intermediate⏱ ~30 minutesai-app

AI App Starter — Cursor + D1 + Railway

Chat interface with streaming, history, and auth

Build your own ChatGPT-style app with streaming responses, conversation history, and usage limits. Powered by Claude or GPT-4.

CursorNext.js 14Cloudflare D1ClerkRailwayTailwind + shadcn/ui

What's Included

🤖

Streaming Chat

Real-time streaming responses from Claude or GPT-4

💬

Conversation History

Full chat history stored in D1

🔐

Auth + Usage Limits

Clerk auth with 50 messages/day free tier

📝

Markdown Rendering

AI responses rendered as formatted markdown

📊

Usage Dashboard

Track token usage and costs per user

Setup Guide

Clone and install

Clone the AI app starter.

$ git clone https://github.com/gptsters/ai-chat-starter my-ai-app && cd my-ai-app && npm install

Set up Clerk

Create Clerk account and add API keys.

Add to .env.local

NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY=

CLERK_SECRET_KEY=

Create D1 database

For conversation history and usage tracking.

$ npx wrangler d1 create ai-chat-db

Add AI API key

Get your API key from Anthropic (Claude) or OpenAI (GPT-4).

Add to .env.local

ANTHROPIC_API_KEY=

💡 Claude Sonnet is the best balance of quality and cost for chat apps.

Deploy

Deploy to Railway.

$ npx railway login && npx railway up

Deploy free →

Cursor Prompt

Paste this into Cursor to generate the full codebase

Build an AI chat application:
- Next.js 14 App Router with streaming
- Cloudflare D1 for conversation history
- Clerk auth with usage limits
- Anthropic Claude API (claude-sonnet-4-20250514)
- Vercel AI SDK for streaming

Database tables:
- conversations: id, user_id, title, created_at
- messages: id, conversation_id, role (user/assistant), content, created_at
- usage: id, user_id, tokens_used, date

Features:
1. Chat interface with real-time streaming
2. Conversation list in sidebar
3. New conversation button
4. Message history per conversation
5. Usage counter (50 messages/day free)
6. Markdown rendering for AI responses

Start with schema, then streaming API route, then chat UI.