Groceryshop opens today in Las Vegas with more than 4,000 grocery and CPG leaders gathering to discuss the technologies, business models and consumer behaviors shaping retail. The agenda is filled with sessions on artificial intelligence, agentic shopping, retail media, personalization, ecommerce and supply-chain productivity.

All are important. But the agenda is largely missing what I believe is the most consequential emerging topic in retail: autonomous commerce.

I do not mean autonomous delivery. Autonomous delivery is one component. Autonomous commerce is the complete system through which demand, inventory, fulfillment, transportation, site access, customer authentication, secure handoff, payment and returns are coordinated with minimal human intervention.

Agentic AI may decide what a customer should buy. Autonomous commerce determines whether the product can be picked, staged, transported, authenticated and delivered profitably. One is intelligence. The other is execution.

Retail has spent several years obsessing over the intelligence layer. The next competitive battle will be fought over the execution layer.

I have stated since 2020 that the biggest impact AI will have on retailers is the acceleration of velocity across the enterprise: product development, procurement, supplier relations, manufacturing, online and store assortment, inventory, pricing and promotions, sales, logistics and last-mile delivery. AI will accelerate everything.

The brutal truth, however, is that retailers are utilizing supply chain and logistics models from the 80s, 90s, 2000s and 2010s. Retailers need a logistics model designed for the modern era. Without it, AI will only enable retailers to do bad things faster.

This is why autonomous commerce is a necessity.

Groceryshop Is Looking at the Pieces Instead of the System

The omission is especially noticeable because several pieces of autonomous commerce are already present at Groceryshop.

The agenda includes multiple sessions on agentic shopping and AI-powered commerce. Amazon is sponsoring a private session on autonomous operations. Nash and Ahold Delhaize USA are presenting a case study titled “Fulfillment Without the Fulfillment Centers.” That Nash session is important. It may be one of the most strategically consequential discussions at the event.

But these subjects are still being presented as separate categories. Agentic shopping is discussed as a customer experience. Fulfillment is discussed as an operational function. Autonomous operations are discussed inside the enterprise. Last-mile delivery is treated as a transportation problem.

They are converging into one system.

The retail industry needs to stop asking how AI will improve shopping and start asking a more important question: Who will own the execution layer of commerce?

What Autonomous Commerce Actually Means

Autonomous commerce begins when an AI agent, consumer or retailer generates demand. The system identifies available inventory, selects the optimal fulfillment location, assigns the work required to prepare the order, chooses the best vehicle, reserves the correct pickup position, authenticates the vehicle, manages loading, monitors the trip, secures the handoff and processes any return.

The goal is not to eliminate every person from commerce. Exceptions, damaged goods, substitutions, inaccessible locations and food-safety issues will still require human judgment. The goal is to remove continuous human coordination from normal transactions and reserve people for the exceptions where they add value.

This distinction matters. A vehicle that can drive itself is not an autonomous-commerce network. A robot that can pick groceries is not an autonomous-commerce network. An AI shopping assistant is not an autonomous-commerce network.

Autonomy emerges only when the components operate as a coordinated system.

The Emerging Autonomous Commerce Stack

Three companies help illustrate how the architecture can take shape: Tesla, Nash and AutoLane.

Tesla represents the mobility layer. Its vehicles, software, charging infrastructure and long-term autonomy strategy create the potential for a large distributed network capable of transporting people, groceries, restaurant orders, pharmacy products, parcels and returns.

Nash represents the orchestration layer. It can connect retailers with fulfillment resources and multiple delivery options while managing dispatch, routing and execution. Its model is important because the future will not depend on a single labor source or vehicle type. Retail associates, independent drivers, third-party fleets, customer-owned vehicles and autonomous vehicles may all participate in the same network.

Nash is also important because it is one of the few credible platforms capable of giving retailers an alternative to structural dependence on Instacart, DoorDash and Uber Eats. I strongly encourage Groceryshop attendees to familiarize themselves with Nash and contact its CEO, Mahmoud Ghulman, to learn more about how Nash can transform their logistics and store operations.

AutoLane represents the physical and geospatial layer. Autonomous vehicles must know more than how to reach a street address. They must know where to enter a property, where to wait, which stall to use, whether that stall is available, how to authenticate, when a trunk can be opened and how the retailer should be notified.

AutoLane combines smart curbside infrastructure with cloud orchestration. Its platform can authenticate arriving vehicles, assign stalls, manage dwell time, coordinate secure trunk access and provide retailers and property owners with real-time operating visibility. This addresses what Ben Seidl, AutoLane’s founder, correctly describes as the last 50 feet of autonomous commerce.

