Production-ready Next.js 16 & Prisma 7 pgvector starter kit with local Ollama RAG and Voyage AI.

I’m Nageshkumar, and I just launched Aether OS Pro — a production-ready Next.js 16 starter kit built to solve a major headache when building AI SaaS apps: high LLM API costs and painful vector DB infrastructure setups.

Instead of spending 40+ hours setting up vector databases, document chunking, and semantic caching from scratch, this boilerplate gives you a complete, local-first RAG foundation ready to ship.

🛠️ Tech Stack & Core Features

• Framework: Next.js 16 (App Router, Server Actions, React 19)

• Database & Vectors: PostgreSQL + Prisma 7 with native 1024-dimension pgvector support (no third-party vector DB lock-in)

• Dual Embeddings Setup: Pre-configured for Voyage AI (production accuracy) and local Ollama runtimes ($0 token testing offline)

• Semantic Query Cache: Sub-millisecond in-memory cache layer to skip redundant LLM calls and cut API bills

• Monetization: Built-in Paddle payment modal integration

🎥 Demo & Code Access

• 🎬 Watch the 6.5-min Technical Walkthrough: YouTube Demo

• ⚡ Instant Codebase Download (.zip via UPI): Kreato Checkout

💬 Let's Discuss

1. How are you currently handling semantic query caching in your AI pipelines?

2. Are you using local vector engines (pgvector) or third-party vector databases (Pinecone, Qdrant) for your MVPs?

Drop your feedback, questions on the Prisma 7 setup, or ideas below!