AI-Enhanced ERP for Driving Manufacturing Productivity

The Challenge

To meet the growing complexity of modern manufacturing, this project integrates three AI modules into an existing ERP platform to address inventory management, quotation processing, and end-user support. The Smart Inventory Module uses Bidirectional LSTM and NARX models to predict inventory needs, production timelines, and supplier delays by analyzing historical data alongside real-time market trends. The Enhanced Quotation Generation Module, built on a fine-tuned RoBERTa NER model, extracts key information from client communications to generate accurate quotes and accelerate response times, with future iterations integrating email plugins for even greater usability. Finally, a Conversational Agent, powered by Llama 3.1 and a retrieval-augmented generation (RAG) architecture, will offer dynamic, AI-driven support to users by interpreting documentation and client-specific content. Together, these modules are expected to generate over $14 million in new revenue for the platform provider and enable $15 million in revenue growth and $137 million in cost savings for up to 150 Canadian manufacturers, while also fostering talent development and reducing time-to-market for AI applications in industrial settings.