Kapitan

Kapitan is a mobile-first finance companion built with Next.js 16, React 19, Tailwind CSS 4, Supabase, and the Vercel AI SDK 6.

Features

• Dashboard, bills, jars, goals, money condition, and chat screens
• Local-first demo flow with optional Supabase-backed shared demo data
• Live AI chat through the Vercel AI Gateway
• Responsive UI with a premium pirate ledger aesthetic

Getting Started

pnpm install
pnpm dev

The app runs at http://localhost:3000.

Scripts

pnpm dev
pnpm build
pnpm start
pnpm lint
pnpm type-check

Configuration

Copy .env.local.example to .env.local.

Demo Mode

Use the default values:

NEXT_PUBLIC_APP_MODE=demo

This keeps chat fully local and uses mock responses from Kapitan.

Live AI Mode

To enable streamed AI chat through /api/chat, set:

AI_GATEWAY_API_KEY=your_api_gateway_key
AI_MODEL=openai/gpt-4o-mini
NEXT_PUBLIC_APP_MODE=live

The client only sees NEXT_PUBLIC_APP_MODE. The API key remains server-only.

Supabase Demo Backend

Kapitan can run with local-only demo state, but the current app also supports a shared Supabase demo workspace behind server API routes.

1. Create a Supabase project.
2. Run the SQL in supabase/migrations/20260429_init_kapitan_demo.sql.
3. Add these server-only env vars to .env.local:

SUPABASE_URL=your_supabase_project_url
SUPABASE_SERVICE_ROLE_KEY=your_service_role_key
SUPABASE_DEMO_WORKSPACE_SLUG=default-demo

If these env vars are missing or the schema has not been applied, the app falls back to local cached demo data and shows a sync warning instead of breaking.

AI Chat

• The client uses useChat() from @ai-sdk/react
• The server route at app/api/chat/route.ts streams responses with the AI SDK UI message protocol
• Model strings such as openai/gpt-4o-mini, anthropic/claude-sonnet-4.5, and google/gemini-2.5-flash work through the Vercel AI Gateway

Tech Stack

• Next.js 16 App Router
• React 19
• Tailwind CSS 4
• TypeScript
• Vercel AI SDK 6