AI agents rewrite the rules of online shopping—but trust remains the final checkout

The technological backbone of agentic commerce is rapidly advancing through innovative protocols and AI tools that enable autonomous agent-to-agent marketing and purchasing. Google's Universal Commerce Protocol (UCP) and OpenAI-Stripe's Agentic Commerce Protocol (ACP), adopted by major players like Shopify, Etsy, and Walmart, empower AI agents to discover products, access real-time inventory and pricing, and complete transactions without human intervention. This foundational infrastructure is complemented by OpenAI's launch of ChatGPT Ads Manager, which integrates commerce and advertising directly into large language models (LLMs), marking a pivotal shift from traditional ecommerce toward AI-driven, agentic interactions.

OpenAI’s recent integration of product feed support into ChatGPT Ads Manager exemplifies the move toward scalable, automated ecommerce advertising within LLMs. By enabling retailers to connect catalogs of up to a million SKUs and automatically generate ad units from structured product data, OpenAI lowers barriers for brands familiar with platforms like Google Shopping. This strategic pivot away from direct checkout inside ChatGPT toward capturing advertising budgets reflects a broader industry trend of blending organic product discovery with paid campaigns inside AI environments, thereby reshaping how retailers engage consumers at scale.

The traditional merchant website is increasingly sidelined as consumers initiate and often complete shopping journeys directly within LLMs, echoing the 'zero click' phenomenon seen in content consumption. While LLMs excel at product discovery, operational limitations currently prevent full transaction and fulfillment capabilities, with 76-77% of consumers still preferring to finalize purchases on brand sites. Nonetheless, platforms like PayPal leverage vast transaction and social sharing data across millions of users to build robust AI-powered commerce graphs, positioning themselves to power the next generation of agentic commerce within these AI interfaces.

Embedding commerce directly into LLMs represents a critical monetization strategy as platforms seek to move beyond mere referral traffic. While integrations by Shopify and Stripe into AI environments have yet to reach mass adoption, the potential to plug product data and shopping experiences into LLMs promises to simplify user access and unlock new revenue streams akin to social media marketplaces. This evolution signals a fundamental shift in ecommerce user interfaces, where AI agents autonomously manage routine purchases, allowing consumers to focus on preferred categories—a transformation likened by Naveen of Glance to the late-1990s Google revolution and underscored by Gary Vee’s call for brands to optimize for AI agent discoverability or risk obsolescence.

AI-referred traffic is fundamentally transforming consumer shopping behavior by shifting the journey from broad browsing to direct, conviction-building interactions at the product detail page level. Contrary to fears that AI might reduce engagement, data from Adobe and Sensor Tower reveal that AI-referred visitors spend more time, view more pages, and convert at significantly higher rates—up to 58% compared to a 21% baseline on Amazon with multiple AI queries—highlighting a deeper, more deliberate decision-making process rather than mere efficiency-driven shopping. This evolution reflects a move toward goal- and constraint-based discovery, where consumers rely on AI to translate pain points into product recommendations, reshaping how brands must position themselves within AI-driven consideration sets. [1, 2, 3, 10, 14]

The rise of AI agents acting autonomously in purchase decisions is redefining brand loyalty and marketing strategies, as consumers increasingly delegate shopping to synthetic agents that follow preset rules but remain influenced by human preferences and marketing touchpoints. Gary Vaynerchuk emphasizes that while agentic commerce automates repeat purchases for convenience, consumers still intervene selectively for products they care about, resetting AI preferences when swayed by compelling ads or trusted brands. This dynamic creates a volatile loyalty landscape where brands must build authentic trust and clear positioning, as AI assistants—like Dust’s 300,000 active AI buyers—scrutinize every detail before shortlisting products, making trust signals and structured data critical for visibility. [4, 5, 6, 17, 18, 23, 26]

Marketing channels are undergoing a significant rebalance as AI-driven brand discovery and agentic commerce rise, prompting brands to double down on first-party data, AI-optimized content, and community engagement to maintain relevance amid automated purchasing behaviors. This shift also paradoxically fuels a resurgence of analog, experiential marketing, as consumers seek tangible, authentic interactions to counterbalance AI’s convenience and selective referral patterns. Industry leaders note that while AI referral traffic reduces reliance on paid ads—given higher organic conversion rates in platforms like ChatGPT and Gemini—investments in real-world activations, such as concerts and in-store experiences, are growing despite measurement challenges, underscoring a 'barbell syndrome' in consumer behavior. [7, 8, 20, 21, 27, 32, 34, 35]

In the AI-driven commerce landscape, brand building emerges as a critical differentiator that extends beyond performance marketing to create trust, preference, and memory, which are essential for sustained success amid functionally similar large language models. Experts like Lindsay Kaplan argue that strong brand equity not only lowers acquisition costs and improves retention but also sharpens conversion and amplifies word of mouth, giving companies a competitive edge as AI models themselves lack inherent brand equity. Consequently, prioritizing brand development early—potentially even before scaling engineering teams—is becoming a strategic imperative for AI startups and established brands alike to thrive in this evolving ecosystem. [36, 37, 38, 39]

The strategic competition among major platforms in AI-driven agentic commerce is defined by fundamentally divergent approaches to controlling the transaction layer. Amazon pursues a vertically integrated model, exemplified by its transformation of the Alexa AI shopping assistant into a high-conversion embedded conversational agent that now powers about 40% of holiday transactions and is being offered as a platform service to other retailers. This closed ecosystem approach includes blocking external AI crawlers and aggressively protecting its marketplace data, aiming to own the entire customer relationship and commerce operating system much like AWS dominates cloud infrastructure. In contrast, Google champions an open, horizontal infrastructure strategy, developing universal protocols such as the Universal Commerce Protocol and Agent Payments Protocol to enable seamless interoperability across AI agents and merchants, spanning platforms like Search, Gemini, YouTube, and Gmail with access to 60 billion product listings. This openness is designed to foster a broad ecosystem where AI agents autonomously negotiate and personalize transactions, reflecting Salim’s assertion that “the retail war is no longer about shelf space. It’s about agent preferences.” Meanwhile, Stripe positions itself as the indispensable payment backbone amid this fragmentation, processing $1.9 trillion annually and building an agent commerce stack to handle the complex, multi-currency, and microtransaction-heavy nature of agentic commerce that current payment infrastructures struggle to support.

Consumer adoption remains a formidable hurdle for agentic commerce, as there is a notable disconnect between the industry's ambitious vision of fully autonomous shopping agents and the current willingness of users to entrust AI with complete purchasing decisions. Scott Friend and others emphasize that regardless of technological advancements, success hinges on consumer buy-in, yet many vendors have diluted the term 'agentic commerce' by labeling offerings that lack true autonomy, thereby fostering confusion and skepticism among potential users.

The evolution of agentic commerce is unfolding incrementally, with Stripe’s 2025 annual letter framing the landscape into five levels of autonomy; however, the industry largely remains at the nascent edges of levels 1 and 2, far from the fully autonomous stages that define true agentic commerce. This gradual progression reflects both structural barriers and behavioral inertia, as consumers continue to favor familiar interfaces and express reluctance toward radically new, immersive experiences that AI-driven commerce platforms propose.

Underlying the slow adoption are deeper concerns about control and transparency as commerce increasingly migrates into AI ecosystems dominated by a handful of major companies. Critics warn that the open web is effectively guarded by four gatekeepers who monetize access, raising fears that without dedicated apps, consumers’ shopping options might be limited or manipulated within chat interfaces, thereby eroding trust and complicating the transition to agentic commerce.