Amazon’s expansion into agent-led commerce, through initiatives like Shop Direct, signals a significant shift in how online purchasing decisions are made. In this emerging model, AI-powered agents embedded in large merchant platforms act on behalf of consumers to discover, compare, and purchase products across the internet as well as within their own supply chains. 

As merchants think about engaging with agentic commerce, they will need to recognise the roles of embedded AI alongside ChatGPT and Claude.  

Opportunity and strategic blind spots

Merchants have embraced AI ecosystems as a path to find and convert new customers and better meet the needs of existing ones. Standards such as Google’s Universal Commerce Protocol enable retailers to integrate AI-driven journeys into their storefronts and checkouts.

However, this shift introduces a critical risk. By opening their ecosystems to AI agents, merchants may also be allowing competitors’ embedded agents control over the purchasing journey, where control over the transaction and associated data has traditionally been a core advantage.

Amazon’s move highlights how new intermediaries are seeking to position themselves between the merchant and the customer, potentially capturing both the customer relationship as well as valuable data insights.

Defining the rules of engagement

As agent-led commerce becomes more prevalent, merchants must establish clear rules governing which AI agents are permitted to operate within their environments. Without defined guardrails, embedded agents, including those acting on behalf of competitors, could influence product selection, pricing visibility, and checkout flows.

The decisions merchants make about access and control will directly shape both the customer experience and their long-term strategic position.

Emerging risks and agentic fraud

Another growing concern is the rise of agentic fraud. In AI-mediated environments, interactions that appear to originate from legitimate sources may in fact be driven by malicious actors leveraging standard consumer agents.

These actors can manipulate purchasing behavior, generate fraudulent transactions, or exploit system vulnerabilities at scale. Traditional fraud controls, designed for human-driven interactions, are not sufficient in this context.

Merchants will need to develop new capabilities to identify, authenticate, and monitor AI agents even from trusted sources to ensure that trust, security, and transaction integrity are maintained.

Retaining control and insight

Agent-led commerce also has significant implications for data ownership and visibility. When AI agents mediate transactions, critical customer data may be abstracted away from the merchant, weakening the feedback loops that inform pricing, marketing, and product strategy.

Maintaining visibility into these interactions is essential. Without it, merchants risk losing not only transactional control but also the insight needed to compete effectively.

Preparing for the future

The evolution toward agent-led commerce will continue to accelerate, with AI agents taking on increasingly complex purchasing journeys. This will require new approaches to oversight, compliance, and risk management, as well as continuous adaptation to emerging technologies.

The shift is already underway. Amazon’s move makes clear that control of the customer journey, including the point of transaction, is now contested territory.

Merchants must define clear rules for AI participation, invest in agent-specific risk controls, and actively manage their ecosystems in order to be best positioned to compete. A failure to do so presents a significant strategic opportunity for competitors who are already investing to take control of customers’ purchasing journeys and their data.