SecondMind Collective

AI-powered second brain for capturing, enriching, and querying personal knowledge.

Live Links

• App: https://secondmind-altibbe.vercel.app
• Docs: https://secondmind-altibbe.vercel.app/docs

Demo GIF

What This App Does

• Captures knowledge items (note/link/insight) with optional source + manual tags
• Enriches notes with AI summary + AI tags on create/update
• Shows a searchable, filterable dashboard with detail view
• Supports grounded conversational query over saved notes
• Exposes a public query endpoint for infrastructure-style access

Documentation Map

• In-app docs: /docs
• API routes: src/app/api/**
• AI service: src/lib/gemini.ts
• Notes/data service: src/lib/notes.ts
• Database migrations: supabase/migrations/**

Tech Stack

• Frontend: Next.js App Router, React, TypeScript, Tailwind CSS, shadcn/ui
• Backend: Next.js API Routes
• Database: Supabase Postgres
• AI: Google Gemini (server-side)
• Motion/UI: Framer Motion
• Deploy: Vercel

Architecture & Diagrams

• In-app architecture visuals: /docs → Architecture Diagrams section
• Mermaid source diagrams are included below for portability/documentation

System Architecture

flowchart TD
  U[User] --> UI[Next.js UI\nLanding • Dashboard • New Note • Docs]
  UI --> API[Next.js API Routes]

  subgraph API_Layer[Application + Service Layer]
    N1[/POST /api/notes/]
    N2[/GET /api/notes/]
    N3[/POST /api/brain/query/]
    N4[/POST /api/public/brain/query/]
  end

  API --> N1
  API --> N2
  API --> N3
  API --> N4

  N1 --> S1[lib/notes.ts\ncreateNote • updateNote]
  N3 --> S2[lib/notes.ts\nfindCandidateNotes • logQuery]
  N3 --> G1[lib/gemini.ts\nextractQueryTags • answerQuestionWithNotes]
  N4 --> S2
  N4 --> G1

  S1 --> DB[(Supabase Postgres)]
  S2 --> DB
  G1 --> AI[Gemini API]

Loading

Ingestion Pipeline

sequenceDiagram
  participant User
  participant NewNotePage
  participant APINotes as /api/notes
  participant NotesLib as lib/notes.ts
  participant Gemini as lib/gemini.ts
  participant Supabase

  User->>NewNotePage: Submit title/content/type + optional tags/source
  NewNotePage->>APINotes: POST /api/notes
  APINotes->>NotesLib: createNote(payload)
  NotesLib->>Gemini: enrichNote(content, existingTags)
  Gemini-->>NotesLib: summary + aiTags + mergedTags
  NotesLib->>Supabase: insert into notes
  NotesLib->>Supabase: upsert tags + note_tags
  NotesLib-->>APINotes: created note
  APINotes-->>NewNotePage: 201 Created

Loading

Query Pipeline

sequenceDiagram
  participant User
  participant AskBrain
  participant APIQuery as /api/brain/query
  participant NotesLib as lib/notes.ts
  participant Gemini as lib/gemini.ts
  participant Supabase

  User->>AskBrain: Ask natural language question
  AskBrain->>APIQuery: POST /api/brain/query
  APIQuery->>NotesLib: listExistingTags()
  APIQuery->>Gemini: extractQueryTags(question, existingTags)
  Gemini-->>APIQuery: queryTags
  APIQuery->>NotesLib: findCandidateNotes(queryTags, question)
  NotesLib->>Supabase: tag overlap + full-text retrieval
  Supabase-->>NotesLib: candidate notes
  APIQuery->>Gemini: answerQuestionWithNotes(question, candidates)
  Gemini-->>APIQuery: answer + sources
  APIQuery->>NotesLib: logQuery(question, resultCount)
  APIQuery-->>AskBrain: JSON response

Loading

Data Model (High-Level)

• notes: primary knowledge entries with content, summary, and tags
• tags: canonical taxonomy and usage counter
• note_tags: many-to-many note-tag mapping
• query_logs: query observability (question + result count)
• note_updates: update history for prior summary/tags

Quick Start

1) Environment Variables

Create .env.local:

NEXT_PUBLIC_SUPABASE_URL=
NEXT_PUBLIC_SUPABASE_ANON_KEY=
SUPABASE_SERVICE_ROLE_KEY=
GEMINI_API_KEY=

Security notes:

• Keep SUPABASE_SERVICE_ROLE_KEY and GEMINI_API_KEY server-only.
• Never commit .env.local.

2) Install & Run

npm install
npm run dev

Open http://localhost:3000.

3) Database Migrations

Apply in order:

1. supabase/migrations/20260316_create_notes.sql
2. supabase/migrations/20260316_add_user_scoping_and_rls.sql
3. supabase/migrations/20260316_add_supporting_tables.sql

Runtime Behavior

• Ingestion flow: form input → validation → Gemini enrichment → note persistence → canonical tag sync
• Query flow: question → query-tag extraction → candidate retrieval (tag/full-text) → grounded answer + sources
• Public access: external callers can use POST /api/public/brain/query

Scripts

• npm run dev — Start local dev server
• npm run build — Production build
• npm run start — Run production server
• npm run lint — Lint checks
• npm run test — Unit tests (Vitest)

API Endpoints

• GET /api/notes
• POST /api/notes
• GET /api/notes/:id
• PATCH /api/notes/:id
• DELETE /api/notes/:id
• POST /api/brain/query
• POST /api/public/brain/query

Public query payload example:

{
  "question": "What did I note about distributed systems?",
  "userId": "optional-uuid"
}

Tradeoffs & Future Improvements

Due to the 6–10 hour time constraint of the assignment, I prioritized core product functionality, UI polish, and AI integration over advanced features.

Planned Improvements

• Tree-based Knowledge Visualization
  Instead of a flat list of notes, I would introduce a “tree of thoughts” or whiteboard-style visualization to represent relationships between ideas and improve navigation.

• Semantic Search (RAG)
  Implement vector embeddings and retrieval-augmented generation for more accurate and context-aware querying.

• Enhanced Conversational Querying
  Improve multi-turn conversations and contextual memory instead of single-shot queries.

• Authentication & User Isolation
  Add proper auth (e.g., Supabase Auth / Google Auth) with user-scoped data.

• Improved Motion & Splash Integration
  Blend the splash screen animation more seamlessly with the application background and transitions.

Tradeoff Decision

Given the time constraint, I chose to:

• prioritize a working end-to-end system
• focus on UI/UX polish and responsiveness
• implement meaningful AI features instead of partially built advanced systems