For many industries, making a large sequence of decisions in succession can be one of the most complicated yet vital components of ensuring daily operations function as intended. The e-commerce industry is no exception, as retailers have to make dozens, if not hundreds, of choices between the point at which a sale was made and the point at which that product is delivered. Such choices can include (but are by no means limited to): which warehouse will ship the order, which shelf the order within that warehouse needs to come off of, who will pick up the order, in what order that item will be delivered alongside others within the delivery truck, and so on.
In other words, successfully running an e-commerce business is necessarily an exercise in decision-making, something medium-to-large retailers simply cannot handle at scale without the right tools given the sheer number of choices involved. Fortunately, many of these retailers do have the right tools in the form of warehouse management systems (WMS), a type of software that streamlines decision-making so businesses can do more work in less time with fewer mistakes.
Over time, however, the warehouse management system has had an increasingly difficult time keeping pace with rising demand, requiring new modes of thinking in regard to what WMS is for and how it should work. Today, that shift in thinking has led to the development of a WMS intended to work smarter, not harder, and while that development is still a work in progress, its existence marks a notable move in a different, potentially more useful direction for the software.
The Issue of Intelligence
WMS was initially created to address a physical question: how will workers know how, when, and where to do their jobs? In practice, this meant finding ways to let workers know what to do next, a process that was previously quite challenging in large warehouses full of inventory that needed to be moved in a short time.
As it turned out, WMS was able to answer these and other operational questions fairly well, allowing modern warehouses to reach new accuracy rates as workers were better able to move from one task to the next. Online commerce continues to grow, however, putting strain on warehouse management systems that can only ever operate so quickly.
Over time, this growth has revealed an important flaw in WMS: the system can execute plans well, yes, but it cannot determine whether there are problems with those plans. In other words, WMS has gotten faster, but it has not necessarily gotten any smarter, potentially limiting its usefulness at a time when the e-commerce industry is the largest it’s ever been.
Some emerging solutions to this issue of intelligence have so far served more as bandages than full-on cures, temporarily covering the problem without addressing it at the source.
For example, much of today’s WMS uses machine learning to help with forecasting and anomaly detection, and while this data may help businesses avoid some issues in the future, it tends to be less effective in the short term. A dashboard may be able to show that pick rates dropped 12% last week, yes, but if it cannot give an explanation for why that happened or provide advice on how to fix the issue for next time, such information may be too little, too late.
From Predictive to Prescriptive
Part of why early attempts at incorporating AI tools like machine learning into WMS struggled was because those tools were often placed outside of the system. This meant analysis was effectively always a step behind the actual data, causing it to almost always be out of date. Recent advancements in AI may have at least partially solved this problem, integrating the analytics stack into the WMS itself so it can run on the same live operational data the warehouse itself runs on.
One company making use of this technology is Deposco, an AI-native supply chain platform designed to provide intelligence at multiple points within a WMS. Its supply chain intelligence suite addresses questions at the executive, inventory, labor, and shipping levels, all of which is done within Deposco’s warehouse management software to keep these streams of data more closely connected.
While this approach is by no means perfect, it may help move AI from a purely predictive role into a more prescriptive one capable of making informed recommendations that pull from real-time data. At Deposco, such AI comes in the form of Felix, the company’s team of AI agents that the company’s customers can enable inside the platform.
This method may make it possible for operators to ask said agents questions about live operations continually, keeping them up to date on the latest goings on using relevant information. According to Deposco, Felix is able to provide these recommendations by drawing on transaction data comparable operators already use. The agent can then configure the system to potentially deliver a similar outcome, reducing the need for extraneous operator input.
In terms of measurable results, such automation may help businesses scale operations by doing more in less time, potentially turning otherwise wasted minutes into periods of productivity in which warehouse operators can better prepare for developments to come.
Making Use of Patterns
The power of modern AI agents lies in their ability to make reasonably accurate predictions based on large samples of data that they may be able to draw clear patterns from.
Provided those patterns hold true, the agents pulling from them may be able to make suggestions that human analysts would take longer to recognize and apply, though it should be noted that AI can still hallucinate or make incorrect assumptions based on gathered data. For these reasons, it remains imperative that even advanced WMS implement human-in-the-loop protocols to ensure automated suggestions are reasonable and applicable.
Still, advancements in WMS AI should be regarded as a positive development within the broader e-commerce industry. With proper implementation, businesses may be able to scale their overall shipping volume without needing to substantially scale their labor in kind. With a system capable of adapting to supply and demand without excessive oversight, businesses could use the time they have to better prepare for surges and droughts alike.
As the e-commerce industry continues to progress and expectations continue to rise, the key differentiator between an operable retailer and a successful one could soon come down to how well their respective WMSs handle data and the fluctuations within it. Data is, after all, the lifeblood of many modern industries, meaning its management will become all the more important as the sheer quantity of information businesses have to handle will only become more cumbersome.
Success, then, may look like the continued ability to uphold the illusion that making purchases online is as simple and effortless as completing the checkout process and picking up the package only a couple of days later.
The information provided in this article is for general informational and educational purposes only. It is not intended as legal, financial, medical, or professional advice. Readers should not rely solely on the content of this article and are encouraged to seek professional advice tailored to their specific circumstances. We disclaim any liability for any loss or damage arising directly or indirectly from the use of, or reliance on, the information presented.
Digital Trends partners with external contributors. All contributor content is reviewed by the Digital Trends editorial staff.
