NewMemo - AI-Powered Second Brain Journal

Turn daily reflections into a searchable second brain using on-device OCR, voice input, and conversational RAG.

Comment

Inspiration

Journaling is one of the most powerful habits for mental clarity and self-reflection, yet over 80% of people abandon it within weeks. Through conversations with entrepreneurs and fellow creators at startup hubs, we identified two fundamental barriers:

  1. High friction of input: Typing thoughts on a mobile keyboard every evening feels like a chore.
  2. Low perceived return: Entries pile up as dead text, rarely revisited or leveraged.

We built NewMemo to solve both problems: making capturing effortless via multimodal AI (Voice & OCR), and transforming past entries into an active "Second Brain" using Retrieval-Augmented Generation (RAG).

What It Does

NewMemo bridges the warmth of physical stationery with cutting-edge AI:

  • Zero-Friction Capture: Dictate thoughts via real-time speech-to-text or snap photos of handwritten notebooks and physical receipts (powered by intelligent OCR).
  • Conversational Second Brain (Synapse RAG): Users can chat with their past self (e.g., "When did I last feel overwhelmed, and how did I overcome it?"). The AI cites specific historical diary cards as verifiable evidence.
  • Emotion & Growth Analytics: Automated sentiment tracking, monthly retrospective reports, and personality insights (MBTI/custom analysis) help users understand their emotional trajectories.
  • Paper Trail UX: Designed with unbleached paper tones (paperCream), sepia typography, and an empathetic companion ("Neu"), creating a calming sanctuary away from hyper-stimulating feeds.

How We Built It

NewMemo was developed solo using a modern, scalable full-stack architecture:

  • Mobile Frontend: Flutter (Dart) for smooth 60fps animations and cross-platform flexibility.
  • Backend & Database: Supabase (PostgreSQL) with pgvector storing 1536-dimensional embeddings for hybrid semantic search.
  • AI & LLM Integration: Gemini API for fast, multimodal analysis and RAG responses.
  • Monetization & In-App Purchases: RevenueCat SDK powering seamless subscription tiers, paywalls, and customer management across platforms.

Monetization & RevenueCat Integration

NewMemo follows a sustainable freemium model:

  • Free Tier: Unlimited text journaling, basic mood tracking, and local storage.
  • NewMemo Pro ($1.99/mo or $19.99/yr):
    • Unlimited AI Voice transcription & OCR parsing.
    • Unlimited Conversational RAG queries with source citation.
    • Deep monthly analytics and custom AI persona configuration.

We leveraged RevenueCat Paywalls and Offerings to rapidly test subscription packages, deliver a frictionless onboarding checkout experience, and handle cross-platform entitlement verification out-of-the-box.

Challenges We Ran Into

  • RAG Latency & Relevance: Querying hundreds of historical entries while keeping response times under 2 seconds required optimizing vector index chunking and hybrid filtering (vector distance + date metadata).
  • Balancing Tech & Emotion: AI apps often feel cold and robotic. We iterated heavily on micro-animations and warm UI shaders to ensure the app feels like opening a cherished leather notebook.

Accomplishments We're Proud Of

  • Designing, building, and deploying a production-grade full-stack Flutter app as a solo creator.
  • Implementing an end-to-end RAG pipeline on Supabase pgvector with real-time multimodal inputs.
  • Launching on the App Store with RevenueCat fully integrated in record time.

What We Learned

Shipping is just the beginning. Through early user feedback, we learned that people don't just want raw AI answers; they want their original words validated. Designing AI as an empathetic facilitator—rather than an intrusive assistant—is the key to long-term habit formation.

What's Next

  • From Passive Log to Proactive Companion: Rather than waiting for users to write, NewMemo will analyze subtle emotional shifts over weeks and gently prompt thoughtful check-ins (e.g., "A year ago, you overcame a similar creative block—would you like to revisit what helped then?").
  • Life Decision Co-pilot: When facing major crossroads (career changes, difficult choices), users can synthesize months of reflections to surface their deeply rooted values and blind spots through targeted RAG queries.
  • Local-First Privacy Architecture: To protect deeply intimate thoughts, we will introduce on-device embeddings and small language models for sensitive processing, reserving cloud calls strictly for heavy synthesis.
  • Data-Driven LTV Optimization with RevenueCat: Implement RevenueCat Experiments and dynamic Paywalls to tailor offerings based on user journaling streaks, ensuring a sustainable, long-term business model without compromising user trust.

Built With

  • dart
  • flutter
  • gemini-api
  • in-app-purchases
  • mobile-app-development
  • openai-api
  • pgvector
  • postgresql
  • rag
  • revenuecat
  • supabase
Share this project:

Updates

Submission history