📘 Personal Notes + AI Chatbot — Next.js · Convex · Vercel AI SDK

A personal notes application built with Next.js, Convex, Tailwind CSS, and the Vercel AI SDK. The app lets you create and manage notes, chat with an integrated AI assistant, and use RAG (Retrieval-Augmented Generation) to query your notes intelligently. This project was created primarily to learn and explore the Vercel AI SDK and RAG implementations.

✨ Features

📝 Notes

• Create, edit, delete, and browse personal notes
• Fully real-time using Convex
• Clean, minimal UI built with Tailwind & Next.js App Router

🔐 Authentication (Convex)

• Convex handles authentication out-of-the-box
• Secure access control for notes and chat
• No custom backend auth required

🧬 Vector Embeddings + Similarity Search (Convex)

• Convex generates and stores vector embeddings for each note
• Implements similarity search as the core of the RAG pipeline
• Ensures fast, accurate retrieval of relevant notes for chat queries

🤖 AI Chatbot (Vercel AI SDK)

• Integrated chatbot built with the Vercel AI SDK
• Streaming text responses in real-time
• Tool-calling ready (planned / work-in-progress)
• RAG-enhanced: the AI can “look into” your notes and answer questions based on their content

🚀 Deployment

• Fully deployed on Vercel
• Zero-config deployments for frontend + static assets

📚 How RAG Works in This Project

1. User sends a chat message
2. Convex computes/query embeddings
3. Convex performs similarity search over stored embeddings
4. Top-matching notes are attached as context
5. The AI model (via Vercel AI SDK) generates a response using your notes
6. Response streams back to the UI in real-time

This allows:

• Q&A over your notes
• Summaries, insights, relationships
• AI-powered recall of anything you’ve written

🛠 Tech Stack

• Next.js (App Router)
• Convex (auth, database, vector embeddings, similarity search)
• Tailwind CSS
• Vercel AI SDK (streaming chat + tool calling)
• TypeScript
• Vercel for deployment

🚀 Getting Started

1. Install dependencies

npm install

2. Environment variables

Create .env.local and include:

• Convex deployment keys
• AI provider API keys
• Embedding model config (if applicable)

3. Run locally

npm run dev

4. Deploy to Vercel

Push to GitHub → “Import Project on Vercel” → deploy Convex deploys backend functions automatically via their dashboard

🎯 Purpose of the Project

This project was built with a primary goal:

• To learn and understand the Vercel AI SDK
• To implement and explore RAG using Convex embeddings + similarity search
• To practice building a full-stack AI-enabled application

🔮 Future Improvements

• Full tool-calling integration