AI POS / Retail ERP

Katalog UMKM

An intelligent web-based POS and inventory catalog system integrated with Generative AI to automate stock tracking, streamline cashier checkouts, and generate predictive demand forecasting.

CategoryCustom ERP & Retail POS
FocusPOS, Inventory, AI Insights
RoleLead Full-Stack Developer & AI Integration Engineer
Katalog UMKM AI POS dashboard preview

Overview

Katalog UMKM was designed for micro, small, and medium retail businesses, especially electronics and computer merchants managing high-value inventory, supplier procurement cycles, and serial-based warranties.

Problem

Retail operators often track sales and stock through journals or disconnected spreadsheets. This causes inaccurate stock levels, weak procurement decisions, inaccessible analytics, and overly broad internal access for cashiers and managers.

Solution

We built a secure multi-role web application with cashier and owner portals, a contextual AI assistant, AI-assisted demand forecasting, and an automated POS workflow with stock deductions, invoices, and warranty tracking.

Grounded AI Retail Operations

The AI assistant maps natural-language questions to targeted database context before generating answers. It detects search intent, extracts relevant sales, stock, supplier, and customer records, then produces grounded business responses without requiring owners to write SQL.

Key Features

  • Dynamic checkout terminal with cash, transfer, and QRIS payment flows.
  • Database-grounded AI assistant for sales milestones, supplier sources, and buyer history queries.
  • Predictive stock planning based on historical sales velocity and fallback moving averages.
  • Role-based access controls separating cashier operations from owner analytics.
  • Automated warranty registry linked to finalized sales details.
  • Supplier, purchase, restock, and threshold warning workflows.

Tech Stack

Laravel 12 MySQL Blade Alpine.js TailwindCSS AI API
BackendLaravel 12, PHP 8.2, role middleware, service layer
DatabaseMySQL models for sales, purchases, warranties, suppliers, and stock
FrontendLaravel Blade, Alpine.js, Tailwind CSS, Bootstrap utilities
AI LayerGrounded natural-language search and forecasting workflows
DocumentsPDF receipts and spreadsheet export support

Contribution

Engineered the database schema and model relationships for sales, purchases, warranties, and stock tracking; built the grounded AI search service; designed role-based route middleware; created the POS checkout workflow; and integrated PDF receipt generation.

Impact

The system helps reduce stockouts through live alerts and predictive restock advisories, protects financial reporting through role-based access, and lets non-technical owners ask business questions in plain language without writing SQL.

Need AI Inside Retail Operations?

We can help connect POS, inventory, forecasting, and grounded AI insights into one practical business system.

Discuss Project