CutCare is an AI-powered barber platform built around a hair analysis pipeline that turns client hair photos into structured hair profile data. Clients upload multi-angle photos, review the generated profile, and then use a RAG-powered chatbot to get detailed, personalized haircut, styling, product, and barber-note recommendations based on their own hair information.
The product also supports the full barber-client workflow: clients can discover barbers, book appointments, manage bookings, and message barbers, while barbers can manage services, availability, client conversations, and booking requests from a dedicated business workflow.
- RAG-style haircut knowledge retrieval using SentenceTransformers embeddings and a curated haircut knowledge base
- FastAPI AI backend that analyzes uploaded hair photos, stores profile predictions, and supports profile-aware recommendations
- Real-time client/barber messaging with deterministic conversation threads
- AI hair profile upload flow with multi-angle photo support
- Client barber discovery, profile browsing, service selection, and booking creation
- Barber service management, weekly availability, and booking status workflows
flowchart TD
A[Client onboarding] --> B[Upload hair photos]
B --> C[AI vision analysis]
C --> D[Structured hair profile]
D --> E[Client reviews and confirms profile]
E --> F[Saved active hair profile]
G[Barber haircut knowledge] --> H[Chunk and clean knowledge]
H --> I[Vector embeddings]
I --> J[(Vector knowledge base)]
K[Client chatbot query] --> L[Embed query]
L --> M[Cosine similarity retrieval]
J --> M
F --> N[Profile-aware prompt context]
M --> N
N --> O[RAG chatbot]
O --> P[Personalized haircut, product, and barber-note recommendation]
The AI backend is built with FastAPI and supports two connected workflows:
- Vision analysis: validates uploaded photos, analyzes available photo angles, creates a unified structured hair profile, and stores the prediction under the client in Firestore.
- Profile-aware recommendations: loads the client's confirmed active hair profile, retrieves relevant haircut knowledge with embeddings, and builds a response grounded in both the user's question and the stored profile.
The RAG knowledge base currently focuses on haircut descriptions, barber communication guidance, and style recommendation context. The long-term direction is to expand the knowledge base with product recommendations, hair-care guides, and barber-facing client notes where appropriate.
Clients can create an account, complete onboarding, search for barbers by profile and service information, book appointments from valid availability slots, manage upcoming bookings, and message barbers in real time.
The AI flow adds a more personalized layer: clients upload front, side, and back hair photos, receive a structured hair profile, review or correct the AI output, and then use that confirmed profile in the recommendation chat.
Barbers can create a business profile, list services, set weekly availability, review incoming booking requests, confirm or cancel appointments, mark appointments complete, and message clients. The barber side is designed around day-to-day business operations rather than a generic social profile.
- Mobile: React Native, Expo, Expo Router, NativeWind / Tailwind
- Backend/data: Firebase Authentication, Cloud Firestore, Firebase Storage
- AI service: FastAPI, Pydantic, OpenAI API integration, Pillow, OpenCV
- Retrieval: SentenceTransformers, NumPy, curated haircut knowledge documents
- Tooling: TypeScript, JavaScript, Python
frontend/ React Native / Expo mobile app
ai/ FastAPI AI service for vision analysis and recommendations
functions/ Firebase Cloud Functions workspace
website/ Web workspaceImplemented:
- Authentication and persistent login
- Role-based routing and onboarding
- Client and barber profile flows
- Barber services and availability management
- Client discovery and booking creation
- Barber booking management
- Real-time messaging
- Hair profile upload, analysis, review, confirmation, and storage
- Profile-aware AI chat with embedded haircut knowledge retrieval