Rohit Laila is a veteran of the logistics world, a man who has spent the better part of three decades navigating the intricate labyrinths of global supply chains and delivery networks. His career has mirrored the evolution of the industry itself, moving from the analog era of clipboards and manual spreadsheets to the high-stakes, data-driven environment we inhabit today. Currently, he stands at the forefront of the retail revolution, championing the integration of artificial intelligence and RFID technology as the ultimate cure for the industry’s oldest ailments. Our conversation explores the profound shift from reactive logistics to a world of predictive, autonomous inventory management, touching on the multi-trillion-dollar stakes involved, the strategic triumphs of global leaders like Zara and Uniqlo, and the rigorous roadmap required to turn raw data into a competitive fortress.
Retailers currently face over $1.7 trillion in annual losses from inventory imbalances and significant shrinkage; how do we bridge the gap between traditional ERP systems and the needs of modern omnichannel retail?
The reality is that for a very long time, retail inventory has been nothing more than a high-stakes guessing game. We see these staggering numbers—$1.73 trillion lost globally to out-of-stocks and overstocks, plus another $112.1 billion vanishing into the void of shrinkage—and it is clear that our advanced ERPs simply aren’t enough on their own. The root of the problem is structural; manual cycle counts and barcode scanning are just too slow and prone to human error to keep up with the breakneck speed of omnichannel demand. We bridge this gap by moving away from “point-in-time” data and toward a continuous, living pulse of inventory. By integrating AI as the analytical brain and RFID as the real-time data backbone, we finally move past the frustration of “phantom inventory” and start making decisions based on what is actually on the shelf right now.
How does the pairing of AI with RFID fundamentally change the nature of the data being collected compared to legacy scanning methods?
To really grasp the change, you have to understand that RFID identifies and tracks products using electromagnetic fields without needing a line-of-sight, which allows us to capture items in bulk across stockrooms and fitting rooms effortlessly. However, that raw data is often noisy; signal interference and read errors are just part of the physical environment in a busy store. This is where Artificial Intelligence becomes indispensable, acting as a sophisticated filter that resolves positional uncertainty through probabilistic modeling. Without AI, RFID just gives you a faster stream of potentially messy data, but with it, you get a refined, actionable stream of intelligence. It is the difference between hearing a thousand voices shouting at once and having a single, clear conversation about exactly where every SKU is located.
In your experience, how are global giants like H&M and Decathlon leveraging these technologies to move beyond simple tracking into true predictive demand forecasting?
What Decathlon and H&M have done is truly transformative because they’ve moved the goalposts from “what do we have?” to “what will we need?” Decathlon, for instance, has achieved a staggering 99.9% inventory accuracy in-store, which allowed them to shift staff from boring stock monitoring to actually helping customers on the floor. They use machine learning models that ingest RFID data alongside historical sales and seasonal patterns to anticipate spikes before they happen. This means stock moves before the gap ever appears on the shelf, drastically reducing those painful out-of-stock incidents during peak sports seasons. Similarly, H&M’s rollout has shown that when you have near-perfect accuracy, the productivity gains in store operations are massive, turning the backroom from a chaotic storage space into a precision-tuned fulfillment hub.
We see Zara moving inventory at full price much more effectively than the industry average; what role does autonomous replenishment play in that success?
Zara is the gold standard here because they’ve spent a decade building an ecosystem where RFID data from fitting rooms and shelf velocity feeds directly into AI models. These models don’t just suggest what to do; they help drive replenishment and even product development decisions in real time. Because they can see which items are being tried on but not bought, or which sizes are moving fastest in specific urban locations, they can automate purchase orders and inter-store transfers without waiting for a human to trigger the process. The result is a sell-through rate that is well above the industry norm, simply because they aren’t forced into the “burn and markdown” cycle that kills margins. It’s a sensory-rich process where the store environment itself is telling the supply chain exactly what the customer wants today.
Uniqlo has been tagging at the source since 2017; how does this level of integration enable a “unified pool” of inventory for omnichannel success?
By tagging every single product at the source for nearly a decade, Uniqlo has essentially turned their entire distribution network into a single, transparent warehouse. When a customer clicks “buy” online, the system doesn’t just look at a central hub; it identifies the nearest store holding that specific color and size in real time and initiates a pick for same-day fulfillment. This level of unification is impossible without the item-level accuracy that RFID provides, as retailers can’t confidently commit to an order they can’t verify. They’ve seen stocking times fall sharply and order accuracy reach near-perfect levels, which creates a seamless experience for the customer who doesn’t care where the item comes from, as long as it arrives on their doorstep within hours.
Many leaders fear the complexity of a multi-phase rollout; what is the most effective roadmap for an enterprise looking to transition from raw data to automated intelligence?
You cannot rush the foundation; if your physical infrastructure is weak, your AI will be hallucinating based on bad data. The first three to six months must be dedicated to tagging at the item level and integrating readers with your existing WMS and ERP systems to ensure clean, continuous data capture. Only once that flow is reliable should you spend the next four to eight months activating the intelligence layer, where you train your models on sales history and seasonal signals to flag discrepancies. The final leap is automation, where the system begins to act on its own outputs to trigger replenishment, and this is where change management becomes vital. It’s a journey from executing tasks to monitoring exceptions, and it requires a cultural shift as much as a technological one.
What are the hidden risks, such as “model drift” or environmental interference, that could derail a massive RFID investment?
There are definitely physical and digital hurdles that can catch a team off guard if they aren’t prepared. For example, UHF signals are notoriously finicky around metal and liquids, which means if you’re tagging electronics or beverages, you need category-level planning and perhaps HF tags to maintain accuracy. On the digital side, “model drift” is a very real threat; an AI trained on last year’s patterns might become less reliable if customer behavior shifts or new store formats are introduced. You need continuous retraining pipelines and constant KPI monitoring to catch these shifts before they impact your bottom line. Furthermore, you have to win over the store teams; if the people on the ground don’t trust the AI’s recommendations and start working around the system, the entire investment loses its value.
What is your forecast for the role of AI-RFID in sustainable retail?
I believe we are moving toward a future where “circular commerce” is the standard, and AI-RFID will be the primary engine driving that sustainability. As we move deeper into this decade, these tags won’t just track a sale; they will track the entire lifecycle of a garment, from the sustainable source of its fibers to its eventual resale or recycling. We are already seeing the groundwork for this as retailers use the 95-99% inventory precision they’ve gained to stop the overproduction that leads to immense environmental waste. By 2028, I expect the most successful retailers will be using this unified data pool to not only maximize profits but to prove their environmental impact with every single item they put into the world.
