Stripe's Annual Letter Proposes Five Stages of Agentic Commerce -- Currently at Levels 1-2, Charting the Path to Fully Autonomous Purchasing

Stripe Co-Founders Analyze the Staged Evolution of AI Shopping in Annual Letter

Stripe's slower view of agentic commerce

The processing giant, like some other digital commerce specialists, sees agent-enabled shopping evolving at a measured pace.

On February 24, 2026, payments infrastructure giant Stripe published its 2025 annual letter. Co-founders Patrick Collison and John Collison presented a five-stage framework for agentic commerce -- the concept of AI agents searching for and purchasing products on behalf of consumers. While the letter expressed strong optimism about the field's long-term potential, it also offered a cautious assessment that the industry remains in its early stages.

Stripe's total payment volume in 2025 reached $1.9 trillion (up 34% year-over-year), equivalent to approximately 1.6% of global GDP. With over 5 million businesses using its infrastructure, Stripe is an indispensable player in any discussion of agentic commerce trends.

Industry Context

Agentic commerce refers to the mechanism by which AI agents -- autonomous software -- discover, compare, purchase, and pay for products on behalf of consumers. The concept has gained rapid attention since the second half of 2025, with major research firms projecting enormous market potential.

A Boston Consulting Group (BCG) survey found that approximately 81% of consumers are open to using agentic commerce tools, with the addressable spending reaching $1.3 trillion. A McKinsey report published in October 2025 estimated that by 2030, up to $1 trillion in the U.S. alone -- and up to $5 trillion globally -- in retail sales could fall within the scope of agentic commerce.

However, reality has yet to catch up. As Payments Dive reported, while bot-initiated payments are trending upward, they still account for less than 1% of total volume. Stripe itself candidly acknowledged in the letter that agentic commerce has been subject to "premature hype in some quarters."

Stripe's Five Stages of Agentic Commerce

The centerpiece of the annual letter was a framework organizing agentic commerce maturity into five levels. Combined with eMarketer's analysis, here is an overview of each level.

Level 1: Auto-fill forms. AI automatically fills in purchase forms, streamlining the checkout process for consumers. This is the primary capability offered by current AI assistants.

Level 2: Descriptive prompts. Consumers describe situations to AI rather than specifying exact product names. For example, telling AI "a gift for my outdoorsy sister" yields suitable recommendations. According to Stripe, the industry currently sits at the boundary of Levels 1 and 2.

Level 3: Long-term memory and preference learning. AI remembers consumers' purchase history and preferences, enabling optimal recommendations without repeated explanations.

Level 4: Delegated purchasing. Consumers trust AI agents enough to delegate purchasing decisions within defined parameters. Once budgets and conditions are set, AI autonomously completes purchases.

Level 5: Fully autonomous purchasing. AI agents fully understand consumers' life patterns and purchase necessary items without any prompts. Stripe's example envisions back-to-school supplies being auto-purchased based on calendar events and past budgets, with only a completion notification sent to the consumer.

Infrastructure Development Through OpenAI Collaboration and Startup Investments

Stripe is not limiting itself to theory -- it has begun concrete technical development. According to the annual letter, Stripe co-developed the "Agentic Commerce Protocol (ACP)" with OpenAI, establishing common technical specifications between AI platforms and merchants. The company also launched the "Agentic Commerce Suite," enabling merchants to support multiple protocols -- including ACP and Google's Universal Commerce Protocol -- through a single integration layer.

A payment token sharing feature has also been introduced, allowing AI agents to complete transactions without directly handling sensitive information such as credit card numbers.

Startup investment is advancing as well. Circuit & Chisel, founded by former Stripe executives Louis Amira and David Noel-Romas, completed a $19.2 million funding round in September 2025. The company is developing "ATXP," a protocol enabling AI agents to autonomously complete everything from product discovery to payment across the web. Stripe itself is among the investors.

Meanwhile, John Collison stated on a podcast appearance that stablecoins (digital tokens pegged to fiat currencies) and high-throughput blockchains will become the infrastructure underpinning agentic commerce. Stripe has already integrated USDC-based AI agent payments via the x402 standard, and the stablecoin platform it acquired through Bridge in 2025 has seen transaction volume increase more than fourfold year-over-year.

Impact and Implications for E-Commerce Businesses

Stripe's five-stage framework serves as a guide for e-commerce businesses to plan their agentic commerce readiness in stages.

Prioritize Level 1-2 readiness in the short term. Specifically, this means organizing structured data (product names, prices, inventory information), publishing APIs, and implementing checkout flows that AI can easily interpret. If AI agents cannot accurately retrieve product information from your site, your products will not even appear as candidates.

Evaluate support for multiple agentic protocols. Multiple standards are emerging in parallel -- Stripe's ACP, Google's Universal Commerce Protocol, and Circuit & Chisel's ATXP, among others. A practical approach is to use a unified integration layer like Stripe's "Agentic Commerce Suite" to support multiple protocols simultaneously.

Advance payment tokenization and security readiness. As more consumers delegate payments to AI agents, systems that prevent card information from being directly shared with agents will become standard. Support for tokenized payments will be an important differentiator for building consumer trust.

Avoid excessive upfront investment and adopt a staged approach. According to eMarketer's forecast, 95% of e-commerce sales generated via AI platforms in 2026 will still be completed on merchant sites outside the AI platform. Fully autonomous purchasing becoming mainstream is still some way off, making infrastructure preparation the optimal strategy for now.

Conclusion

Stripe's annual letter brings a measured perspective to the industry's outsized expectations for agentic commerce while clearly articulating its long-term potential. In the letter, the Collison brothers draw parallels to the mid-1990s dawn of the internet, characterizing the current moment as analogous to the era of experimentation before HTTP and HTML were standardized.

What stands out is that despite its cautious stance, Stripe is steadily advancing real infrastructure development -- co-developing protocols with OpenAI, expanding stablecoin payments, and investing in startups. For e-commerce businesses, building an environment where AI agents can "discover" their products and monitoring new payment protocol developments will be the critical actions over the next 12 months.