“Augment” versus “Rip and Replace” - How AI Can End Supply Chain Technology Buyer’s Remorse

^{“Augment” versus “Rip and Replace” - How AI Can End Supply Chain Technology Buyer’s Remorse}

On a recent webinar with Slync, Jim McCullen, the CIO of Century Supply Chain Solutions made an important point while discussing the potential of Artificial Intelligence. He said, “AI allows us to engage in all the conversations, without requiring everyone to do it the same way.”  

Century is a large global logistics provider, helping many Fortune 1000 importers and he’s experiencing first hand what the future holds with AI in the supply chain. Success with any IT initiative goes beyond software functionality and in the supply chain, the challenges of coordinating with large networks of partners has vexed IT projects for decades. With AI, maybe things are going to get easier.

‍

Opportunities and Investments on the Rise

The supply chain technology market has matured into a robust ecosystem of software and hardware providers over the past three decades. According to Gartner, companies spend more than $20 Billion each year on Logistics and Supply Chain Management (SCM) software, making it the fastest growing enterprise application software segment.

The ecosystem that has emerged around the SCM software category is vast, which is not surprising considering the complexity and scale involved. To date, hundreds of billions of dollars have been invested to implement solutions that cope with the extraordinary amounts of wasted time, breakdowns, paper, email, and missed opportunities. 

For those of us that work in the supply chain, the evolution and growth of IT is not a surprise.  Attend any major conference and you will find the educational agenda dominated by sessions devoted to the various slices of the technology functional pie.

‍

Hype vs Reality

From the vast array of planning systems, to warehouse management, to order management to transportation management, to trade compliance, to robotics, to supply chain finance, there is no shortage of successful vendors and powerful case studies that showcase results and value.  A Google search of “supply chain technology” results in more than 1 billion results.

The plethora of proven SCM technology is great, but it is also a problem. 

A single large enterprise supply chain could involve more than 20,000 partners, scattered across the globe. Suppliers, carriers, freight forwarders, brokers, consolidators, consultants – all different companies – play vital roles in getting goods to stores, factories and consumers. Many partners have their own technology systems, while many others are small and use Microsoft office to operate. 

There are thousands of different supply chain networks, each designed to support a specific business or commodity. In some cases, partners such as ocean carriers and 3PLs are common, which adds further complexity to those providers. How do they satisfy their customers’ IT needs, while also running their own, large business?

The achilles heel of software is data. The dashboards and reports showcased in conference demos look fantastic, but without the underlying data, they’re just pretty pictures. For software to work as advertised, the data must be accurate, complete, and timely. The challenge of accumulating data across partner networks is a daunting task. 

When you combine the realities of needing to digitally engage with a large diverse network of partners with the need for high octane data to fuel the software that has been implemented to improve a supply chain workflow, the challenges expand exponentially. This is why email and Excel are still pervasive.

‍

The Single Version of Truth Myth

Another common theme from the various SCM influencers and conferences over the years is the concept of universal data model, a single version of supply chain truth.

“Data is the new Oil” was a session I personally helped develop with a $50B hi tech customer at the Gartner Supply Chain Executive Conference back in 2012. Our POV was simple. Social media network information models could be used for SCM, where an update to a shipment or order could be “posted” on the network so all interested parties could see and act on the same thing. We compared it to posting a professional update on LinkedIn versus emailing all of your professional contacts with the news. It made perfect sense at the time, but little has changed since.

There have been a number of data exchange standards such as EDI, XML, and APIs. Other companies launched portals, where partners can login and provide updates. Suppliers are often told (or threatened) to use portal X to receive and acknowledge purchase orders, or they won’t get paid. I’m sure there are 500 case studies out there proclaiming victory and how missing data is no longer a problem. However, what’s working on the edges is not universal, it’s the exception. 

The path to a single version of truth for all supply chain data is not through coercion, standards bodies, the next iteration of XML, or programming language. 

‍

AI to the Rescue

Here’s an idea. Instead of seeking ways to force your way on partners, why not try something else? Allow them to operate and communicate as they do today and use AI to consume, translate and standardize the information for the application that needs it.

Large language models (LLMs) are maturing rapidly. We have all seen how they can write essays, contracts, and emails. With the right teaching libraries, LLMs can also read and interpret complex business documents, regardless of format or language. If a small supplier prefers to communicate by emailing a PDF, let the AI do the work of updating the importer systems.

This becomes especially important for companies that have already invested in technology. As systems evolve, obsolescence is an issue. Unlike upgrading your iPhone every three years, it’s not easy to rip and replace an enterprise system that has been implemented at a significant cost. With AI as the data interpreter, the ability to augment existing system investments is more attainable than ever before.

Imagine a large automotive OEM, who spent more than $100 million rolling out a global transportation management system to support all global logistics operations. That is not going anywhere. Five years ago, the partner network data quality issue might have rendered that solution only partially effective. 

Today, it is possible for a startup with a state of the art AI solution design to bring together the unstructured data that is common in global logistics and significantly upgrade any installed TMS system. For that automotive OEM, their investment is not only preserved, it’s expanded. 

The AI hype machine is now being compared to Blockchain, which fizzled after a few spectacular years atop the bold predictions ladder. Nobody understood Blockchain, and there were few examples of tangible business uses to be had.  AI is a different beast. In a very short amount of time, the business use cases are not only taking place, but CIOs are doing webinars talking about how it is already working, breaking boundaries for the first time. 

Nobody wants to start over and admit defeat. A CIO at a large global company will do everything they can to preserve and build upon their investments because it goes beyond just the cost of the IT. The change management process is a big hurdle so if AI can in fact improve the performance of software and do it in a way that involves minimal change management, that shows real promise.

It’s still early, but it’s a big deal. Believe the hype.

‍

‍