Original Article Title: A Year Inside Agentic Payments: The Uncomfortable Truth

Original Article Author: @13yearoldvc

Translator: Peggy

Editor's Note: This article provides a relatively sober 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, and Google making strategic moves. Concepts such as stablecoin microtransactions, x402, machine-to-machine settlements, and agentic e-commerce have been gaining momentum. However, the author discovered after actually building products, engaging with merchants and developers, that the true demand has not materialized at scale.

The article breaks down several typical scenarios: agentic shopping is not inherently better than traditional e-commerce in most categories, as users still require images, comparison, and browsing; machine API payments may seem suitable for stablecoin microtransactions, but most developers have already addressed this through subscriptions, prepaid credits, and existing billing systems; inter-agent payments, although a long-term vision, are currently in an early stage with a lack of real transaction volume.

Comparatively, 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 provide tangible enhancements such as real-time monitoring and automated rebalancing. However, this market also favors traditional institutions that already possess licenses, compliance, and client relationships.

The author's final verdict is: what the agentic economy truly lacks is not just a payment layer, but a more complex coordination ability — how to enable agents to collaborate with humans, validate task completion, and settle outcomes. Payment is just one piece of the puzzle. For incumbents, early positioning is a defensive choice; but for startups, the key is to identify existing markets that are ready.

The following is the original text:

Over the past year, I have been building infrastructure for the Agent economy and collaborating with teams advancing Agent businesses at Stripe, Visa, Coinbase, Google, and dozens of startups. I have navigated this space, launched products, and attempted to find true market fit.

However, the reality is: the real demand has not yet emerged. For startups looking to enter this space, there are still many structural challenges.

Stripe unveiled 288 new products at last month's Sessions conference, with Agent-related documentation now accounting for almost 40% of all documentation views. Its Agent Business Market has onboarded over 1,000 merchants. However, at the Sessions event, the actual number of Agents 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 $2.5 billion to qualify for access. Today, only companies at the level of Amazon and Walmart have the capability to close the identity verification loop.

Coinbase previously reported that as of April, there were 69,000 active Agents and 165 million transactions on x402. However, independent on-chain analysis shows that the actual daily transaction volume is around $17,000, with about half of it still being test transactions (CoinDesk, March 2026).

What We Learned from Building shop.fast.xyz

From Agent to Merchant, or Agentive Commerce

We built shop.fast.xyz with the goal of positively validating Agentive Commerce. Real products, real merchants, real transactions.

However, for most product categories, the current AI shopping experience is clearly inferior to traditional e-commerce. When buying clothes, electronics, or furniture, users want to see pictures, browse options, and compare side by side. A chatbot-style conversation is a step backward: you replace a rich visual interface with a text conversation. Human shopping is first and foremost visual shopping.

The Agent performed well in the part we originally thought would be the most difficult. It can understand what users want and can handle requests like "something like this, but cheaper" very well. The model layer is effective. But it cannot replace the experience of "looking at ten products at once and then choosing one." The chat interface can incorporate product carousels and interactive displays, but at that point, you are essentially rebuilding an e-commerce frontend within a chat window. For shopping scenarios that require visual comparison, we have not yet found a compelling answer as to why a chat shell would be better than the original e-commerce interface.

We did see demand on the merchant side, but this demand is more defensive. Merchants want their stores to be queriable by Agents, not because many consumers are already shopping through Agents today, but because they are concerned that if Agents become the mainstream channel in the future, they will be left behind. This is the so-called Agentic Engine Optimization opportunity, but it is currently only a "nice-to-have" rather than a "must-have." Merchants are preparing preemptively for a wave that has not yet arrived.

The true power of conversational commerce is in scenarios that are high-frequency, low decision cost, and where users already know what they want to buy. The clearest example is ordering food. The market is large enough, the frequency is high enough, and the decisions are quick enough, such as "Help me order Pad Thai from the last restaurant I liked." In this scenario, a conversational agent may come out on top. However, major food delivery platforms do not provide open APIs. The only path is through computer use, meaning having AI interact with apps visually like a human. This process is slow, fragile, and simply does not justify the reasoning cost for a $15 lunch.

Another opportunity lies in online stores that are so complex that they truly frustrate users. For example, layers of discounts, promo codes, loyalty points, and a chaotic checkout process. An agent that can understand "Help me apply a coupon, redeem my points, find the cheapest shipping option, and complete the transaction in my language" could indeed streamline today's already broken shopping experience. This is particularly crucial for elderly users, non-native speakers, especially when shopping across regions, or in very specific scenarios where users have extremely niche and complex requirements.

However, both of these opportunities require significant B2C distribution capabilities. You are competing with DoorDash and Amazon for user entry points. The consumer-scale distribution capability is an advantage held by existing giants. The supply side of agent-based commerce is already prepared, but the demand side is constrained by user experience and distribution channels, and more infrastructure cannot solve these two issues.

