Software Engineering Radio - the podcast for professional software developers SE Radio 707: Subhajit Paul on ERP Automation and AI
Feb 12, 2026
Subhajit Paul, an ERP implementation leader with 20+ years in electronics manufacturing and supply chain, and an IEEE author. He explains core ERP flows like order-to-cash and procure-to-pay. He recounts real-world rollouts—what succeeds and what fails. He explores practical ML uses such as inventory and anomaly detection, plus emerging generative AI agents and integration challenges.
AI Snips
Chapters
Transcript
Episode notes
Pre‑Receiving Rollout That Initially Failed
- A pre-receiving SAP BTP rollout failed initially because physical inventory and reporting were missed.
- The team rebuilt processes, added reports, and then successfully rolled the solution to 40 sites.
ML Enhances; Gen AI Adds Conversation
- ML augments ERP by adding predictive analytics and pattern detection to static, rule-driven processes.
- Generative models add conversational and content-creation layers beyond prediction.
Train ML On ERP Contextual Data
- Train ML models with ERP data (inventory, production, layout, schedule) and store it in data lakes for feature preparation.
- Validate models like k-means, XGBoost, or regressions against historical ERP data before deployment.