I believe AutoLane is well positioned to become a fleet-management and site-orchestration layer for Tesla’s eventual entry into retail logistics. My expectation is that AutoLane and Nash can leverage their retail relationships to enable Tesla vehicles to operate within geofenced delivery networks surrounding selected stores. If executed successfully, this model can transform retail.

I was the first person to write an article about AutoLane. In July 2025, I knew that AutoLane was going to change commerce, and I predicted that the company would soon attract interest from the largest investors and global corporations. It’s happened. AutoLane is truly an exceptional company.

A fourth component is also necessary: secure cargo infrastructure. Temperature-controlled vehicle lockers can create protected compartments for frozen food, refrigerated groceries, prepared meals, pharmacy orders, parcels and returns. The locker is not simply storage. It provides chain of custody and allows a vehicle to carry multiple orders or serve both passenger and commerce demand without exposing one customer’s products to another. AutoLane has designed and manufactures a temperature-controlled locker system for selected Tesla vehicles that can also be customized for other vehicle platforms.

No individual company owns the complete stack. That is the point. The emerging opportunity is to connect mobility, orchestration, property infrastructure and secure custody into a common execution layer.

Why Austin Matters

This is not a forecast about a distant future.

Ben Seidl has confirmed to me that AutoLane will expand its Austin autonomous delivery service to include multiple leading retailers and quick-service restaurants before the end of 2026.

That is the type of development the retail industry should be examining closely. Autonomous commerce will not emerge nationally on day one. It will develop through dense, geofenced operating networks where retailers, restaurants, vehicles and consumers interact repeatedly in the same market.

Austin has the ingredients to become one of the first cities where autonomous commerce operates as an ecosystem rather than as a collection of disconnected demonstrations. The strategic value will come from the operating data: vehicle arrival patterns, loading time, stall utilization, handoff reliability, order density, exception rates and customer adoption.

The companies that learn fastest inside these early networks will define the operating standards adopted elsewhere.

Early operating evidence indicates that many customers will choose autonomous delivery when it offers greater security, predictable service and freedom from tipping. Austin will help determine how strong that preference remains as autonomous delivery expands across retailers, restaurants and customer segments.

I’ve written multiple articles since 2018 in which I stated that consumers are suffering from “delivery fatigue.” This is a condition fueled by third-party delivery platforms, primarily DoorDash and Uber Eats, whose delivery models cannot consistently guarantee food integrity, tamper protection or an auditable chain of custody once an order leaves the restaurant. I coined the phrase “The DoorDash Buffet” in response to reported incidents of delivery drivers eating customers’ food. Customers are also fatigued by drivers who create uncomfortable confrontations over tips.

There is also the safety issue. Incidents involving harassment, threats and assault by delivery drivers have heightened consumer concerns about personal safety.

My argument is this: Customers will choose safety over convenience. When given a credible alternative to an unwanted interaction with a delivery driver, many customers will choose autonomous delivery. Retail analysts and investors who assume customers will reject autonomous delivery because they must go outside to retrieve their delivered food or groceries are underestimating the value consumers place on security and control. When price and service levels are comparable, a meaningful share of customers—especially those concerned about unwanted contact—will choose the option that provides greater safety.

The Most Important Idea Is Immediate Commerce

The industry frequently makes the mistake of treating autonomous commerce as something that cannot begin until every vehicle is driverless. That is unnecessary and strategically dangerous.

I use the term Immediate Commerce to describe a network that begins with the vehicles, labor and regulatory permissions available today and becomes progressively more autonomous over time.

A human-driven vehicle can execute an order today using the same retailer integrations, dispatch logic, parking instructions, authentication protocols and secure cargo systems that a fully autonomous vehicle will use later. The driver can gradually exit the workflow as autonomy improves and regulators approve broader deployment.

Immediate Commerce does not eliminate every driver on day one. It can, however, eliminate the unwanted interaction. Secure lockers, authenticated handoffs and contactless retrieval allow a human-driven vehicle to use the same protected chain-of-custody model that will eventually operate without a driver.

This approach allows retailers and their partners to build demand density, integrations, operating procedures, customer trust and unit-economic knowledge before universal autonomous driving arrives. It also avoids the worst possible strategy: waiting for the vehicle technology to become perfect and only then beginning to design the commerce network around it.

Autonomy should be a migration, not a starting gun.

Why Grocery Is the Ideal Proving Ground

Grocery is one of the hardest retail categories to execute and therefore one of the best categories in which to build an autonomous-commerce platform.

Demand is frequent and local. Stores already hold inventory close to consumers. Delivery radiuses can be compact. Orders contain products requiring ambient, refrigerated and frozen temperatures. Customers have little tolerance for late delivery, poor substitutions or broken chain of custody.

