The Agent Shopper: Brand Strategy for the Agentic Commerce Era

1. The Agent Shopper Has Arrived

The pace of change is striking. In December 2025, Klarna launched its Agentic Product Protocol—an open standard that made over 100 million products across twelve markets instantly discoverable by AI agents. Weeks later, Google announced the Universal Commerce Protocol (UCP—an open standard for agent-merchant interoperability) at the National Retail Federation conference, backed by more than twenty partners including Shopify, Walmart, Target, Mastercard, and Visa. OpenAI and Stripe co-developed the Agentic Commerce Protocol (ACP), an open standard for AI-enabled payments and commerce—though OpenAI’s initial attempt at direct checkout inside ChatGPT was quietly discontinued in early March 2026 after low transaction completion rates, a revealing signal that consumer purchasing behavior in agent environments is still forming. Perplexity expanded its AI shopping assistant to more than 5,000 merchants through PayPal integration. And in a revealing legal confrontation, a federal court in California granted Amazon a preliminary injunction against Perplexity in March 2026, finding that AI agents accessing Amazon’s systems on behalf of users did not constitute authorized access by the retailer’s standards.

These are not incremental upgrades. They represent the emergence of a new commercial architecture in which a growing share of buying decisions are evaluated, filtered, and sometimes executed by systems operating on behalf of human buyers—what this article calls the agent shopper. The scale projections are substantial: Gartner projects that 40% of enterprise applications will incorporate task-specific AI agents by the end of 2026, up from less than 5% in 2025. Adobe reported a 693% increase in traffic to U.S. retail sites from generative AI tools during the 2025 holiday season, with AI-referred shoppers converting at 31% higher rates and delivering 254% higher revenue per visit. Morgan Stanley estimates that agentic commerce could reach $190–385 billion in U.S. e-commerce spending by 2030. McKinsey frames the global opportunity at $3–5 trillion in AI-orchestrated commerce revenue by the same year. For brand leaders, these figures raise a question that decades of marketing theory did not prepare them for: what happens to brand strategy when the buyer is not a person but an agent shopper acting on a person’s behalf?

2. Who Builds the Agent Shoppers—and What Constitutes Their Intelligence

Before asking what brands should do, executives need to understand who is building the agent shoppers—because most of them are not being built by brands. Third parties define the agent’s intelligence; brands must respond.

The Builders

Five categories of firms are constructing the agent ecosystem. AI platforms and assistants (OpenAI, Google, Apple) control conversational interfaces and sit at the moment of intent capture. Google’s UCP connects Gemini and AI Mode in Search directly to merchant backends. These platforms are building the general-purpose agent that interprets what the user wants and orchestrates the path to fulfillment. Commerce infrastructure providers (Stripe, payment networks, identity services) build the transaction plumbing beneath the agent’s decisions—how purchases are authorized, routed, and completed. Retailers, marketplaces, and merchant ecosystems are building their own agent-compatible systems. Klarna’s protocol makes merchant catalogs machine-readable. Walmart deploys its own Sparky assistant while simultaneously integrating with Google’s Gemini and OpenAI’s ChatGPT—a deliberate strategy to secure decision slots across multiple agent ecosystems. Aggregators and data providers organize supply, comparison, and review data at scale, serving as normalized, trusted sources of commerce data for agent shoppers. And brands themselves—but selectively. The pattern of digital history suggests caution: consumers have consistently migrated to general-purpose tools that serve their interests. But some brands will build narrower agents around high-consideration categories, concierge service, replenishment, or post-purchase guidance.

What Constitutes Agent Intelligence

An AI shopping agent is not a blank search engine. It carries a set of core components that together constitute what we call agent intelligence—the full set of design parameters, functional capabilities, and learning inputs that determine how an agent shopper decides. Understanding these components is critical because the question of who influences each determines who holds power in agentic commerce.

The Brand Intelligence framework (Sun, 2026) provides the conceptual parallel. The Command Center—the brand’s algorithmic decision engine—generates real-time marketing decisions by optimizing who to engage, what to offer, when to act, where to reach them, and how to present it, all to maximize User Lifetime Value (ULTV) over the entire customer relationship. The consumer’s agent shopper solves a symmetric problem: who to buy from, what to buy, when to buy, where to transact, and how to execute—optimizing for whatever objective the user defines. The seven components fall into three layers—Architecture (the agent’s design parameters), Capabilities (its functional reach), and Learning (the inputs that make it smarter over time).

Together, these seven components constitute agent intelligence—the commercial mind of the agent shopper. The three-layer classification matters strategically: Architecture determines which brands can compete for the agent’s attention. Capabilities determine which brands the agent can actually evaluate and transact with. Learning determines which brands the agent develops loyalty to over time.