Original Title: a year inside agentic payments: the uncomfortable truth
Original Author: @13yearoldvc
Original Translator: Peggy
Editor's Note: This article provides a relatively calm builder's perspective: over the past year, agentic payments have become a hot narrative in the intersection of AI and payments, with companies like Stripe, Visa, Coinbase, Google, and others making moves, while concepts like stablecoin micropayments, x402, inter-machine settlements, and agentic e-commerce are increasingly in the spotlight. However, the author found that real demand has not yet emerged on a large scale after getting involved in product development and engaging with merchants and developers.
The article dissects several typical scenarios: agentic shopping is not significantly better than traditional e-commerce in most categories, as users still want images, comparisons, and browsing; machine API payments seem suitable for stablecoin micropayments, but most developers have already solved their issues through subscriptions, top-ups, and existing billing systems; payments between agents, while a long-term vision, are still at an early stage with a lack of real transaction volume.
Relatively speaking, agentic finance is one of the few directions with existing demand. Funds, treasury teams, and DeFi users are already paying for financial tools, and AI can bring real-time monitoring and automated rebalancing capabilities. However, this market is also more favorable to traditional institutions that already possess licenses, compliance, and customer relationships.
The author's final judgment is that what the agentic economy truly lacks is not simply a payment layer, but more complex coordination capabilities—how to enable agents to collaborate with humans, validate task completion, and complete result settlements. Payments are just one part of the equation. For giants, laying groundwork in advance is a defensive choice; but for startups, what is truly important is finding the existing market that already exists.
The following is the original text:
Over the past year, I have been building infrastructure for the Agent economy and communicating with teams at Stripe, Visa, Coinbase, Google, and dozens of startups pushing forward with agent business. I have sorted out this field, launched products, and tried to find real markets.
But the reality is: real demand has not emerged. For startups looking to enter this field, there are still many structural issues present.
Last month, Stripe launched 288 new products at the Sessions conference, and the access traffic for agent-related documents has approached 40% of total document reading volume. Its agent business marketplace has onboarded more than 1,000 merchants. However, at the Sessions event, the actual number of agents that registered and completed transactions was only in the single digits.
Visa mentioned that its agent token currently requires 3 to 9 months of KYC approval, and the basic requirement is that companies must have an annual revenue of at least $250 million to be eligible for onboarding. Today, only companies at the level of Amazon and Walmart have the capability to close identity verification links.
Coinbase reported that as of April, there were 69,000 active agents and 165 million transactions on x402. However, independent on-chain analysis shows the real daily transaction volume is approximately $17,000, with about half of that still being test transactions (CoinDesk, March 2026).
What We Learned While Building shop.fast.xyz
Agent to Merchant, which is Agency Business
We built shop.fast.xyz with the goal of validating agency business directly. Real products, real merchants, real transactions.
However, for most product categories, the current AI shopping experience is significantly inferior to traditional e-commerce. When buying clothes, electronics, or furniture, users want to see images, browse options, and compare side by side. Chatbot-style conversations are actually a regression: you are replacing a rich visual interface with a stream of text dialogue. Human shopping is primarily visual shopping.
Agents performed well in what we initially thought would be the most difficult part. They can understand what users want and handle requests like "something similar but cheaper" effectively. The model layer is effective. But it cannot replace the experience of "seeing ten products at once and then selecting one." The chat interface can include product carousels and interactive displays, but at that point, you are essentially reconstructing an e-commerce frontend in a chat window. For shopping scenarios that require visual comparison, we have yet to find a compelling answer to why a chat shell would be better than the original e-commerce interface.
We do see demand on the merchant side, but this demand is more defensive. Merchants want their stores to be queryable by agents, not because there are already many consumers shopping through agents today, but because they fear being left behind if agents become a mainstream channel in the future. This is what we call the opportunity for Agentic Engine Optimization, but it is currently "better to have" rather than "must have." Merchants are preparing in advance for a wave that has not yet arrived.
The real improvement that conversational commerce can offer is in high-frequency, low-decision-cost purchases where the user already knows what they want. The clearest example is ordering food. The market is large enough, frequency is high enough, and decisions are quick, such as "help me order a Pad Thai from that restaurant I liked last time." In this scenario, a conversational agent may emerge victorious. However, major food delivery platforms do not open APIs. The only path is computer use, which means having AI interact with apps in a human-like visual manner. This process is slow, fragile, and the reasoning cost does not hold for a $15 lunch.
Another opportunity lies in online stores that are complex to the point of causing real pain for users. For example, layered discounts, coupon codes, loyalty points, and a confusing checkout process. An agent that can understand "help me apply coupons, redeem points, find the cheapest delivery method, and do it all in my language" could indeed simplify a shopping experience that has already gone awry. This is particularly important for elderly users, non-native language users, especially in cross-regional shopping; or in some very specific scenarios, users may have extremely niche and complex needs.
However, both of these opportunities require significant B2C distribution capabilities. You are competing with DoorDash and Amazon for user entry points. Distribution capabilities in terms of consumer scale are the advantage of existing giants. The supply side of agency business is ready, but the demand side is limited by user experience and distribution channels, and more infrastructure cannot solve these two problems.
