Agentic Commerce: Turning Retail Intelligence Into Real-Time Decisions

The Retail Decision Gap

Retailers have never had more data. Point-of-sale transactions, loyalty programs, delivery telemetry, and support tickets stream in continuously. What most organizations lack is the ability to turn that stream into action.

Across customer experience, merchandising, and supply chain, decisions still depend on manual review. A pricing error is left unnoticed for a shift. A stockout goes unaddressed until a customer complains. A service ticket queues behind a hundred others that could have been resolved automatically. This results in lost sales, inconsistent service, and slow response when conditions change.

Agentic commerce solves this problem by putting the decision inside the system itself. AI agents sense a signal, weigh the options, act, and then report what they did. McKinsey estimates that even under moderate scenarios, AI agents could mediate $3 trillion to $5 trillion of global consumer commerce by 2030.¹ That change is already visible in specific, high-impact deployments across retail and logistics.

Intelligent Basket Building in Grocery

A leading North American grocery retailer is putting this into practice, investing in AI-powered shopping assistants to change how customers complete their purchase journey. Rather than making the customer work through the funnel step by step, the system takes a stated intent, such as meal planning or restocking staples, and turns it into a finished, order-ready basket. It does this by:

• Recommending related items based on the stated intent
• Substituting products according to availability and preference
• Checking the basket against delivery windows and fulfillment constraints

The pattern is showing up across the industry. Albertsons Companies launched a similar AI-powered shopping assistant across its store banners in December 2025, saying the tool can cut a typical online grocery trip, about 46 minutes according to Capital One Shopping research, down to roughly four.² The shift moves basket construction from the shopper to the system and treats speed and accuracy as the product, not just the interface.

Agentic commerce moves basket-building from the shopper to the system, redefining what convenience means in grocery retail.

AI-Led Insight and Merchandising Optimization

Retailers are applying the same logic to merchandising, where AI agents continuously read demand signals, customer behavior, and inventory data instead of waiting for a weekly planning cycle.

McKinsey's Global Merchant Survey found that merchants who adopt agentic AI could reclaim up to 40% of their time from manual reporting and reconciliation work, redirecting it toward strategy and vendor negotiation.³ In retail environments where agentic AI is already delivering results, it detects demand shifts in real time, adjusts assortments and promotions, and keeps inventory closer to actual demand. These early applications show how AI can move merchandising from periodic analysis to continuous decision-making, making it a practical starting point for broader adoption across the business.

From Reactive Support to Proactive Service Operations

A large North American energy infrastructure company re-engineered its service desk with AI-led automation, shifting from a reactive model to a proactive one. Instead of waiting for a ticket, the system watches for early signals of trouble across end-user environments and flags them before they escalate.

This mirrors a broader pattern. Deloitte reports that AI adoption in customer service climbed from 46% of organizations in 2023 to 61% in 2025, with the clearest gains showing up in resolution time, customer satisfaction, and cost.⁴ Deloitte's research on IT operations describes a similar transition, with generative and agentic AI moving service desks from ticket response toward self-healing and proactive problem management.⁵

Within a year, the energy company saw a significant rise in customer satisfaction and a clearer view into its own operations, the kind of improvement that is hard to reach through headcount alone.

The strongest agentic systems do not add a feature. They close the loop between signal, decision, and action.

Core Systems Modernization in Transportation

A freight and logistics network leader is modernizing decades-old core systems without disrupting daily operations. This is a common starting point. McKinsey estimates that 70% of the software running Fortune 500 companies was built 20 years ago or more.⁶

Rather than a single high-risk replatforming, the company is layering AI-driven automation onto high-volume workflows such as billing and invoicing, aiming for processing that needs less manual review. McKinsey's work on legacy modernization suggests agentic AI can accelerate this kind of transformation by 40% to 50%, while also re-engineering the business processes tied to the old systems, not just refreshing the code around them.⁶

The approach trades speed for stability and lets the enterprise build new digital capability in modules, without betting the business on one large migration.

Optimizing Last-mile Delivery Decisions

Parcel delivery networks are increasingly applying AI to optimize last-mile routing. Real-time analysis of conditions such as traffic, delivery density, and route dependencies enables these systems to minimize transit time, reduce fuel and operational costs, and improve delivery predictability.

UPS offers a well-documented example of what this looks like at scale. Its ORION routing system cuts roughly 100 million miles and 10 million gallons of fuel every year, worth an estimated $300 million to $400 million in annual savings once fully deployed.^7,8 A few miles saved per driver each day can translate into meaningful savings when applied across tens of thousands of routes.

The impact extends beyond cost. It reaches both operational efficiency and the customer experience.

Solving the Core Retail Problem

What ties these five examples together is a transition from generating insight to executing decisions. Traditionally, insight required a person to interpret it and decide what to do next. Agentic systems compress that cycle. They sense a signal, interpret the context, and act across systems, without waiting for the next planning meeting.

This addresses long-standing retail challenges: fragmented decisions, slow responses to demand and supply changes, underused real-time signals, and disconnected customer and supply chain systems.

The Closed-loop Advantage

The strongest implementations share a closed-loop operating model. The system receives signals from customer interactions, frontline input, and enterprise systems. AI models analyze those signals and rank them by priority in real time. Actions execute across merchandising, inventory, and fulfillment while outcomes are monitored continuously and fed back into the next decision. This creates a system that improves with use, while keeping customer demand and operational execution aligned.

From Digital Commerce to Autonomous Commerce

The next phase of retail transformation will hinge on how decisions are made, supported by connected data, AI, and real-time intelligence.

As agentic systems mature, more enterprises will run environments where multiple AI agents operate across the value chain, managing decisions as they happen rather than on a fixed schedule. Gartner projects that 40% of enterprise applications will carry task-specific AI agents by the end of 2026, up from less than 5% in 2025, evidence that this shift is already underway.⁹

For customers, that means more predictable, less friction-heavy service, while operations gain systems that respond to changing conditions and enable real-time action.

Building the Foundation

None of these use cases scale into an enterprise-wide capability without groundwork. Retailers preparing for agentic commerce need:

• data that different systems can actually share and trust
• architecture built to integrate rather than bolt on
• governance that defines how AI-led decisions get made and reviewed
• alignment across functions, so a decision made by one agent does not create a problem for another

Without that foundation, AI is just a set of isolated tools instead of a transformation engine. McKinsey's merchant survey backs this up directly: 61% of merchants say their organization is not well prepared to scale AI across the function, and the gap is rarely about the model.³ It is about the data, process, and governance underneath it.

Scaling agentic commerce is a data and governance problem before it is an AI problem.

A Shift That is Already Underway

Agentic commerce is not a future concept. It is running today in targeted deployments across grocery, merchandising, service, transportation, and last-mile delivery, with each use case demonstrating a different part of the model.

The organizations seeing the most value are embedding AI into how decisions are made. They use it to sense signals, prioritize actions, and respond in real time, bringing decisions closer to the point where they have the greatest impact.

The opportunity now is to move from pilot to scale, and from insight to outcome. That starts with the same groundwork behind every example in this piece: clean data, integrated architecture, and governance that keeps pace with what the agents are doing.