NexBook AI | Voice Booking React Native App

NexBook AI is a production-ready, commercial-grade appointment booking ecosystem designed for modern salons, luxury spas, wellness centers, barbershops, and aesthetic clinics. The complete platform unites an autonomous conversational AI Voice Booking Agent (@ai-agent), a high-performance React Native Customer Mobile App (built on the latest Expo SDK), an interactive Super Admin Web Portal & Real-Time POS Register (React 19 / Vite / Tailwind CSS v4), and a centralized Node.js Express TypeScript REST API with live Socket.io bi-directional synchronization, Stripe PaymentSheet integration, and automated push notifications.

⭐ Key Highlights & Commercial USPs

• ️ Autonomous Conversational Voice AI Concierge: Multi-turn spoken booking agent powered by Groq Cloud (openai/gpt-oss-120b) and LangGraph. Automatically extracts services, specialists, dates, and times from natural voice speech and prepares bookings in under 15 seconds.
• Interactive Real-Time POS Register & Walk-In Billing: Dedicated Point of Sale terminal for front-desk staff to handle walk-ins, apply custom percentage or fixed discounts, select specialists, take split cash/card payments, and generate 80mm thermal receipts.
• Stripe Payment Gateway & Native Mobile PaymentSheet: Full support for online prepayments, 25% booking deposits, Apple Pay, Google Pay, card elements, and cryptographic signature-verified Stripe webhook listeners.
• Multi-Channel Push & In-App Notifications: Real-time Expo push notifications sent directly to customer devices on booking confirmation, status changes, and payment completions, accompanied by live Socket.io admin sound alerts and HTML email receipts.
• Staff Roster, Shift & Vacation Leave Management: Configure specialist working hours, buffer times between treatments, and manage staff leave requests with automatic live calendar blocker enforcement.
• Universal React Native Mobile Experience: Built with the latest Expo SDK and NativeWind / Tailwind CSS v4, featuring animated bottom sheets, haptic feedback, service search filters, and digital QR appointment check-in passes.
• ️ Pre-LLM Security Guardrail Shield: Built-in input shield that blocks prompt injections, jailbreaks, gibberish, and off-domain chatter before reaching LLMs — saving 100% token costs on non-booking interactions.
• Executive Revenue Analytics & Financial Reporting: Live financial metrics including Gross Revenue, Completed Bookings, Specialist Performance Metrics, and downloadable CSV invoice reports.
• Zero-Code Dynamic Rebranding: Salon administrators can dynamically upload custom business logos and favicons stored on Cloudinary CDN directly via the Admin Dashboard without rebuilding or redeploying code.

IMPORTANT NOTICE TO BUYERS: NexBook AI integrates with external third-party cloud services for payment processing, media hosting, database storage, and AI inference. You must register your own accounts with these services. While generous free tiers are available for setup and testing, commercial production usage carries ongoing transaction and subscription costs as detailed below:

• Stripe Payment Gateway (Paid Service): Stripe charges pay-as-you-go per-transaction processing fees. The standard fee is typically 2.9% + $0.30 per successful transaction (rates vary by region and payment method). There are no setup fees or monthly minimums for standard accounts.
• ☁️ Cloudinary Media Storage (Free Tier with Paid Caps): Cloudinary is used for hosting service catalog images, staff profile avatars, and salon branding logos. It includes a free tier (25 monthly credits, approx. 25GB storage/bandwidth). High-volume media uploads will require a paid plan (starting at $19/month).
• MongoDB Atlas Database (Free Sandbox with Paid Tiers): MongoDB Atlas hosts your cloud database. It includes a permanent free M0 sandbox cluster (512MB storage), ideal for testing. Production deployments with large client records will require a paid cluster (starting at ~$9 to $57/month).
• ⚡ Groq Cloud AI (Free Developer Tier Available): Groq provides free developer quotas for ultra-fast LLM inference (openai/gpt-oss-120b). Commercial high-concurrency voice traffic may require paid API token usage.
• Firebase & Firebase Cloud Messaging (FCM) (Free Spark Plan): Firebase Cloud Messaging (FCM) and Expo Push Notifications provide 100% free unlimited push notification delivery under the standard Firebase Spark tier with zero ongoing message fees. Advanced optional Google Cloud / Firebase Blaze usage (e.g. high-volume Phone SMS verification) is billed pay-as-you-go.
• Expo Push Notification Service (Free): Push notifications are delivered via Expo Server SDK with no per-message fees.
• ✉️ Nodemailer SMTP Email Service: Supports free Gmail App Passwords, SendGrid, Mailgun, and AWS SES. Includes a local development console simulator for testing without SMTP credentials.

Autonomous AI Agent Stack & Architecture (@ai-agent)

NexBook AI features an enterprise-grade, stateful conversational booking agent engineered using LangChain, LangGraph, and Groq Cloud for lightning-fast sub-second voice interactions:

• ⚡ Inference Engine & Speech-to-Text: Powered by Groq Cloud (openai/gpt-oss-120b) and Groq Whisper audio transcription (/api/v1/ai/voice/transcribe) for near-instantaneous spoken dialogue.
• Compiled LangGraph 6-Node Workflow: Operates as a stateful cyclic state machine coordinating 6 specialized execution nodes:
  • classifyIntent: Natural Language Understanding (NLU) extracting requested services, specialists, dates, 24h normalized times, and add-on treatments with automatic language detection.
  • queryDatabase: Real-time MongoDB catalog search and dynamic slot availability calculation accounting for specialist working shifts, buffer times, and approved vacations.
  • executeBooking: Atomic, double-booking collision-free reservation creation with tax and add-on price computation.
  • cancelBooking: Secure cancellation tool enforcing authenticated customer scoping and booking reference verification.
  • rescheduleBooking: Conflict-checked appointment shifting within transactional MongoDB sessions.
  • generateResponse: Empathetic response generation with speech properties (rate, pitch, language), suggested quick-action chips, and pronunciation safety filters.
• Stateful Multi-Turn Session Memory: In-memory session store with TTL auto-expiry and MongoDB write-through persistence that remembers draft parameters (services, dates, times, specialists, extras) across multi-turn voice conversations.
• ️ Pronunciation-Safe Speech Synthesis: Automatically removes hyphens before price amounts to ensure client text-to-speech engines never vocalize ”$185” as “minus 185 dollars”.