Plumbly

AI-powered review assistant that walks customers through leaving a Google review after a plumbing service call.

Problem

Small service businesses like plumbing companies rely on Google reviews for growth, but customers often don't leave reviews — not because they're unwilling, but because the process has too much friction. They forget, get confused by the steps, or give up halfway through.

Solution

Plumbly is an AI chat agent that guides customers step-by-step through leaving a Google review. After a job is completed, the customer receives a link. They open it, and an AI agent — powered by Google Gemini — walks them through the entire review process conversationally, adapting to their device (iPhone/Android) and answering questions in real-time.

Features

• Conversational review guidance — AI agent detects device type and provides tailored step-by-step instructions
• Smart Q&A — Handles common questions ("I don't have a Google account", "Where do I tap?")
• Session persistence — Redis-backed conversation state with automatic expiry
• Customer pipeline tracking — Track customers from contacted → in-progress → review completed
• Unique customer links — Each customer gets a personalized review session URL

Tech Stack

Layer	Technology
AI Agent	Google Gemini 2.0 Flash via GenAI SDK
Backend	Python + FastAPI
Session State	Redis
Frontend	Web chat interface
Hosting	Railway
Containerization	Docker

Architecture

Customer → Browser Chat UI → FastAPI Backend → Gemini GenAI SDK → Gemini 2.0 Flash
                                    ↕
                                  Redis
                            (sessions, history,
                             pipeline tracking)

Setup

Prerequisites

• Python 3.11+
• Redis instance (local or cloud)
• Google Cloud account with Gemini API access

Local Development

# Clone the repo
git clone https://github.com/YOUR_USERNAME/plumbreview.git
cd plumbreview

# Create virtual environment
python -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Set environment variables
cp .env.example .env
# Edit .env with your API keys

# Run Redis (if local)
redis-server

# Start the app
uvicorn app.main:app --reload

Deploy to Railway

# Push your changes to GitHub, then connect the repo to Railway.
# Railway should use the repo root and build from the root Dockerfile.

Required Railway environment variables:

• GEMINI_API_KEY
• REDIS_URL
• GOOGLE_REVIEW_URL
• BUSINESS_NAME

Recommended Railway service settings:

• Root directory: /
• Builder: Dockerfile
• Auto-deploy: enabled for your main branch

Built For

• Gemini Live Agent Challenge Hackathon — #GeminiLiveAgentChallenge
• Demonstrates real-world application of Google Gemini multimodal AI with Railway deployment

License

MIT