A two-to-five-mile operating zone around a supermarket, mass merchant or restaurant cluster can support far more than grocery delivery. The same network can transport prepared meals, pharmacy items, retail purchases, parcels and returns. Vehicles can serve passenger and cargo demand at different times of day, raising utilization and distributing ownership costs across more transactions.

This is where the economics can change. Today, most delivery networks dedicate a person and a trip to a single customer order. Autonomous commerce allows retailers to redesign the transaction around route density, shared capacity, secure compartments and very low marginal transportation cost. Nash can orchestrate the myriad of tasks required throughout the process.

The objective should be delivery at a cost low enough to become an ordinary part of commerce rather than a premium service dependent on high fees and tips. A target such as $2 per order should be treated as a design requirement to be proven through pilots—not as an assumption. That discipline would force retailers to measure picking cost, loading time, vehicle utilization, dwell time, failed handoffs, insurance, cleaning, refrigeration and remote intervention.

Retailers Are Outsourcing Too Much

For years, retailers have relied on Instacart, DoorDash, Uber Eats and other platforms to close gaps in digital demand and last-mile execution. Those relationships created convenience and speed, but they can also transfer margin, customer ownership, transaction data and service control to a third party.

Autonomous commerce creates an alternative. It does not require retailers to eliminate every third-party platform. It gives them the ability to control more of the execution system and choose among labor, carriers and vehicles transaction by transaction.

Nash is strategically important for this reason. Its potential is larger than delivery aggregation. Nash can become the control plane connecting retailer demand with store fulfillment, independent drivers, existing carriers and eventually autonomous fleets. Its Load-and-Go model also challenges the assumption that ecommerce fulfillment must always be performed inside a dedicated fulfillment center.

AutoLane extends that control plane across the private-property boundary. Retailers cannot operate autonomous fleets reliably if vehicles reach a parking lot and then depend on improvised instructions, occupied pickup spaces and manual phone calls. Parking, staging and authentication must become machine-readable infrastructure.

The retailer that controls these capabilities controls more of its economics and customer relationship. The retailer that ignores them risks replacing dependence on human delivery marketplaces with dependence on autonomous delivery marketplaces.

The Store Must Become a Node

The grocery store of the future will still serve customers who want to shop inside. But it must also function as a high-velocity node in a distributed fulfillment network.

That requires more than adding pickup spaces. Stores will need redesigned staging areas, rapid loading processes, accurate inventory, machine-readable site instructions, secure vehicle access and clear exception-handling rules. Micro-fulfillment automation may play a role at high-volume locations. Store associates, flexible picking models and mobile automation may be more economical elsewhere. Some grocers should selectively attach micro-fulfillment systems to high-volume stores or operate standalone MFCs serving clusters of nearby locations. These facilities can interact directly with autonomous vehicles.

The winning architecture will not impose one fulfillment model on every store. It will select the right combination of people, automation and inventory for each market and order.

This is why autonomous commerce belongs at Groceryshop. It connects store operations, ecommerce, real estate, supply chain, labor, AI, customer experience and transportation. It is not a last-mile topic. It is a retail operating-model topic.

What Groceryshop Should Put on the Agenda

Next year’s event should include a main-stage session titled “Autonomous Commerce: The Execution Layer Agentic AI Is Missing.” The panel should include a retailer, Nash, AutoLane, an autonomous-vehicle platform and a secure-cargo or mobile-fulfillment technology company.

The session should not consist of vague predictions. It should follow one order from beginning to end: an agent or consumer places the order; the system selects inventory and fulfillment; the store picks and stages it; a vehicle is assigned; the property authenticates and positions the vehicle; the order is loaded into a secure temperature-controlled compartment; the customer completes an authenticated handoff; and the same network manages a return.

That demonstration would reveal the industry’s real challenge. The hard problem is not teaching software to recommend cereal. It is coordinating hundreds of physical and digital decisions so the cereal, and everything else in the basket, arrives safely, profitably and on time.

The Next Retail Platform Will Execute

Groceryshop describes itself as the place where the future of everyday retail gets decided. If that is the mission, autonomous commerce must become part of the discussion.

Agentic AI will generate demand, make recommendations and increasingly initiate transactions. But intelligence without physical execution is incomplete. Every promise made by an AI agent eventually collides with inventory, labor, a loading area, a vehicle, a parking lot and a customer waiting for an order.

Tesla, Nash and AutoLane illustrate the architecture. Secure lockers close the chain-of-custody gap. Store-based fulfillment provides local inventory. Immediate Commerce provides a path to begin now rather than waiting for a perfectly autonomous future.

The companies that connect these elements will not merely improve delivery. They will build the execution layer of commerce.

That is the emerging topic Groceryshop is missing. It may also become the most important platform battle in retail.