What We Learned in x402 and MPP
Agent to Web/API, which is Machine Business
We have communicated with dozens of developers about their real payment needs. The patterns are almost entirely consistent: today's agent API usage is essentially for recurring consumption, such as computing power, inference, and data sources. Developers already have subscriptions, API keys, bound accounts, and billing relationships with core service providers.
A typical argument for stablecoin payments is: the effective minimum cost of credit card payments on Stripe is around 2.9% plus 30 cents, making API calls below $1 uneconomical. But at today's low transaction volume, the problem can be solved with top-up points. Developers preload their accounts, and this issue ceases to exist.
The deeper issue lies in the supplier market. Most large SaaS companies do not want to offer piecemeal API access for a fraction of a cent. Their business models are multi-year enterprise contracts. Companies relying on large committed revenues will resist new pricing methods that circumvent this model.
Machine business is structurally a long-tail market. It serves small services, vertical data sources, independent developers, MCP servers, etc. Protocols like MPP and x402 are very suitable for this niche market. But by definition, this is a market aimed at professional demand users; and developers have traditionally been one of the most reluctant groups to pay.
When Stripe Projects launched, it onboarded 32 service provider partners, including Vercel, Supabase, Cloudflare, Twilio, etc., covering most of the core services used by developers when building and deploying software, and all of which can be accessed through existing billing systems. The top of the developer tech stack has already been well served. The opportunity for new payment tracks lies in everything beyond those first 30 service providers: it exists, but its scale is naturally smaller than the market space that grand narratives suggest.
Content access follows the same logic. Agents are continuously scraping and summarizing articles, and publishers are starting to fight back. However, when content monetization truly arrives at scale, it will likely be realized through CDN service providers already positioned between publishers and the internet, such as Cloudflare, which has already launched AI auditing tools; or through batch licensing agreements between publishers and AI labs. Infrastructure opportunities will flow to existing players who already have distribution capabilities.
What We Learned in Agent to Agent Payments
Business between agents is a long-term vision but is currently almost entirely theoretical. No one has generated meaningful transaction volume. The real difficult parts are being pushed forward by various startups, including agent discovery, trust building, terms negotiation, and dispute resolution.
Once this transaction structure truly takes shape, it will look completely different from existing payment tracks. No party involved will have a human identity; latency requirements will be under one second; transaction amounts can range from a fraction of a cent to millions of dollars; and it will involve multi-party settlements rather than the bilateral buyer-seller model that existing payment tracks default to. When it does happen, we believe it will explode at an extremely fast pace and enormous scale.
This is precisely the long-term bet on dedicated settlement infrastructure, and this bet is real. But "real long-term bets" and "current markets" are not the same thing. We were also among those who claimed for several months that this market would arrive and built a whole set of infrastructure around it over the past few years, including our distributed network. Theoretically, it can scale to over 1 billion TPS, with latency under 50 milliseconds and average consistency time of 10 milliseconds. But we must return to where the market currently stands.
What We Learned in Agent Finance
It can be said that this is the only category where real demand already exists. Customers are already present and paying. Fund managers, treasury management teams, and DeFi users are already spending money on financial tools today. Inserting AI into existing workflows is a natural product path.
Agent finance will also create entirely new behavior patterns. Agents that can autonomously monitor and rebalance hundreds of positions can operate in ways that humans cannot manually replicate. There are real capability improvements here, not just automation.
The challenge lies in the competitive landscape. The financial industry is highly regulated and relies on existing relationships. Existing institutions have licenses, compliance infrastructure, and customer relationships. Startups can enter areas with lighter regulation, such as DeFi; or they can look for fields where existing institutions are slow to act, or where AI can create new capabilities that giants have yet to possess. But overall, the competitive dynamics in this field are more favorable to existing players, because layering AI on top of existing products and customer bases is much easier than the reverse—starting with AI and then supplementing it with products and customers.
Honest Summary
So, why does everyone continue to pursue this endeavor? There are two reasons.
The first is the incentive mechanism. Large companies have enough cash flow to bet on a future that may take years to manifest. For them, the cost of entering five years early is just a decimal point error; but the cost of entering a year late could be catastrophic. So they must do it.
The second is cognitive blind spots. When your business is payments, every problem seems like a payment problem. The agent economy requires a payment layer, so everyone goes about building that payment layer.
But payments are only part of a larger issue. The truly difficult problem is not making money flow between agents but how to coordinate work between agents and humans, how to verify tasks are completed, and how to settle results. Payments are just part of settlement. Settlement is just part of coordination. And coordination is the real prize.
Large-scale coordination will naturally create a demand for settlement mechanisms. Payments will become one instrument in this concerto, rather than the whole piece itself. Companies that genuinely solve coordination problems will ultimately incorporate payments into their operations, rather than the other way around, where payment companies absorb coordination.
Most existing giants are defensively building a future of "machine-scale transactions." For them, timelines are not critical because they have almost unlimited runway.
But startups do not have this luxury. We must find where the market truly lies today. We cannot just wait for the wave to arrive.
A year of construction has led us in an unexpected direction. There is indeed activity, and growth is fast, yet under-served. It exists beyond the four categories we have outlined.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
