**Status:** Draft **Mantra:** *Keep your vibe, no interruptions.* ~~*Stop 'catching up' on reading. Start capturing what matters.*~~ ## 1. The Problem Reading is valuable, but logging insights from physical books is filled with friction. Readers must stop their flow to find a pen, open a notes app, or just let the insight fade. Existing "reading" apps focus on social tracking (Goodreads) or heavy-duty digital capture (Readwise), not the _in-the-moment_ experience of reading a physical book. The core problem: **Valuable insights are lost because the cost-of-action in capturing them is too high.** *or* **Why does it take so many clicks to complete a simple task?** And even when insights are captured, they sit in a notes app, unconnected. The second problem: **Captured insights don't compound.** They're just data. There's no synthesis, no reflection, no "what did I actually learn from this book?" ## 2. The Solution (v1 MVP) An Android-first app to capture words and quotes from a physical book as-frictionlessly-as-possible, then use AI to surface patterns and generate reflections from what you've captured. The app is built around the **"Reading Session."** When a session is active, the app's core functions are available from the lock screen via a **Quick Settings Tile** and a foreground notification with action buttons. The user pulls down the notification shade, taps the tile, and captures. The original MediaSession "music player" approach was retired due to platform reliability issues (see [[Technical Decisions#TDR-008]]). This creates a "non-interruptive" flow, letting the user capture an insight and return to their book in seconds. **The AI layer** takes those raw captures and does something with them. Weekly "vibe" reflections that surface thematic patterns. Vocabulary trends. Reading rhythm insights. The captures become inputs to something larger, not just a graveyard of highlights. ## 3. User Stories (v1) - **As a Reader, I want to...** - ...start a "Reading Session" to track my time and enable quick-capture. - ...tap a single button on my lock screen to define a word I'm reading. - ...tap a single button on my lock screen to save a quote by reading it aloud. - ...have the app "read back" the quote it heard so I can confirm its accuracy. - ...have every word and quote I save automatically timestamped and tagged to my current session. - ...end my session and see my total time read automatically logged. - ...browse all words and quotes I've saved, filterable by session/book. - **As a Reader who wants reflection, I want to...** - ...generate a "Weekly Vibe" that tells me what themes I've been exploring. - ...see patterns in my vocabulary (am I reading philosophy? science? fiction with dense prose?). - ...get a sense of my reading rhythm without having to manually journal about it. - ...feel like my captures are *doing* something, not just piling up. ## 4. Key Performance Indicators (KPIs) Our "North Star" is **Activation & Engagement**. We succeed if users who try the app use it consistently. - **North Star KPI:** - **Capture Rate:** (Total Words Defined + Quotes Saved) / Total Active Sessions. This measures if the app's core promise is being used. A high number means we are successfully reducing friction. - **Primary KPIs (v1):** - **W1 Retention:** % of users who start a session in Week 1 who also start a session in Week 2. - **Session Start Rate:** % of users who open the app who successfully start a "Reading Session." - **Lock Screen Capture:** % of all captures that are initiated from the lock screen (validates our core `MediaSession` feature). - **AI Engagement KPIs:** - **Vibe Generation Rate:** % of users with 3+ sessions who generate at least one Weekly Vibe. - **Vibe Re-engagement:** % of users who generate a Vibe and then start another session within 7 days. - **Secondary (Guardrail) Metrics:** - **Quote Accuracy:** % of saved quotes that are _not_ deleted within 24 hours (measures speech-to-text verification quality). - **Session Length:** Average time per active session. ## 5. Phase 1: MVP Timeline Estimate This is a high-level estimate for a small, focused team (e.g., 1-2 engineers). | Phase | Est. Time | Key Tasks | | -------------------------- | ----------- | --------------------------------------------------------------------------------------------------------------------------------------------- | | **Week 1: Foundations** | 1 Week | - Setup Android project. - Build database schema (Room DB). - Basic UI layout (Compose). | | **Week 2-3: Core Service** | 2 Weeks | - Implement `MediaSession` & `foregroundService`. - Build lock screen controls (Define/Quote). - Implement Session logic (start/stop, timer). | | **Week 4: Capture Flows** | 1 Week | - Integrate `SpeechRecognizer` (word/quote capture). - Integrate `TextToSpeech` (quote verification). - Connect 3rd-party Dictionary API. | | **Week 5: UI & Review** | 1 Week | - Build the "Library" screen (browsing quotes/words). - Onboarding & permissions flow (Notification, Mic). - Internal testing & bug fixing. | | **Week 6: AI & Polish** | 1 Week | - Integrate Gemini API for Weekly Vibe generation. - Final polish. - Prep for Play Store internal test track. | | **Total Estimated Time** | **6 Weeks** | | ## 6. The AI Angle The AI isn't a bolt-on. It's the reason capturing matters. Most note-taking apps are write-only. You dump things in and never look at them again. The AI layer closes the loop. It reads what you've captured, finds the threads, and gives you something back. The hypothesis: **If the app can tell you what you've been thinking about, you'll capture more.** The reflection creates the motivation to keep feeding it. Technical approach: - Use Gemini Pro for quality (thematic analysis benefits from reasoning capability) - Keep prompts focused on being a "reading companion," not a productivity coach - Generate weekly, not daily (enough data to find patterns, not so frequent it feels noisy) ## 7. Out of Scope (v2 / Future) To keep v1 focused, we are _not_ building: - Social features (friends, sharing, feeds). - Book search / ISBN scanning (v1 sessions are just text titles). - Advanced reading goals (e.g., "30 mins a day"). - OCR/Camera-based quote capture (more friction than voice). - iOS version (must validate on Android first). - Per-book AI summaries (start with weekly aggregate, expand later). - On-device AI (Gemini Nano) for offline use (explore if API latency is a problem).