What We Learned in x402 and MPP

From Agent to Web/API, Aka Machine Commerce

We engaged with dozens of developers to understand their real payment needs. The pattern is nearly identical: Today's agent APIs are essentially for recurring consumption, such as computational power, inference, and data sources. Developers are already accustomed to subscriptions, API keys, linked accounts, and billing relationships with core service providers.

The typical argument for stablecoin payments is that the effective lowest cost for credit card payments on Stripe is about 2.9% plus 30 cents, making API calls below $1 uneconomical. However, with low transaction volumes today, topping up balances can solve this issue. Developers pre-fund their accounts, eliminating this problem.

A deeper issue lies in the supplier market. Most large SaaS companies are not keen on offering disparate API access for a fraction of a cent. Their business model revolves around multi-year enterprise contracts. Companies reliant on large committed revenue will resist new pricing models that bypass this structure.

The Machine Commerce is structurally a long-tail market. It serves small services, vertical data sources, independent developers, MCP servers, and more. Protocols like MPP and x402 are well suited for this niche market. However, by definition, this is a market aimed at professional demand users; and developers have always been one of the least willing to pay groups.

When launched, Stripe Projects onboarded 32 service provider partners, including Vercel, Supabase, Cloudflare, Twilio, covering most of the core services developers use to build and deploy software, all accessible through existing billing systems. The top of the developer stack has already been well served. The opportunity for a new payment rail lies in everything beyond those top 30 service providers: it is real but naturally smaller in scale than the market spaces implied by grand narratives.

A similar logic applies to content access. Agents have been continuously crawling and summarizing articles, and publishers have started to fight back. However, when content monetization truly arrives at scale, it is likely to flow through existing CDN providers already positioned between publishers and the internet, such as Cloudflare, which has already introduced AI auditing tools; or through bulk licensing agreements between publishers and AI labs. The infrastructure opportunity will flow to existing players with distribution capabilities.

What We Learned from Agent-to-Agent Payments

Business between Agents is a long-term vision but is currently almost entirely theoretical. There is no significant transaction volume yet. The truly challenging parts are being driven by various startups, including Agent discovery, trust establishment, terms negotiation, and dispute resolution.

Once this transaction structure takes shape, it looks to be entirely different from the current payment rails. Neither party to the transaction has a human identity; latency requirements are less than a 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 default bilateral buyer-seller model of existing payment rails. When it does materialize, we believe it will happen at an extremely rapid pace and massive scale.

This is indeed a long-term bet on dedicated settlement infrastructure, and this bet is real. However, a "real long-term bet" is not the same as the "current market." We have been among those who claimed for months that this market was coming and have built an entire infrastructure around it over the past few years, including our distributed network. In theory, it can scale to over 1 billion TPS, with latencies below 50 milliseconds and average consistency time of 10 milliseconds. But we must return to where the market is now.

What We Learned in Agent Finance

One could argue this is the only category with existing real demand. The customer already exists and is already paying. Fund managers, asset management teams, and DeFi users are already spending on financial tools today. Inserting AI into existing workflows is a natural product path.

Agent Finance will also create entirely new behavior patterns. An Agent capable of autonomously monitoring and rebalancing hundreds of positions in real-time can operate in a way that humans cannot replicate manually. There is a real step change in capability here, not just automation.

The challenge lies in the competitive landscape. The financial industry is highly regulated and relies on existing relationships. Incumbents have licenses, compliance infrastructure, and customer relationships. Startups can enter lighter regulated areas like DeFi, look for slower-moving areas within incumbents, or find AI capabilities that can create new abilities that incumbents do not yet possess. However, overall, the competitive dynamics in this space are more favorable to existing players than the first three categories because layering AI on top of existing products and customer bases is much easier than starting with AI and then supplementing with products and customers.

Honest Recap

So, why is everyone still doing this? There are two reasons.

The first is the incentive structure. Large companies have enough cash flow to bet on a future that may take years to materialize. For them, the cost of entering five years early is just a rounding error; but the cost of entering one year late could be catastrophic. So they have to do it.

The second is a cognitive blind spot. When your business is payments, every problem looks like a payment problem. Agent economics need a payment layer, so everyone goes to build a payment layer.

However, payments are only a small part of a larger issue. The truly difficult problem is not about moving money between agents, but about how to coordinate between agents and humans, how to verify if things are done, and how to settle outcomes. Payments are just part of settlement. Settlement is just part of coordination. And coordination is the real prize.

Large-scale coordination naturally leads to a need for settlement mechanisms. Payments will become one instrument in this symphony, not the composition itself. Companies truly addressing the coordination problem will ultimately incorporate payments into it, rather than having payment companies swallow coordination.

Most existing incumbents are defensively building towards a future of "machine-scale transactions." For them, the timeline is inconsequential, as they have nearly infinite runway.

But startups don't have that luxury. We have to find out where the market really is right now. We can't just wait for the wave to come.

A year of building took us in an unexpected direction. There was indeed activity, and the growth was rapid, with underserved demand. It existed outside of the four categories we had mapped out.

[Original Article Link]

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia