The future of e-commerce is AI-to-AI

AI agents are ushering in a new era of e-commerce, transforming the foundational principles of online payments. According to Sherman Jiang, a seasoned product executive in the payments industry, the shift from traditional machine-to-machine (M2M) payments to truly autonomous robot-to-robot (R2R) payments is already underway. Unlike M2M systems that rely on simple, rule-based instructions, R2R payments leverage agentic AI, allowing them to reason, learn, and act with purpose. For example, instead of merely re-ordering a previously purchased item, a personal AI agent could analyze a user’s needs, check real-time stock levels, find the best-rated product at the most competitive price, and then autonomously complete the transaction. As more consumers adopt these intelligent agents, payments and e-commerce companies must rapidly adapt to the new demands of this sophisticated technology.

The growing consumer demand for agent-assisted shopping highlights the urgency of this transformation. A July 2025 survey by Boston Consulting Group (BCG) found that an overwhelming 81% of US consumers expect to use agentic AI for their shopping needs. BCG projects that agent-assisted spending could exceed one trillion US dollars, representing about half of today’s total e-commerce expenditure. This massive shift underscores a critical need for payments companies to align their strategies with this evolving consumer behavior.

This new commerce paradigm presents both significant opportunities and complex challenges for the payments industry. For decades, the primary heuristic for fraud prevention was that “bot-like” behavior, characterized by rapid, high-frequency transactions, was a clear red flag. However, with the rise of agentic commerce, this behavior is now a feature of a legitimate user. AI agents operate at machine speed and scale, necessitating a complete overhaul of traditional fraud prevention strategies. The long-standing “Know Your Customer” (KYC) model is becoming insufficient, giving way to a new imperative: “Know Your Agent” (KYA). Fraud experts are already seeing an uptick in AI-enabled fraud, with one study revealing that 65% of experts have already been targeted. The key challenge lies in developing new infrastructure capable of distinguishing legitimate AI from malicious ones. This will require new systems that can authenticate both the user’s identity and the agent’s identity, securely linking the two.

The adoption of agentic AI also introduces novel regulatory and liability questions. Traditional liability laws are often based on the intent of the actor. However, an AI agent lacks intent, creating a gray area when it makes an erroneous purchase that technically follows its authorization but goes against the user’s ultimate wishes. This issue is compounded by the “agentic loyalty problem,” where an agent might be programmed to prioritize the platform that deployed it rather than the user. For instance, an agent could choose a more expensive item from a partner company even when a cheaper, more suitable option is available elsewhere.

Furthermore, the rise of agentic commerce reveals a fundamental incompatibility between the new AI-driven economy and our existing financial infrastructure. Current payment systems, such as credit cards and digital wallets, are designed for human-centric, high-value, and low-frequency transactions. In contrast, autonomous agents increasingly require a “pay-per-use” model to handle micro-transactions for individual API calls, data, or computing resources. This has created a demand for purpose-built payment rails that can securely and cost-effectively handle a high volume of small, fast transactions between AI robots. Jiang suggests that blockchain technology may be the answer. Micropayment systems built on blockchain are emerging as foundational infrastructure for this new economy, enabling agents to buy data or rent computing power without any human intervention. With the use of smart contracts, payments can be automatically triggered and executed the moment predefined conditions are met.

The emergence of AI agents is not just an incremental improvement to e-commerce but a fundamental re-engineering of the entire digital payments ecosystem. As AI-to-AI transactions become the norm, the industry must move beyond outdated fraud models, address complex liability issues, and build new payment infrastructure capable of handling a world of autonomous, high-frequency micropayments. The companies that successfully navigate these challenges and embrace the “Know Your Agent” paradigm will be the ones to thrive in this new era of agentic commerce.