social feeds, or a generic chat assistant unrelated to vocabulary.

Future instrumentation - deck creation, session length, cards reviewed per session, and AI generation success versus skip rate - so improvements target study quality, not vanity counts.

Privacy posture - minimize retained content from AI calls; prefer aggregated product metrics over raw prompts in analytics backends.