Solstice Pilates AI Receptionist

An AI-powered receptionist for Solstice Pilates studio that handles class bookings, availability checks, cancellations, and general inquiries via text chat and voice calls.

Features

• Class Booking — Check availability, book, and cancel classes
• Smart Suggestions — Suggests alternatives when a class is full
• Contact Management — Tracks callers in Google Sheets with call logs
• Human Handoff — Routes billing disputes, refunds, and special requests to staff
• Voice Calls — Real-time voice interaction via Vapi integration

Tech Stack

• Backend: Python, Flask
• LLM: Google Gemini 2.0 Flash (text chat), OpenAI GPT-4.1 (voice via Vapi)
• Calendar: Google Calendar API
• CRM: Google Sheets API
• Voice: Vapi AI
• Auth: Google Service Account

Project Structure

config.py              — Shared configuration (IDs, studio info, pricing)
calendar_service.py    — Google Calendar operations (availability, booking, cancellation)
sheets_service.py      — Google Sheets operations (contacts, call logging)
agent.py               — Gemini-powered agent with tool calling (Phase 1)
app.py                 — Flask server, chat UI, and Vapi webhook (Phase 1 + 2)
seed_calendar.py       — Seeds test classes into Google Calendar

Setup

1. Install dependencies

python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

2. Configure credentials

• Create a Google Cloud project and enable Calendar + Sheets APIs
• Create a Service Account and download the JSON key as credentials.json
• Create a .env file:

GEMINI_API_KEY=your_gemini_api_key

3. Set up Google resources

• Create a Google Calendar named "Solstice Pilates"
• Create a Google Sheet named "Solstice Pilates Contacts" with headers: Name | Phone | Email | Last Call Date | Call Log | Notes
• Share both with the service account email (found in credentials.json)
• Update CALENDAR_ID and SPREADSHEET_ID in config.py

4. Seed test data

python seed_calendar.py

5. Run

python app.py

Open http://localhost:8080 — use Text mode for chat or Voice mode for Vapi calls.

Phase 2: Voice (Vapi)

To enable voice calls:

1. Create a Vapi account at vapi.ai
2. Create an assistant with the tools defined in the Vapi dashboard
3. Expose the server publicly (e.g., cloudflared tunnel --url http://localhost:8080)
4. Point each tool's Server URL to https://your-tunnel-url/vapi/webhook
5. Add your Vapi Public Key and Assistant ID in app.py