Retail AI benchmark for choosing the economically right LLM by workflow
-
Updated
Jun 13, 2026 - Python
Retail AI benchmark for choosing the economically right LLM by workflow
Open-source FBO replenishment planning toolkit for marketplace sellers.
Free Business Central extension: recommends the Reordering Policy per item from its demand pattern (ADI/CV2). Now part of the free Inventory Planning app.
AI-powered retail inventory intelligence platform that forecasts product demand, identifies stockout and overstock risks, optimizes replenishment decisions, and simulates inventory cost and service-level trade-offs.
Proposes min/max replenishment rules, writes them back to stock.warehouse.orderpoint.
Plans the day's transfers on a six-tier priority ladder under transfer capacity, and reads inbound ETAs so it does not move stock that is about to arrive.
结合销量、库存、交期和季节性,设计可解释的预测与补货决策。
Caso simulado de supply chain en Valencia: análisis y previsión de demanda, planificación de inventario y reposición bajo incertidumbre. Proyecto desarrollado en Python para apoyar decisiones logísticas y evaluar distintos escenarios operativos.
Simulation-based evaluation and monitoring framework for ML-driven retail replenishment decisions - Software Engineering Graduation Project
Improve inventory management by adjusting ordering thresholds thanks to trusted data and insights.
Hierarchical retail demand forecasting and replenishment in Go: per-series model tournament (seasonal naive, Holt-Winters, ridge promo uplift, pluggable external models), explicit promo and cold-start handling, OLS/WLS reconciliation, P10/P50/P90 bands and explained order-up-to orders.
Supply and Inventory management software application
◫ Preact inventory planner with demand analysis, safety stock, shortage forecasts, suppliers, and purchase orders.
Decision-support toolkit for replenishment, built on traditional ML models.
跨境电商海外仓智能补货与实时仿真原型:预测、补货、分仓、策略对打与数字孪生沙盘。
LLM-in-the-loop shelf replenishment POC: deterministic rules handle ~90% of restock decisions, a local LLM reasons over the ambiguous rest. Kafka, Postgres, FastAPI, Streamlit, Docker.
Lubricant distributors tie up months of cash in stock: DDMRP buffers vs min-max over 800 SKUs, judged at matched fill rate
S&OP replenishment decisions computed from inventory policy - every number in the explanation traceable to the arithmetic, nothing self-approves
Warehouse management system dashboard: stock value and utilisation, ageing and ABC classes, replenishment cover, storage cost, an in-browser CBM calculator and inbound commitments. Static React and TypeScript, seeded synthetic data, 2,493 reconciliation checks. Live demo included.
承接 Legacy ERP Headless Bridge,將客戶與銷售資料轉成補貨、分群、交叉推薦與預測表單的架構 Showcase。
To associate your repository with the replenishment topic, visit your repo's landing page and select "manage topics."