Intellectual Property India Publishes Patent Application for A Decentralized Edge-Ai System For E-Commerce Demand Forecasting.

MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202621066108 A) filed by Prof. Prashant Laxmanrao Mandale; Mangesh Shivaji Andhale; Gahesh Shivanand Argulwar; Ashraf Suhail Ahmed; and Prem Kishor Bankar on May 26, 2026, for A Decentralized Edge-Ai System For E-Commerce Demand Forecasting..

Inventors include Prof. Prashant Laxmanrao Mandale; Mangesh Shivaji Andhale; Gahesh Shivanand Argulwar; Ashraf Suhail Ahmed; and Prem Kishor Bankar.

The application for the patent was published on August 07, 2026, under issue no. 32/2026.

Abstract: The invention provides a decentralized Edge-AI demand forecasting and retail optimization framework engineered to streamline inventory logistics and campaign strategies for resource-constrained enterprises. The system architecture integrates an on-premise transactional data ingestion pipeline, an automated machine learning (AutoML) forecasting engine, an Explainable AI (XAI) narrative module, a multi-stage campaign builder, and an interval-bound data comparison module into a unified, serverless application. Raw transactional data streams are securely stored and verified within a localized relational database layer, eliminating dependencies on centralized cloud infrastructures and preserving absolute data sovereignty. The multi-model forecasting framework concurrently backtests and executes a native Python- based algorithm family including Simple Exponential Smoothing, Double Exponential Smoothing, ARIMA AI, Prophet AI, and XGBoost AI variants to synthesize a robust, performance-weighted ensemble forecast line surrounded by a statistical uncertainty confidence interval. A natural language processing abstraction layer automatically translates complex data matrices into context-aware plain-English executive summaries and operational recommendations directly on an interactive user interface. The system frontend further integrates dynamic category filtering matrices, stock efficiency alert consoles, a risk-free " What-If' promotional planning sandbox, and a modular reporting hub. Every computational block operates efficiently on basic local processors without requiring high-performance graphics processing units, successfully bridging the infrastructure-actionability gap for digital retailers.

Disclaimer: Curated by HT Syndication.