Amazon weighs strategy as AI shopping agents threaten to reshape online retail

Amazon is reassessing its approach to artificial-intelligence shopping agents as the technology begins to alter how consumers search for and buy products online. The company’s response ranges from blocking external bots to developing in-house tools, while simultaneously exploring potential partnerships that could safeguard its dominance in e-commerce.

Chief Executive Officer Andy Jassy signaled the shift in June 2023, telling employees that AI agents would soon become part of everyday tasks, including shopping. By October, he informed investors that Amazon was already in discussions with third-party providers and intended to collaborate where it made strategic sense. A recent corporate development job posting seeking a leader for “agentic commerce” underscores the priority the company now places on the issue.

The urgency is driven by a wave of AI products released over the past year by OpenAI, Google, Microsoft and startup Perplexity. These systems promise to handle the comparison shopping process, identify the best deals across the internet and complete purchases without sending users to a retailer’s website. Consulting firm McKinsey & Company projects that such “agentic commerce” could generate up to US$1 trillion in U.S. retail revenue by 2030.

Under this emerging model, AI platform operators collect a fee for each completed transaction. That arrangement threatens Amazon’s margins and its direct relationship with shoppers, according to industry analysts. Retailers risk paying what has been described as a toll to reach customers who would previously have navigated directly to their sites.

Defensive measures and legal action

Amazon’s first line of defense has been to restrict data access. The company recently modified its website code to block 47 different AI crawlers, preventing rivals from using Amazon’s product listings, customer reviews and sales rankings to train their models. Those reviews and rankings are viewed internally as critical, proprietary data that underpin the quality of Amazon’s own recommendations.

The company also took legal action in November, filing suit against Perplexity over an agent embedded in the startup’s Comet browser. Amazon alleges that Perplexity attempted to conceal its bots to continue scraping Amazon’s site without authorization. Perplexity has dismissed the lawsuit as an intimidation tactic.

Parallel investment in proprietary tools

While shielding its data, Amazon is increasing investment in its own AI-driven shopping experiences. The company introduced a chatbot named Rufus in February 2023 and has since extended its placement across more parts of the Amazon storefront. Recent tests allow Rufus to automatically purchase items for Prime members once a desired price is reached, and to generate customized shopping guides that reference products both on and off Amazon.

Another experimental feature, Buy For Me, enables users to select products from external websites without leaving the Amazon mobile app, positioning the company to remain involved even when a purchase originates elsewhere.

Market adoption still early

Research by Morgan Stanley estimates that nearly half of U.S. shoppers could employ AI agents by 2030, adding roughly US$115 billion to domestic e-commerce sales. At present, only a small single-digit percentage of consumers begin a purchase with AI assistance, though 40% to 50% already use the technology for product research.

Imagem: Internet

Conversions remain inconsistent. Early tests reveal technical limitations: one industry observer attempting to buy a cable-knit sweater through Perplexity’s Instant Buy tool encountered repeated errors, and a separate search on ChatGPT surfaced a Breville espresso maker that displayed an unrelated garden rake image when clicked. Despite growing traffic from chatbots during the 2023 holiday season, traditional web search still delivers superior conversion rates and revenue per session.

Competitors opt for a mixed strategy

Several large retailers are striking a balance between collaboration and competition with AI platforms. Walmart, Shopify and others have announced agreements that allow select agents to tap their catalogs while setting rules to protect sensitive data. Shopify, for example, has committed to building its own infrastructure layers while simultaneously providing controlled access to external developers.

Amazon is experimenting on a smaller scale through subsidiaries. Zappos, Shopbop and deal site Woot do not currently block AI agents, suggesting the parent company may be using these brands to test outcomes without exposing its main marketplace.

A high-stakes calculation

Industry consultants describe Amazon’s position as a “leader’s dilemma.” With a dominant share of U.S. online retail, the company stands to lose more than most if shoppers migrate to third-party agents. Allowing unfettered access could erode proprietary advantages, yet refusing engagement risks driving consumers toward rival platforms that integrate seamlessly with AI tools.

For now, Amazon continues to refine its protective barriers, litigate perceived overreach and accelerate development of its own conversational commerce products, all while evaluating which partnerships might secure a role in the rapidly evolving landscape of AI-mediated shopping.

Crédito da imagem: Getty Images