NextJS OpenAI Integration Monorepo (WIP)

Status: 🏗️ Work in Progress

This is a Turborepo-based monorepo setup for building AI-enhanced mood analysis applications using modern web technologies. This monorepo contains two Next.js applications — platform and public — and shared packages managed with Turborepo. The project integrates tools like OpenAI, Prisma, and Contentful to deliver a rich and interactive user experience.

📦 Apps & Packages Structure

apps/
  ├── platform/   # Authenticated, AI-enhanced journaling platform
  └── public/     # Public-facing PWA with content from Contentful

packages/
  ├── global-analytics/  # Analytics utilities
  ├── global-ui/         # Shared UI components
  ├── global-utils/      # Common utilities
  ├── eslint-config/     # Shared ESLint configurations
  └── typescript-config/ # Shared TypeScript configurations

✨ Features

• AI-powered mood analysis from journal entries using Vercel AI SDK with structured outputs
• Semantic Q&A over journal entries using LangChain's RAG capabilities
• Detailed mood statistics displayed in interactive charts
• Integration with Prisma for database management
• Integration with Contentful Headless CMS
• Progressive Web App (PWA) support
• Monorepo structure with shared UI components
• Error tracking and monitoring with Sentry
• Internationalization (i18n) support

🧰 Technologies

• Next.js (App Router, Turbopack)
• Turborepo — high-performance monorepo management
• TypeScript — strict typing
• Tailwind CSS — modern utility-first CSS
• Prisma — ORM and DB schema
• Vercel AI SDK — structured AI generation with type-safe outputs
• LangChain — RAG (Retrieval Augmented Generation) and vector embeddings
• Contentful — headless CMS
• Next-PWA — for offline-first capabilities
• Vitest & Cypress — unit and E2E testing
• Sentry — error tracking and monitoring
• i18n — internationalization

📦 Scripts (root)

yarn dev         # Run all dev servers
yarn build       # Build all apps/packages
yarn lint        # Lint all apps/packages
yarn test        # Run all tests

Each app has its own set of scripts inside apps/platform and apps/public.

🚀 Getting Started

git clone https://github.com/yourusername/nextjs-openai-integration-demo.git
cd nextjs-openai-integration-demo
yarn install
yarn dev

🤖 AI Architecture

This project uses a hybrid AI approach:

• Vercel AI SDK (generateObject) — For structured journal entry analysis with Zod schemas
  • Mood detection
  • Sentiment scoring
  • Summary generation
  • Color representation
• LangChain — For advanced RAG operations
  • Vector embeddings (OpenAIEmbeddings)
  • Semantic search over journal entries
  • Q&A refinement chains
  • In-memory vector store

This separation provides the best of both worlds: simple, type-safe structured generation with Vercel AI SDK, and powerful semantic search with LangChain.

📁 Environment Variables

Environment variables are managed via .env files and/or Vercel dashboard. Refer to turbo.json for globalEnv configuration used during builds:

• Contentful CMS configuration
• OpenAI API key (OPENAI_API_KEY)
• Database connection
• Cron job secrets

🧑‍💻 Author

Alex Suprun
Email: alex.suprun.email@gmail.com

📄 License

This project is licensed under the MIT License. See the LICENSE file for details.