Complete Feature Overview & Technical Documentation
🎯 Core Features
- Natural Language Processing
"Book 15 mins with Peter" - Smart Contact Collection
Name + Email OR Phone required - Intelligent Time Management
Real-time availability & conflict detection - Multi-Format Meeting Support
30-90 minute sessions
🎥 Google Meet Integration
- Automatic Video Conference Creation
Detects keywords automatically - Real Google Meet Links
Generated via Calendar API - Smart Detection
"Google Meet", "video call", "online meeting" - Universal Compatibility
Works with all meeting types
📧 Email System
- Dual Email Invitations
Sent to both Peter & user - Rich Calendar Attachments
.ics files with Meet links - Smart Email Collection
Optional but encouraged - Professional Templates
Branded communication
🧠 AI Capabilities
- Groq Llama-3.1-8b Integration
Fast, efficient LLM - Context-Aware Conversations
Remembers user details - Information Extraction
Complex multi-part requests - Intelligent Error Handling
Graceful conflict resolution
🎙️ Speech-to-Speech (NEW!)
- Real-time Voice Input
WebSocket STT @ 16kHz - AI Voice Responses
Text-to-Speech output - Natural Conversations
Hands-free booking experience - GPU-Accelerated STT
Fast, accurate transcription
📋 Setup Guide
1️⃣ Prerequisites
- Python 3.9+
- Google Cloud Project
- Anthropic API key
2️⃣ Google Calendar Setup
- Enable Google Calendar API
- Create OAuth2 credentials
- Run: python3 complete_oauth.py
3️⃣ Environment Configuration
- Copy .env.example to .env
- Add ANTHROPIC_API_KEY
- Add Google Calendar ID
- Add contact information
4️⃣ Dependencies & Launch
- pip install -r requirements.txt
- uvicorn app.api.main:app --reload
- Visit: localhost:8000/simple-voice
💡 Usage Examples
🔧 Technical Architecture
Backend
- FastAPI REST API
- Redis/JWT session management
- Google Calendar API
- SMTP email service
Frontend
- HTML/JS voice widget
- WebSocket STT integration
- Real-time audio processing
- Mobile-responsive design
AI & Speech
- LlamaIndex ReActAgent
- Groq Llama-3.1-8b (primary)
- WebSocket Speech-to-Text
- Text-to-Speech output
Deployment
- HuggingFace Spaces
- Docker containerization
- Environment-based config
- Production-ready setup
🚀 Development Journey
Phase 1: Core Scheduling (anthropic-sonnet branch)
- Replaced Gemini with Claude Sonnet 4
- Built intelligent contact collection
- Added email invitation system
Phase 2: Google Meet Integration (google-meet branch)
- Added video conference detection
- Integrated Google Calendar conferenceData API
- Enhanced email templates with Meet links
Phase 3: Speech-to-Speech AI 🎙️ (voiceCal-ai-v3)
- Real-time Voice Conversations - Natural speech interactions with the AI
- Speech-to-Text Integration - WebSocket-based STT service at 16kHz
- Text-to-Speech Output - AI responds with both text and voice
- Advanced LLM Models - Groq Llama-3.1-8b-instant with Claude & Gemini fallbacks
- ReAct Agent Pattern - Intelligent tool calling with LlamaIndex
- WebSocket-based STT - Connected to GPU-accelerated service
- Voice-First UX - Simplified "Just speak into your microphone" interface
🎯 Key Benefits
For Users
- Natural conversation scheduling
- No complex forms
- Automatic calendar integration
- Google Meet links on demand
For Business
- 24/7 availability
- Professional brand presence
- Reduced scheduling overhead
- Enhanced client experience
For Developers
- Modern Python/FastAPI stack
- Clean, extensible architecture
- Production-ready deployment
- Comprehensive documentation
For AI Enthusiasts
- Claude Sonnet 4 integration
- Real-world AI application
- Advanced prompt engineering
- Natural language processing
LinkedIn Post Caption Ideas
Option 1 - Technical Focus:
"🚀 Just completed VoiceCal.ai - an AI-powered scheduling assistant that transforms how we book meetings! Built with Claude Sonnet 4, it handles natural language requests, creates Google Meet links automatically, and sends professional email invitations. The AI extracts contact info from conversations like 'This is Betty, call me at 630-800-1180' and books meetings intelligently. #AI #Python #GoogleMeet #Automation"
Option 2 - Business Value:
"💡 Solving the scheduling headache with AI! VoiceCal.ai lets clients book time using natural conversation - no forms, no back-and-forth emails. Just say 'Book a Google Meet with Peter tomorrow at 2pm' and it handles everything: calendar conflicts, Google Meet links, email invitations. The future of client onboarding! #ClientExperience #AI #BusinessAutomation"
Option 3 - Developer Journey:
"🔧 From idea to production: Building VoiceCal.ai taught me advanced Google Calendar API integration, AI prompt engineering with Claude Sonnet 4, and creating seamless user experiences. The system now handles complex requests like 'Book 15 minutes Friday morning' and creates real Google Meet conferences automatically. Open to discussing the technical architecture! #SoftwareDevelopment #AI #GoogleAPI"
🎙️ Built with ❤️ using Groq LLama-3.1, Speech-to-Speech AI, Google Calendar API, and modern Python
A complete voice-powered AI scheduling solution from concept to production
Deployed on HuggingFace Spaces • Version 1.1.6