From my previous blog post – “Many IBM i systems store vast amounts of transactional data in DB2 for i. AI can automate data extraction, transformation, and analysis to generate real-time business insights. For example, A machine learning model can categorize customer complaints stored in IBM i databases, helping businesses identify trends and address concerns faster.”

Many organizations running on IBM i have accumulated decades of transactional data in DB2 for i. These data silos often contain essential details like customer behavior patterns, product performance insights, service gaps that go unnoticed due to the legacy nature of the systems and their complex data structures. However, post-modernization, when data is accessible via APIs or replicated into cloud environments like Google BigQuery or AWS Redshift, AI and ML can transform how businesses extract value from that historical and real-time data.

The challenge is:  Too Much Data, Not Enough Insights

Before modernization, many IBM i systems store data in physical files with obscure field names (CUSF01, ORDH02) and embedded business logic in RPG or CLP programs. Extracting meaningful information often involves manual work or custom report development. For example, a customer support table may look like this:

Analyzing this data, recognizing patterns in complaints, or determining priority all require efficient automation.

How AI transforms this Data into Real-Time Insights

After modernization, organizations can expose DB2 for i data via APIs or data pipelines, making it accessible to cloud-native AI services. Here's how AI steps in: 

•  Automated Data Extraction and Transformation (ETL)

AI tools like Google Cloud's Dataflow or AWS Glue can automate the understanding of legacy field patterns and recommend transformations. For example, ORDSTS = 05 (Order Status Code) can be transformed to  Order_Status = "Ready for Shipment"

• Natural Language Processing (NLP) for Customer Complaints

A machine learning model trained on categorized complaints can classify new entries from DB2 for i automatically. For example, a simple chart can be illustrated as follows:

AI not only classifies complaints in real-time but also notifies managers when there is an unexpected surge in a particular category.

• Predictive Analytics for Operational Decisions

By training models on historical purchase and inventory data from DB2 for i, businesses can forecast demand, reduce overstock, or anticipate support needs. An example scenario could be: "Considering historical seasonal trends and recent orders, there is a 40% chance of stockout for Product X next week in Region Y."

From IBM I to AI-powered dashboards, a typical implementation flow can be as follows:

This flow represents the transition from traditional storage to modern, intelligent analytics.

Some of the business benefits can be listed as:

• Quick decisions with AI-tagged insights on real-time dashboards
• Proactive Problem solving with early detection of customer service or inventory issues
• Increased productivity with Automation (reducing manual data processing)
• Enhanced customer experience by quickly addressing categorized complaints 

In conclusion, modernizing an IBM i system is a significant step, but integrating it with AI can enhance its capabilities. AI can analyze raw data and provide meaningful insights, allowing business users to make quicker and more informed decisions. This integration can assist in tasks such as classifying customer complaints, forecasting sales, or identifying anomalies in operational data. The combination of modern architecture and AI offers a robust solution for enhancing legacy systems.