If you're a founder building in ecommerce, you've likely noticed a shift: your customers are turning to ChatGPT and Claude not just for research, but for recommendations—and they're ready to buy based on what these platforms suggest. Ecommerce gave consumers power, but it also created paralysis. Purchasing involves navigating dozens of tabs, comparing endless specifications, and second-guessing decisions. AI agents are beginning to solve this. Purchases will increasingly happen inside chat interfaces, productivity tools, and ambient computing environments. But the infrastructure enabling this shift doesn’t exist yet.

If you can eliminate the friction that slows agentic transactions today, you’ll be able to define the next era of commerce.

Fully Automated Purchasing

The difference between ecommerce and agentic commerce isn't technical, it's conceptual. The consumer's role shifts from executor to strategist, which breaks most of the digital infrastructure we currently rely on.

According to Stripe's Head of Information Emily Glassberg Sands, "AI isn't just answering questions. It's starting to do things, including buying things for people. That shift is reshaping commerce, with a new set of tools driving the infrastructure for agentic commerce that will help businesses grow."

Current state is when someone books a vacation, they open multiple browser tabs, compare flight prices across airlines, cross-reference hotel reviews, check rental car availability, coordinate dates, and make multiple purchases across multiple sites. Hours of research and dozens of micro-decisions before even a single transaction can take place.

With agentic commerce, all they have to do is state intent: "I want to take a weeklong family vacation for spring break." An agent researches destinations, evaluates options against preferences and budget, builds an itinerary, handles logistics, and presents final options. The user reviews and approves, empowering the agent to make all bookings. Hours of activity collapses into two touch points.

If you're a founder building the infrastructure that makes this possible, you're building for a fundamentally different transaction model.

Where the Opportunities Are

Today, companies like Stripe are developing some of the foundational tools for this future, like the Agentic Commerce Protocol (ACP) for programmatic commerce and the Shared Payment Token (SPT) for agent-initiated payments. But this is just the beginning—the architecture for agentic commerce remains incomplete.

If you're a founder looking at this space, we believe two areas define the opportunity: discovery and payments.

Building for Discovery

Agents need to find products and evaluate options on behalf of users, which requires machine-readable information from trusted sources. This has created Generative Engine Optimization (GEO), focused on making products discoverable to agents rather than human shoppers. Agents need ways to access merchant sites, machine-readable product catalogs, real-time inventory and pricing, and robust APIs. If you can standardize how agents access product information, you're building critical plumbing.

Discoverability leads to another challenge: disintermediation. When agents handle purchases, merchants lose their direct relationship with customers. A customer might buy a product without ever visiting the merchant's website, reading their marketing, or joining their email list. Some approaches are emerging—like loyalty programs that work through agents and merchant-specific agents that preserve brand connection—but most solutions remain to be built.

We see tremendous opportunity around designing new commercial models for the agentic era. Content creators need ways to get compensated when agents use their reviews and comparisons. Merchants need new affiliate and referral structures that work when agents, not humans, are making decisions. Loyalty programs must function when customers never directly interact with brands. Small creators in particular need access to attribution systems, paywalls that detect non-human traffic, and payment mechanisms that make small transactions economically viable.

Rethinking Payment Infrastructure

The promise of agents is not just that they will find products but that they’ll purchase them automatically. This upends traditional payment flows, with an API call completing checkout rather than a human clicking purchase.

Allowing agents to connect directly to merchant systems and complete transactions programmatically requires solving for two needs: programmability and transparency.

Programmability means embedding rules directly into payment instruments. Consumers need to be able to set parameters around their purchasing behaviors, like specifying which cards they want used for which types of purchases, what price threshold requires human approval, and which merchants they prefer. Agentic systems need to enforce these constraints automatically.

Transparency means both parties need clear signals about what's happening. Merchants need to distinguish legitimate agent purchases from fraud. Consumers need detailed logs of what their agent purchased, why, and from whom, as well as straightforward ways to fix mistakes.

Together, programmability and transparency creates trust and control layers, which are essential to creating secure agentic purchasing systems. People need to be comfortable delegating authority and purchasing decisions to agents, which means the systems need to detect fraud, adapt to human contexts, and offer granular control over spending if or when desired.

Building the Future of Commerce

The technical challenges to creating the architecture of a new era of commerce are real: enabling discovery, supporting programmable automation, securing payments, and ensuring trust all require sophisticated infrastructure.

But we believe the opportunity is immense. The founders who solve these problems will define the future of commerce, and the time to build that future is now.