A complete, production-ready retail management system with a built-in AI engine — deployable in minutes, full source code included.
 
What It Is
Articsledge AI Point of Sale Software is a full-stack retail platform that combines a real-time point-of-sale interface with seven purpose-built AI features in one self-contained system. The database, AI engine, web interface, background workers, and reverse proxy all ship pre-configured and start with a single command.
 
This is not a subscription. You receive the complete source code, own it outright, and run it on your own infrastructure. No recurring platform fees. No external AI API dependencies for the core feature set — your transaction data never leaves your server.
 
The Point-of-Sale Interface
The POS terminal runs in any modern browser — no software installation needed on individual machines. Real-time updates are delivered over a live WebSocket connection, so inventory counts and order status stay current across every open session without manual refreshes.
 
Seven Built-In AI Modules
Each module runs on your server using proven open-source models and degrades gracefully if a dependency is unavailable — the system always keeps running.
 
Demand Forecasting — Predicts 30-day sales for each product using a two-model ensemble: Prophet (captures weekly rhythms and seasonal trends) blended with XGBoost (learns from recent sales velocity). Falls back automatically if either model is unavailable.
 
Anomaly Detection — Scans sales data for unusual patterns in order totals, item counts, discounts, and timing. Flagged anomalies surface in the manager dashboard with a risk score.
 
Fraud Detection — A PyTorch neural network (Autoencoder) learns what a normal order looks like from your own transaction history, then flags orders that deviate significantly for human review.
 
Customer Segmentation — Clusters your customer base using RFM analysis (Recency, Frequency, Monetary value) and automatically labels customers as VIP, Regular, New, or At Risk.
 
Dynamic Pricing — Surfaces price adjustment suggestions based on product velocity and stock levels. Always a manager decision — never automated price changes.
 
Semantic Product Search — Staff find products using natural language. Powered by FAISS and sentence-transformers, the search understands meaning, not just exact SKU matches — all running on-device.
 
Conversational Business Assistant — A chat interface that answers questions using your live business data. Works without any OpenAI key using a built-in rule-based fallback. The OpenAI key is entirely optional and only enhances this one feature.
 
Day-One AI Accuracy
An included seed script generates 180 days of realistic synthetic sales history — covering seasonal patterns, product categories, payment methods, and customer segments — so every AI model produces meaningful results from the moment you first log in. No real customer data is required to get started. Models retrain in the background at any time without interrupting live operations.
 
Role-Based Access
Cashiers — POS terminal and customer lookup.
Managers — Analytics, inventory, order history, and all AI insight views.
Admins — Full access including employee management, settings, and AI model retraining.
 
Technology at a Glance
The backend runs on FastAPI (Python) with PostgreSQL 16 for data storage, Redis for caching and task queuing, and Celery for background AI jobs. The frontend is React 18 with TypeScript, TailwindCSS, and Recharts for analytics visualizations. Nginx handles the reverse proxy. Every component is containerized — Docker Compose orchestrates all of them together, handles health checks, and manages startup order automatically. You do not need to install Python, Node.js, or PostgreSQL individually.
 
Security Foundations
Passwords are stored as bcrypt hashes. All endpoints require JWT authentication. Role checks are enforced at the API layer. The software ships with a production checklist covering JWT secret rotation, database password hardening, CORS configuration, SSL/TLS setup, and backup guidance. A full interactive API reference is available at /api/v1/docs once deployed.
 
What's Included
Complete backend and frontend source code — all seven AI modules, Docker Compose orchestration, PostgreSQL schema with Alembic migrations, Celery worker setup, 180-day seed data script, backend test suite (pytest), frontend test configuration (Vitest), environment variable template, and a production deployment checklist.
 
Requirements
Docker Desktop (8 GB RAM recommended)
OpenAI API key — optional; only needed for the enhanced AI assistant. Every other AI feature runs entirely on-device.