VibeMap Agent

A Next.js + React app that integrates Google Maps with AI-powered location discovery using Ollama. Displays interesting Manhattan locations discovered by local language models.

Features

• Interactive Google Maps centered on downtown Manhattan
• AI-powered location discovery via Ollama LLM integration
• Custom markers for AI-suggested venues
• Responsive design with dark mode support
• Graceful fallback when API keys are missing
• Local-first architecture - runs locally with Ollama

Prerequisites

1. Google Maps API Key (required)

   • Get one from Google Cloud Console
   • Enable Maps JavaScript API and Geocoding API

2. Ollama (optional but recommended)

   • Download from ollama.ai
   • Ensure it's running locally (default: http://localhost:11434)

Environment Variables

Copy .env.example to .env.local and fill in the required values:

cp .env.example .env.local

Required Variables

• NEXT_PUBLIC_GOOGLE_MAPS_API_KEY - Your Google Maps API key (required for map to display)

Optional Variables

• NEXT_PUBLIC_OLLAMA_BASE_URL - Ollama endpoint (default: http://localhost:11434)
• NEXT_PUBLIC_OLLAMA_MODEL - Ollama model to use (default: llama2)

Getting Started

1. Install dependencies:

   npm install

2. Set up environment variables:

   # Edit .env.local with your Google Maps API key
   NEXT_PUBLIC_GOOGLE_MAPS_API_KEY=your_key_here

3. (Optional) Start Ollama:

   ollama serve
   # In another terminal, pull a model:
   ollama pull llama2  # or any other available model

4. Run the development server:

   npm run dev

5. Open http://localhost:3000 in your browser

Usage

• The map displays downtown Manhattan by default
• Click "Discover Locations" to fetch AI-suggested venues
• Click on location names to pan the map
• Markers are placed at coordinates returned by the AI

Architecture

app/page.tsx                    # Main page component
components/MapContainer.tsx     # Google Maps + Ollama integration
lib/ollama.ts                   # Ollama API client
types/location.ts               # TypeScript types

How It Works

1. Google Maps renders an interactive map centered on downtown Manhattan
2. Ollama generates location suggestions using a local LLM
3. Markers are placed on the map at the coordinates returned by Ollama
4. UI Panel displays discovered locations and error states

Notes

• All environment variables are public (NEXT_PUBLIC_ prefix) since they're client-side
• Ollama connectivity is graceful - errors don't break the app
• The map shows a helpful message if the Google Maps API key is missing
• Works offline once the map tiles are cached