AI-driven e-commerce will progress in stages. It will start with human-prompted purchases, then move to agents proactively suggesting items, and ultimately culminate in autonomous agent-to-agent transactions based on predefined budgets and inferred needs, requiring no human intervention.

Rather than relying entirely on user subscriptions or ad auctions, AI shopping agents like Meta's Muse are poised to capture merchant revenue via payment processing. While standard card rails charge low percentages, agentic checkout channels could demand processing fees of 5% to 7% directly inside apps. This allows agent providers to take a substantial cut from existing merchants by controlling the final conversion channel alongside ad spend.

Users increasingly expect to complete purchases within AI chat interfaces. This trend, called "agentic commerce," requires new tools like Stripe Projects that allow agents to programmatically sign up for and pay for services like Vercel or Cloudflare to complete end-to-end tasks for users.

Microtransactions have historically failed because the 'mental load' of a human deciding on a small payment outweighs the value. AI agents, which can scrutinize tiny decisions without cognitive cost, can enable a new economy of per-use payments for data, content, and APIs.

While human payment habits are entrenched (e.g., Visa), AI agents have no such loyalty. They will ruthlessly optimize for cost and efficiency, making near-free, programmable stablecoin transactions the default choice for the 99%+ of future transactions they will conduct, sidestepping legacy financial infrastructure.

Enabling AI agents to purchase goods autonomously creates entirely new consumer demand rather than merely boosting conversion efficiency for existing users. People delegate tasks and initiate transactions they previously avoided due to manual friction, expanding commercial volume. Consequently, agentic commerce enlarges the total addressable market as software entities open dedicated user accounts.

Consumers will be slow to trust AI with subjective shopping decisions (e.g., buying clothes). However, they will more quickly adopt AI agents to handle the objective task of payment optimization—choosing the right card or financing—making it the first major frontier for agentic commerce.

The idea that AI agents require entirely new financial infrastructure for payments is flawed. Existing, developer-centric platforms are perfectly positioned to be called by agents. The agents will simply present payment options like Affirm to the user, obviating the need for a separate "agent economy" payment system.

The financial system is unprepared for the coming wave of AI agents. These agents will perform tasks and require payment, creating trillions of micropayments. Current infrastructure from Stripe, Visa, or Mastercard cannot handle this volume, creating a massive opportunity for new protocols to facilitate the 'agent economy'.