Reattend - AI Memory Operating System

An AI-native "shared memory + decision OS" that compounds into a living memory graph over time.

Quick Start

Prerequisites

• Node.js 18+
• npm

Local Development

# Install dependencies
npm install

# Initialize database + seed demo data
npm run db:seed

# Start dev server
npm run dev

Open http://localhost:3000

Demo login: demo@reattend.com / demo1234

Production Build

npm run build
npm start

Environment Variables

Create a .env.local file:

# Auth (required)
NEXTAUTH_SECRET=your-secret-here
NEXTAUTH_URL=http://localhost:3000

# Database
DATABASE_URL=./data/reattend.db

# AI Provider (mock | openai | anthropic)
AI_PROVIDER=mock
AI_API_KEY=
AI_MODEL=

# Embedding Provider
EMBEDDING_PROVIDER=mock
EMBEDDING_API_KEY=
EMBEDDING_MODEL=

# App
APP_URL=http://localhost:3000

Tech Stack

Layer	Technology
Framework	Next.js 14 (App Router)
Language	TypeScript
Database	SQLite via better-sqlite3
ORM	Drizzle ORM
Auth	NextAuth.js v5
UI	Tailwind CSS + shadcn/ui + Radix
Graph/Board	React Flow (@xyflow/react)
Animations	Framer Motion
State	Zustand
AI	Provider-agnostic (mock mode included)

Architecture

src/
  app/                     # Next.js App Router
    (auth)/                # Login, Register
    (app)/app/             # Authenticated app routes
    api/                   # API routes
  components/
    app/                   # App shell (sidebar, topbar)
    ui/                    # shadcn/ui components
  lib/
    db/                    # Schema, migrations, seed
    ai/                    # LLM interface, prompts, agents
    jobs/                  # Background job queue
    auth/                  # Auth configuration
  stores/                  # Zustand stores
  types/                   # TypeScript definitions

AI Agent Pipeline

Ingest Contract

POST /api/ingest
Content-Type: application/json

{
  "text": "string (required)",
  "source": "string (default: manual)",
  "workspace_id": "string (required)",
  "occurred_at": "ISO 8601 datetime",
  "author": { "name": "string", "email": "string" },
  "metadata": { ... }
}

Pipeline Flow

1. Ingest → Raw item created in inbox
2. Triage Agent → Analyzes text, decides store/ignore, extracts entities/tags
3. Embedding Job → Generates vector embedding for semantic search
4. Linking Agent → Finds similar records, creates graph edges
5. Project Agent → Suggests project groupings

Mock Mode

Set AI_PROVIDER=mock for local dev. The mock provider:

• Returns deterministic triage results based on keyword analysis
• Generates consistent 384-dim embeddings from text hashing
• Creates similarity-based links

Plugging in a Real Provider

Edit src/lib/ai/llm.ts and implement the LLMProvider interface:

interface LLMProvider {
  generateJSON<T>(prompt: string, schema: ZodType<T>): Promise<T>
  generateText(prompt: string): Promise<string>
  embed(text: string): Promise<number[]>
}

Pages

Route	Description
/	Landing page
/login	Authentication
/register	Account creation (30-day trial)
/app	Dashboard with daily summary
/app/inbox	Raw captures + AI triage
/app/memories	Curated records list + detail
/app/projects	Project management + AI suggestions
/app/graph	Interactive memory graph
/app/boards	Whiteboard canvas (Miro-like)
/app/search	Semantic + text search
/app/integrations	100 integration directory
/app/settings	Profile, workspace, members, agent logs
/app/billing	Plan selection, billing stubs

Database

SQLite database stored at data/reattend.db. Tables:

• users, workspaces, workspace_members, workspace_invites
• sources, raw_items (inbox)
• records (curated memories), entities, record_entities
• record_links (graph edges), embeddings
• projects, project_records, project_suggestions
• boards, board_nodes, board_edges
• comments, activity_log
• plans, subscriptions
• integrations_catalog (100 items), integrations_connections
• job_queue, inbox_notifications

Reset Database

rm data/reattend.db
npm run db:seed

Features

• Multi-tenant workspaces (personal + team)
• Dark mode + light mode
• AI triage with entity extraction
• Semantic search via embeddings
• Interactive memory graph with filters
• Whiteboard/canvas with sticky notes
• 100 integration stubs
• Billing stubs with 30-day trial
• Invite team members by email
• Agent activity logs
• Ask AI (conversational Q&A over memories)
• Notification inbox