AI Stack

The Production-Ready AI Application Framework

Build AI-powered applications with zero vendor lock-in.

Get Started · Documentation · Stack Builder · GitHub

Why AI Stack?

Building production AI applications requires more than just connecting to an LLM. You need authentication, vector databases, RAG pipelines, and infrastructure that scales. AI Stack gives you all of this out of the box.

Challenge	AI Stack Solution
Vendor Lock-in	Swap LLMs or Vector DBs with one environment variable
Production Readiness	Built-in auth, error handling, observability, and cost tracking
Type Safety	Full TypeScript + Pydantic across the entire stack
Infrastructure	Docker Compose dev, Kubernetes prod - ready to deploy

Quick Start

npx create-ai-stack-starter@latest my-ai-app

You'll be prompted to select:

Architecture: Next.js Fullstack, FastAPI + Next.js, or TypeScript Backend
Database: Neon, Supabase, Turso, or SQLite
Auth: Better Auth, NextAuth, Clerk, or None
LLM Provider: OpenAI, Anthropic, or Novita AI
Add-ons: Tailwind, Biome, PWA, Analytics

Architecture Options

🌐 Next.js Fullstack

Best for: Rapid prototyping, Vercel deployment

Vercel AI SDK
Drizzle ORM
Better Auth
Server Actions

🐍 FastAPI + Next.js

Best for: ML teams, complex AI pipelines

Async Python backend
SQLAlchemy 2.0
Celery for background jobs
Alembic migrations

📦 TypeScript Backend

Best for: TypeScript teams, edge deployment

Hono / NestJS / Fastify
Drizzle or Prisma
Edge-ready
tRPC ready

Core Features

🔓 Zero Vendor Lock-in

Abstract interfaces for every integration. Switch providers with one env variable:

# Switch LLM provider
LLM_PROVIDER=openai      # or: anthropic, gemini, ollama

# Switch Vector DB
VECTOR_DB=qdrant         # or: weaviate, pgvector, pinecone

⚡ Production RAG Pipeline

Complete document ingestion, chunking, embedding, and retrieval:

// Ingest documents
await rag.ingest(documents, { chunkSize: 512, overlap: 50 });

// Query with context
const response = await rag.query("What is the refund policy?", {
  topK: 5,
  rerank: true,
});

🔐 Enterprise Authentication

Multiple auth strategies with unified interface:

Better Auth - Modern, type-safe, self-hosted
NextAuth.js - Flexible OAuth, social logins
Clerk - Managed auth with great DX
JWT - Service-to-service authentication

📊 Built-in Observability

Track everything that matters in production:

Token usage and cost tracking
Request latency monitoring
Error aggregation
LLM response quality metrics

Use Cases

💬 AI Customer Support Build intelligent chatbots that understand your product docs and provide accurate answers with citations.	📚 Document Q&A Create internal knowledge bases that let employees query company documents naturally.
🔍 Semantic Search Replace keyword search with AI-powered semantic search across your content.	🤖 AI Assistants Build domain-specific AI assistants with memory and tool use.

Tech Stack

Layer	Technology
Frontend	Next.js 15, React 19, Tailwind CSS
Backend	FastAPI / Hono / NestJS
Database	PostgreSQL (Neon/Supabase), SQLite (Turso)
Vector DB	Qdrant, Weaviate, pgvector
ORM	Drizzle / Prisma / SQLAlchemy
Auth	Better Auth, NextAuth, Clerk
AI	OpenAI, Anthropic, Google Gemini, Ollama
Infrastructure	Docker, Kubernetes, Vercel

Documentation

Section	Description
Quick Start	Get running in 5 minutes
Architecture	System design and patterns
LLM Providers	Configure AI providers
Vector Databases	Vector storage options
RAG Pipeline	Document ingestion and retrieval
Deployment	Production deployment guides
Security	Security best practices

Project Structure

my-ai-app/
├── app/                    # Next.js App Router
│   ├── (auth)/            # Auth pages
│   ├── (dashboard)/       # Protected pages
│   └── api/               # API routes
├── components/            # React components
│   ├── chat/              # Chat UI components
│   └── ui/                # shadcn/ui components
├── lib/
│   ├── ai/                # LLM client abstraction
│   ├── auth/              # Auth configuration
│   ├── db/                # Database schema & client
│   └── search/            # Vector search client
└── types/                 # TypeScript types

Security

AI Stack is built with security as a first-class concern:

✅ Authentication - Multiple battle-tested auth providers
✅ Authorization - Role-based access control ready
✅ Data Encryption - TLS in transit, encryption at rest
✅ API Security - Rate limiting, CORS, input validation
✅ Secrets Management - Environment-based configuration
✅ Audit Logging - Track all sensitive operations

Read our Security Documentation for details.

Contributing

We welcome contributions! See our Contributing Guide for details.

# Clone the repo
git clone https://github.com/princepal9120/ai-stack.git

# Install dependencies
pnpm install

# Run development
pnpm dev

License

Built with ❤️ for the AI developer community

Website · Documentation · GitHub · Twitter

princepal9120/ai-stack

README