Promptly

A chatbot web app with streaming Gemini replies, saved chats organised into notebooks, and a try-without-signing-up demo mode.

Features

• Streaming chat: Gemini replies stream from /api/chat, rendered as markdown, with auto-generated chat titles and a mock mode that skips model calls.
• Chats and notebooks: Chats and messages are saved per user in Postgres. Notebooks group related chats and have their own pages.
• Prompts library: 12 starter prompts plus your own saved prompts at /prompts, with search and categories. Use fills the chat input of a new chat.
• Export: Download any saved chat as a Markdown file from its three-dot menu in the sidebar, on notebook pages or on the chat page.
• Sidebar: Search, new chat, pinned and recent chats, and rename, pin and delete menus with delete confirmation.
• Input extras: File attachments and voice dictation.
• Guest chat: Signed-out visitors can chat in memory without saving.
• Demo mode: "Try the demo" signs visitors into a temporary, freshly seeded user that a daily cron job cleans up.
• Usage limits: Per-minute, daily and global limits on Upstash Redis. Over the limit, replies switch to a sample reply.
• Settings: Dark and light themes, plus a free-tier Gemini model choice saved per user.

Tech stack

• Next.js 16 (App Router), React 19, TypeScript
• Tailwind CSS v4 and shadcn/ui
• Clerk for authentication
• Prisma with Postgres on Neon
• Vercel AI SDK with Google Gemini
• Upstash Redis for rate limiting
• Vitest for unit tests

Getting started

1. Install dependencies (this also runs prisma generate):

   npm install
   

2. Copy .env.example to .env and fill in the values (see below).

3. Apply the database migrations:

   npx prisma migrate dev
   

4. Optionally seed the development database with sample notebooks, chats and messages:

   npm run db:seed
   

5. Start the dev server and open http://localhost:3000:

   npm run dev
   

Environment variables

Variable	Purpose
NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY, CLERK_SECRET_KEY	Clerk authentication
GEMINI_API_KEY	Gemini API access
USE_AI_MODEL	Set to false to return mock replies and skip model calls
DATABASE_URL, DATABASE_URL_UNPOOLED	Pooled and direct Postgres connection strings
SEED_ALLOWED_HOST	Host of the development database; db:seed refuses any other host
UPSTASH_REDIS_REST_URL, UPSTASH_REDIS_REST_TOKEN	Redis for rate limits
GLOBAL_DAILY_MODEL_LIMIT	Total model calls per day across all users (default 500)
CHAT_DAILY_LIMIT	Optional. Model replies per person per day (default 10)
CRON_SECRET	Protects /api/demo/cleanup; Vercel Cron sends it as a bearer token
DEMO_LIMIT_PER_IP, DEMO_LIMIT_GLOBAL, DEMO_USER_TTL_HOURS	Optional. Demo limits per address and overall, and hours before a demo user is deleted

Scripts

Command	What it does
npm run dev	Start the development server
npm run build	Build for production
npm start	Run the production build
npm run lint	Run ESLint
npm test	Run the Vitest unit tests once
npm run test:watch	Run Vitest in watch mode
npm run db:seed	Seed the development database

Project structure

• app/: routes (/, /chats/[chatId], /notebooks/[notebookId], /prompts, auth pages) and API routes (/api/chat, /api/demo)
• components/: UI by feature (chat, sidebar, notebook, prompts, layout) and shadcn components in ui/
• actions/: Server Actions for chats, messages, prompts and settings
• lib/: AI setup, rate limits, validation, database and helper code
• prisma/: schema, migrations and seed data
• context/: project specs, coding standards and feature history

Deployment

Deploy on Vercel. Run prisma migrate deploy before the app starts in production, and set the environment variables above. vercel.json schedules a daily cron job (04:00) that calls /api/demo/cleanup to delete expired demo users.