Author: Robbie Petersen, Dragonfly Junior Partner

Translated by: Gu Yu, ChainCatcher

Whenever a new emergent narrative enters public discussion, mainstream arguments tend to be simplified into the most easily digestible forms for the general public. Intuitively, when no one can empirically prove what will happen, provocations are often more rewarding than meticulous analyses.

The recent discussions surrounding "agent commerce" are no exception. There has emerged a consensus in the market: the number of agents is surging; agents need to conduct transactions; agents cannot hold bank accounts but can hold e-wallets; card organizations charge a fee of 2-3%; therefore, stablecoins prevail.

This logical chain has flaws on many levels. Agents can hold bank accounts under a FBO (Financial Operator) structure. Furthermore, the 2-3% fee reflects credit risk and fraud risk, which blockchain does not solve.

However, the debate around "which payment method will prevail?" actually stems from a premise question that is largely overlooked in discussions:

Do most agents actually conduct transactions?

The scale of the agent economy will be massive, but the proportion of agents that actually engage in transactions will not be nearly as high.

The agent economy resembles an organizational chart rather than a market

Fundamentally, artificial intelligence is a form of automation technology. It can perform certain tasks—such as searching, aggregating, and synthesizing—and do so more efficiently than humans. Agents are operational derivatives of artificial intelligence. They do not merely return output results; they perform actual actions.

The underlying assumption of the entire agent commerce theory is: execution comes at a cost. In other words, for most agent tasks, they need to spend funds to autonomously obtain external resources, pay for computation and data based on usage, and interact as independent economic entities with other agents.

This fundamentally contradicts the actual application of agents.

Overall, agent deployment can be categorized into two types: commercial agents deployed on behalf of enterprises, and consumer agents that enhance our personal lives. For different reasons, it is unlikely that both types of agents will autonomously conduct transactions.

Business agents are the inevitable evolution of SaaS

A reasonable concept of business agents represents the inevitable evolution of SaaS. They do not enhance workflows but replace existing workflows. Just as over 95% of software spending comes from enterprises and governments, over 95% of large-scale agent application scenarios may also be deployed within similar organizations.

This is the first subtlety that the current mainstream theory of agent commerce overlooks: the vast majority of agent demand does not come from agents booking flights for consumers but from top-down deployment within enterprises. More importantly, agents that automate task execution within closed organizations are fundamentally different from agents that operate as independent economic entities.

Take sales agents as an example. They connect to CRM systems, research potential clients, write personalized marketing copy, and arrange follow-ups. They do not spend autonomously, nor do they interact with external agents from other organizations. They merely execute a task—sales expansion—within a closed environment and automate it.

Intuitively, this scenario applies to nearly all organizational functions. Financial agents audit and verify expenses; accounting agents record journal entries, reconcile accounts, and prepare reports; legal agents review contracts and identify exceptions; coding agents write code.

In almost all use cases, agents themselves do not spend and are not given spending authority. They are deployed top-down in a controlled organizational environment with permissions in place.

Even if they do need to interact across organizations and pay for their API calls or data, costs may not be reflected as payments made autonomously by agents. Any usage-based costs may be abstracted by the software sellers. This is how the enterprise software stack operates. Platform providers negotiate customized partnerships with data providers, compute providers, and other infrastructure partners, packaging access into the platform cost, and passing it off as a single aggregated entry.

Moreover, they can achieve this with unit economics that no single agent can independently replicate. Computing resources are obtained through reserved capacity agreements with AWS, Azure, or GCP. Model inference pricing is based on bulk agreements with companies like Anthropic, OpenAI, or Google. Data augmentation is facilitated through vendors like Bombora or Clearbit. All of this is pre-negotiated and abstracted.

In other words, the 40,000 API calls, model inferences, and data queries of an agent do not result in 40,000 separate payments but generate one invoice. The granularity of consumption has never aligned with the granularity of settlement, and corporations may prefer to keep it that way.

Consumer agents will be responsible for coordination, not consumption

While business agents may not engage in autonomous transactions because enterprises won't allow it, consumer agents will also refrain from autonomous transactions because people do not want them to do so.

Take an example often cited by proponents of smart commerce: you have your agent book a trip to Tokyo. It searches hundreds of hotels, cross-references reviews, checks your calendar, applies your preferences, and then automatically books the room. You need to do nothing. Of course, those who advocate for agent-based business models will extend this user experience across almost all areas of consumption, from groceries to household items to clothing.

