This site uses cookies to enhance your browsing experience and deliver personalized content. By continuing to use this site, you consent to our use of cookies.
COOKIE POLICY
Retailers have nightmares about merchandising system malfunctions on peak shopping days, causing in-store and online bottlenecks, pricing problems, and promotional errors. The potential costs from unprepared associates, confused shoppers, and lost sales can be extensive. One retailer wanted to build in system resilience to eliminate that risk.
A large retailer was in the middle of a broader modernization initiative. This included updating its merchandising software platform, built in Talend. The platform handled company-wide promotional deals by pushing pricing and other details to 2,000+ retail store point-of-sale systems.
The existing system had become brittle, slow, and caused frequent bottlenecks in scale and performance. Its limitations included:
The retailer asked UDig for a “like-for-like” system that could process and integrate the same promotions the same way, but without bogging down or – even worse – threatening to fail. They wanted to quickly onboard a resilient platform, with no disruption to current operations.
UDig envisioned a more robust deliverable: A faster, updated infrastructure that was not only resilient, but scalable. UDig leveraged AI to complete the project far more quickly than anticipated and help the retailer efficiently shift from its old platform to remove merchandising bottlenecks.
AI was able to accelerate this complex modernization project by helping analyze and rewrite over 180,000 lines of legacy Java code. This paved the way for a faster, scalable platform, deprecating the old Talend platform.
But parsing code with AI isn’t as simple as uploading program data and writing a clever prompt. The 180,000 lines of code were far more than the 1,000 lines of code that the typical AI platform can hold in memory. UDig developed accommodations which included:
To best leverage AI’s capabilities, the UDig team combined their technical knowledge to develop a unique process. The team compartmentalized each of the overall job components into AI-friendly sizes for analysis and optimization.
AI helped to uncover and streamline inefficiencies in the old program. Additional enhancements to logistics and computations helped speed up code translation and comprehension.
Leveraging AI efficiently allowed for the project to be completed in half the estimated time.
Early one Tuesday morning, the retailer piloted the new system by pushing promotions to one test store. The system was programmed to send a version of every type of possible deal, plus new gift card capabilities. In the testing process, 10x more SKUs were added than the retailer had ever used or planned to use, but the new system didn’t even blink.
With the test cleared, the retailer smoothly rolled out the platform to over 2,000 stores.
This improvement eliminated the risk of system freezes during high-traffic events, removing any worries about downtime or merchandising bottlenecks. More reliable performance under pressure means this retailer now has faster processing, more resilient infrastructure, and readiness for future scale.
