AI as the operating system of fulfilment

For operations managers, the greatest value of AI is not simply in accelerating delivery speeds, but in strengthening efficiency, reliability, transparency and trust across the entire fulfilment chain. 

For years, AI conversations in retail have centred on marketing innovation. Personalised recommendations, dynamic content and creative automation have dominated the headlines. Yet in 2025, the part of retail under the greatest pressure is the one consumers rarely think about until something goes wrong: fulfilment. Today’s shoppers expect orders to move through the system quickly, predictably and without friction, but crucially, they expect all of this at no extra cost. Research from Scurri shows that while 38% of UK consumers now use AI somewhere in their online shopping journey, and younger shoppers increasingly rely on it, 44% still refuse to pay more for AI-enhanced delivery. At the same time, 60% of consumers expect AI to improve delivery updates, 57% expect AI to improve order allocation, and 59% want retailers to use AI during peak periods.

This creates a new operational mandate. AI must enhance reliability and efficiency without eroding margin. It must make fulfilment systems smarter, more transparent and more resilient, while remaining essentially invisible to the consumer. 

The Scurri report highlights that AI is no longer just a point solution applied to individual parts of the fulfilment process. Instead, it is evolving into a system-wide operating layer that connects forecasting, inventory placement, carrier decisions, returns management and post-purchase communication. It behaves like a silent orchestration engine, managing thousands of decisions every minute to ensure that every order has the best chance of arriving on time, in full and at the lowest possible cost to serve. 

The impact of this systems-level approach is already evident in leading global operations. Companies such as McKinsey, DHL and Unilever have demonstrated that AI can reduce inventory levels by between 20 and 30 percent, cut warehousing costs by 10 to 20 percent, and lower returns handling costs by up to 25 percent. When applied across forecasting, routing and last-mile execution, AI can halve delivery delays, increase fulfilment capacity by around 40 percent and reduce supply chain emissions by more than 60 percent. These are not speculative projections; they are being achieved in real-world environments, and they offer a clear blueprint for retail. For omnichannel retailers AI can go one step further and leverage the store network for stock availability and order fulfillment.

A consistent theme throughout the Scurri research is that consumers increasingly equate fulfilment quality with fulfilment dependability. For many shoppers, speed is no longer defined solely by next-day or same-day delivery. It is defined by whether the promised delivery time is accurate, whether tracking updates are reliable, and whether any issues are communicated proactively. Scurri’s findings show that 60 percent of consumers want AI-powered real-time tracking, 54 percent expect retailers to improve post-purchase communication through AI-driven systems, and 50 percent would prefer their returns to be automated entirely. What consumers want is not necessarily faster fulfilment; they want better visibility, fewer surprises and fewer service failures. 

AI enables this shift from reactive firefighting to predictable reliability. Predictive models can identify issues such as depot delays, failed pick-ups or weather disruptions before they escalate into customer complaints. Rather than waiting for a parcel to be declared “lost,” retailers can intervene early, with AI recommending the most cost-efficient remedy. This has a direct effect on cost-to-serve, reducing failed deliveries, minimising inbound customer queries and preventing last-minute re-routing expenses. The consumer may never know that AI helped avert a problem, but they feel the benefits through a smoother, more trustworthy delivery experience. 

Efficiency remains the primary way AI pays for itself. Because consumers are unwilling to pay more for AI-enabled service enhancements, operational gains are essential to the business case. Forecasting accuracy is one of the strongest levers, with McKinsey estimating that AI can reduce forecasting errors by 20 to 50 percent. These improvements cascade through the supply chain, leading to fewer stockouts, fewer overstocks, lower warehousing costs and a reduction in the administrative burden associated with constant manual adjustments. Even a one percent improvement in forecasting accuracy can save millions in overstock or stockout waste.

Warehousing is another major area where AI delivers measurable value. AI-driven slotting, picking and packing systems can increase capacity by 10 to 20 percent without additional labour, while significantly lowering the risk of mis-picks and damaged goods. Similarly, the application of AI to last-mile optimisation—whether through dynamic carrier allocation, predictive ETAs or intelligent routing—reduces delivery costs and emissions while improving on-time performance. UPS’s long-standing routing optimisation programme, which saved 10 million gallons of fuel annually by reducing left-hand turns, illustrates the power of incremental efficiency gains. AI multiplies these gains at scale across every parcel moving through the network.

The Scurri report makes it clear that trust has become a critical currency within fulfilment, particularly in emerging channels like social commerce. While consumers enjoy the discovery and immediacy of platforms such as TikTok Shop and Instagram, they remain sceptical about fulfilment quality. Seventy-two percent of shoppers believe AI could fix the pain points associated with social commerce fulfilment, and 57 percent expect AI to manage the logistics of social orders. Trust in these channels is not built through content; it is built through operational performance. When an order arrives reliably, consumers blame neither the platform nor the retailer. When it doesn’t, they blame both.

The practical steps for retailers are clear. Improvements in forecasting accuracy should come first, as they create a flywheel effect across inventory, warehousing and delivery. Warehouse automation should be scaled next, followed by AI-driven carrier optimisation, predictive returns management and the introduction of AI-driven integration layers to eliminate data gaps between systems. Retailers should also embed sustainability metrics—such as CO₂e per parcel—so AI’s operational benefits align with regulatory and ESG expectations. 

Ultimately, the future of fulfilment will be decided not by how fast retailers can get parcels through the door, but by how consistently and efficiently they can deliver on their promises. As Rory O’Connor, CEO of Scurri, puts it, AI in fulfilment must work “invisibly, reliably and cost-neutrally across the supply chain.” It is this foundation of operational trust that will determine which retailers stay competitive in the decade ahead.

For more insight into AI and the future of fulfilment, download Scurri’s report.