The issue is that preferences are not static. They manifest in the choices themselves. When you book a hotel, you are not just looking for the lowest-priced accommodation. The judgments you make reflect your mood, context, risk tolerance, and a variety of qualitative factors you may not even be aware of before reviewing the options.

In practice, agents will search, ask follow-up questions, and return options. You will check images, inquire about the surrounding area, and perhaps read some reviews. Then you will make a choice and authorize the agent to use the credit card information it has on file for the payment. In other words, the agent is merely a research assistant, not an independent economic entity.

Aside from certain predictable repetitive purchases, this user experience is likely consistent across nearly all areas of consumption, precisely because consumer decision-making rarely depends solely on price. The entire consumption economy is built on product differentiation. Whether it’s clothing, hotels, household goods, or groceries, the decision-making process involves countless qualitative factors that agents cannot capture—and more importantly, these factors exist within the process of user discovery itself.

Agents will play a powerful coordinating role during the discovery phase, but at critical moments, they will return decision-making authority to humans. Semantically, this does not constitute agent commerce and does not necessitate the establishment of new payment channels.

Where crypto payments truly excel: bottom-up agents

While the combined proportion of these two types of agents may account for over 95% of agent deployments in the next five years, there is a third type that deserves attention.

In recent months, a new type of bottom-up agent has begun to emerge. Driven by the OpenClaw phenomenon, these agents belong to a distinctly different category. Unlike the aforementioned business and consumer agents, they operate truly autonomously, independent of any organizational entity. These agents necessitate actual payments, and the granularity and frequency of those payments are such that manual authorization becomes impractical. Although the bottom-up agent economy is currently small, it is likely to grow rapidly with the appearance of some unforeseen emerging use cases.

Thus, only in this extremely narrow context does the debate over which is the better underlying architecture—crypto payments or card networks—become compelling. While everyone presents technological arguments for the superiority of crypto payments, it seems to me that the reason they may ultimately prevail lies elsewhere—specifically, the permissionless nature of crypto.

The reality today is that neither of these payment methods are optimized for agent commerce from a technical standpoint. While blockchain theoretically offers superior unit economics for micropayments, it lacks identity verification and risk scoring mechanisms—elements that may prove crucial in the agent era ahead. Moreover, while immediate settlement is often mentioned, it merely means that fraudulent transactions would settle on-chain immediately. Conversely, card organizations maintain complex fraud maps and tokenized credentials for agents to inherit, but these tools are trained on human behavioral patterns and cannot be directly mapped to autonomous agent transactions. Additionally, for cross-border transactions, agents would also be subject to card organization settlement timelines.

Perhaps counterintuitively, the reason crypto payment methods may become the default infrastructure for these types of agents is that blockchain is open, permissionless, and unregulated.

This is its ultimate structural advantage. While I believe that existing card organizations like Visa and MasterCard will continue to adapt through initiatives like Visa Intelligence Commerce and MasterCard's AgentPay, they are after all publicly traded companies that must comply with regulatory obligations, meet customer onboarding requirements, and collaborate with institutional trading counterparts. Blockchain has no such limitations. Anyone can develop on the blockchain, and any agent can transact without any approvals.

Intuition tells us that emerging, experimental categories will develop where friction is minimized.

The bottleneck lies not in the infrastructure, but in ourselves

However, the longer-term question is how the speed of this experimental development can ultimately yield greater impacts. The bottom-up agent economy will only become truly popular when autonomous agent organizations clearly outperform human organizations enhanced by agents; this advantage is not marginal but sufficiently significant that top-down human restrictions on agents become a competitive disadvantage. At that point, agents will no longer merely automate human tasks in controlled environments, but will become the organization itself.

However, we might be far from that future. The bottleneck will not lie in the technology itself. Moreover, what is truly "unsuitable for machines" may not be the payment systems themselves but rather all the other elements not designed for an autonomous agent economy: regulatory frameworks, institutional bureaucracies, legal structures, and the social inertia surrounding human decision-making. These limitations have far-reaching impacts that far exceed any technological details within the payment stack. Unfortunately, protocol upgrades cannot solve these issues.

The scale of the agent economy will be massive, with most of it billed monthly.

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。