With decades of experience navigating the complexities of the logistics industry, Rohit Laila has become a leading voice on the intersection of supply chain efficiency and technological innovation. His insights shed light on how modern distribution centers are evolving beyond traditional methods to embrace data-driven strategies. In our conversation, we delve into the sophisticated science behind ‘best pallet matching,’ exploring how this technology not only streamlines the picking process but also fundamentally enhances accuracy. We’ll examine real-world applications, such as the success seen at Associated Wholesale Grocers, and discuss the critical role of hardware flexibility in deploying these advanced systems. Finally, we’ll look toward the horizon to understand how artificial intelligence is poised to further revolutionize warehouse operations.
The article mentions it’s inefficient to assign pallets based on location alone. How does your “best pallet matching” data science model determine the optimal match, and what specific factors beyond destination does it analyze to boost productivity by 15% to 30%?
That’s really the core of the problem we set out to solve. For years, the default logic was simple: group orders going to the same area. It seems intuitive, but it leaves a massive amount of efficiency on the table. Our model operates on a much more sophisticated level. It’s a dynamic system that constantly re-evaluates the entire pool of open orders. It looks at the physical location of every single item for a potential group of orders, calculating the most efficient travel path a worker could take. It also considers order priorities, the physical size and weight of items for pallet stability, and even the due-out times. By crunching all these variables, it creates a picking assignment that minimizes travel, which is where warehouses lose the most time. It’s this holistic optimization of the travel path, not just the destination, that delivers those significant productivity gains of 15% to 30%.
Ken Ramoutar noted that warehouses often fear sacrificing accuracy when picking for multiple stores. Could you walk us through the specific features of your technology that not only prevent but actively increase accuracy in these complex, multi-pallet picking scenarios?
That fear is completely understandable and one we took very seriously. When you ask a picker to manage orders for two or three different stores on two or three different pallets, the potential for error skyrockets if you’re using traditional paper-based methods. The key is removing the mental burden and guesswork. Our system uses voice-directed applications, which act as a co-pilot for the worker. At every pick location, the system tells the worker exactly how many units to grab and, crucially, directs them to place the items on a specific pallet—for example, “place two cases on pallet one.” This closed-loop confirmation process makes it incredibly difficult to make a mistake. Furthermore, by incorporating features like paperless equipment check-ins, we build a foundation of compliance and accountability from the start of the shift. Instead of being a trade-off, we find that the structured guidance actually makes workers more focused, leading to an increase in accuracy alongside the boost in efficiency.
Associated Wholesale Grocers successfully used this tech to group diverse orders and reduce travel. Could you provide a step-by-step example of how your system would optimize the travel path for a worker picking for three completely different stores at once?
Certainly. Imagine a picker at an AWG facility, which serves over 1,100 member companies with diverse needs. The system won’t just grab the next three orders in the queue. First, it analyzes the entire order pool and identifies three distinct orders that, when combined, create a perfect logistical puzzle. For example, it might pair a large order for a supermarket with two smaller, fill-in orders for convenience-style stores. The system then generates a single, optimized travel path. The picker starts their journey and is directed to Aisle 5 for the first pick. The voice command might be, “Pick ten units of canned tomatoes, place on pallet two.” The next stop might be Aisle 12, where the instruction is, “Pick three units of cereal, place on pallet one.” Then, in the same aisle, “Pick five units of bottled water, place on pallet three.” The worker is moving in one continuous, fluid path through the warehouse, fulfilling three orders simultaneously without ever backtracking or having to decide what to do next. It transforms a potentially chaotic task into a simple, guided process.
Richard Kearns of AWG praised the ability to use “agnostic multi-modal hardware.” How does this hardware flexibility impact the implementation process for a new client, and what common challenges of proprietary systems does this approach help them avoid?
Hardware flexibility is a game-changer for operations leaders. Being “agnostic” simply means our software can run on a wide variety of devices, regardless of the manufacturer. For a new client, this dramatically lowers the barrier to entry. They don’t have to discard their existing scanners or voice headsets; they can leverage the hardware they’ve already invested in. This accelerates the implementation timeline and significantly reduces upfront costs. The biggest challenge with proprietary systems is vendor lock-in. You’re forced to buy their specific, often overpriced, hardware. If that hardware becomes outdated or breaks, you have no other option but to go back to them. An agnostic approach gives the client freedom and control. They can choose the best device for the job, mix and match different types of hardware, and aren’t held hostage by a single supplier’s roadmap or pricing. It future-proofs their investment and gives them far greater operational agility.
What is your forecast for the role of AI and data science in warehouse optimization over the next five years?
I believe we are at the very beginning of a major transformation. The pallet matching we’ve discussed is a powerful application, but it’s just scratching the surface. Over the next five years, I forecast that AI will become predictive, not just reactive. Systems will analyze historical data and upcoming promotions to anticipate order volumes and pre-slot fast-moving inventory in the most optimal locations before the orders even drop. We’ll also see hyper-personalization of work. The AI will learn the strengths and patterns of individual workers, assigning tasks that best fit their performance to maximize their efficiency and job satisfaction. The ultimate goal is to create a fully dynamic, self-optimizing warehouse environment where everything—from inventory placement to labor assignment to picking paths—is continuously adjusted in real-time by an intelligent engine, making the entire operation more resilient and incredibly efficient.
