Dragonfly Partners: Most agents will not engage in autonomous trading, how can crypto payments prevail?

Author: Robbie Petersen, Dragonfly Junior Partner

Compiled by: Gu Yu, ChainCatcher

Whenever an emerging narrative enters public discourse, mainstream arguments are simplified into the most easily accepted forms by the public. Intuitively, when no one can empirically prove what will happen, provocation is more likely to yield returns than detailed analysis.

The recent discussions around "agent commerce" are no exception. There is a consensus in the market: the number of agents is skyrocketing; agents need to transact; agents cannot hold bank accounts but can hold e-wallets; card organizations charge a 2-3% fee; therefore, stablecoins win.

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

However, the debate about "which payment method will prevail?" actually stems from a premise that has been largely overlooked in the discussion:

Will most agents really engage in transactions?

The scale of the agent economy will be massive, but the proportion of agents that actually engage in transactions is unlikely to be that high.

The Agent Economy is More Like an Organizational Chart than a Market

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

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

This fundamentally contradicts the actual application of agents.

Overall, agent deployment can be divided into two categories: commercial agents deployed on behalf of businesses and consumer agents that enhance our personal lives. For different reasons, both types of agents are unlikely to autonomously engage in transactions.

Commercial Agents are an Inevitable Evolution of SaaS

A reasonable concept of commercial agents is an inevitable evolution of SaaS. They do not enhance workflows; they replace existing workflows. Just as over 95% of software spending comes from businesses and governments, over 95% of large-scale agent application scenarios are likely to be deployed within similar organizations.

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

Take sales agents as an example. They connect to CRM systems, research potential customers, write personalized marketing copy, and arrange follow-ups. They do not spend autonomously and do not interact with external agents from other organizations. They simply execute a task—sales expansion—within a closed environment and automate it.

Intuitively, this situation applies to almost 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 nearly all use cases, the agents themselves do not spend and are not granted spending authority. They are deployed top-down in a controlled organizational environment with permission control mechanisms.

Even if they do need to interact across organizations and pay for their API calls or data, the costs may not be reflected in the form of autonomous payments by the agents. Any usage-based costs may be abstracted by the software vendor. This is how enterprise software stacks operate. Platform providers negotiate customized partnerships with data providers, computing providers, and other infrastructure partners, packaging access into platform costs and passing it off as a single aggregated item.

Moreover, they can achieve this with unit economics that no single agent could autonomously replicate. Computing resources are acquired through reserved capacity agreements with AWS, Azure, or GCP. Pricing for model inference is based on bulk agreements with companies like Anthropic, OpenAI, or Google. Data augmentation is done 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 by agents do not generate 40,000 payments but rather a single invoice. The granularity of consumption has never aligned with the granularity of settlement, and enterprises may prefer to maintain this state.

Consumer Agents Will Coordinate, Not Consume

While commercial agents may not engage in autonomous transactions because businesses will not allow it, consumer agents also will not engage in autonomous transactions because people do not want them to.

An example often cited by advocates of smart commerce: you let your agent book a trip to Tokyo. It searches hundreds of hotels, cross-references reviews, checks your calendar, and applies your preferences. Then, it automatically books a room. You do not need to do anything. Of course, those promoting the agent-based business model will extend this user experience to almost all consumer domains, from groceries to home goods to clothing, and so on.

The problem is that preferences are not static. They manifest in the choice behavior itself. When you book a hotel, you are not just looking for the cheapest accommodation. The judgments you make reflect your mood, context, risk tolerance, and other qualitative factors that you may not even be aware of before reviewing options.

In practice, agents will search, ask follow-up questions, and return options. You will look at pictures, inquire about the surroundings, 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 payment. In other words, the agent is merely a research assistant, not an independent economic entity.

Except for certain predictable repeat purchases, this user experience is likely to remain consistent across almost all consumer domains, precisely because consumer decisions rarely depend solely on price. The entire consumer economy is built on product differentiation. Whether it’s clothing, hotels, home goods, or groceries, the decision-making process involves countless qualitative factors, which not only cannot be captured by agents—more importantly, these factors exist within the process of user discovery itself.

Agents will play a powerful coordinating role in the discovery phase, but at critical moments, they will return decision-making power to humans. Semantically, this is not agent commerce, nor does it require the establishment of new payment channels.

The Real Advantage of Crypto Payments: Bottom-Up Agents

While these two types of agents may account for over 95% of agent deployments in the next five years, there is a third type worth noting.

In recent months, a new type of bottom-up agent has begun to emerge. Driven by the OpenClaw phenomenon, these agents belong to a fundamentally different category. Unlike the aforementioned commercial and consumer agents, they are truly autonomous actors, independent of any organizational entity. These agents require actual payments, and the granularity and frequency of payments are so high that manual authorization becomes impractical. Although the bottom-up agent economy is currently small, it is likely to grow rapidly with the emergence of some unforeseen new use cases.

Thus, only in this extremely narrow context does the debate about which is the best underlying infrastructure for crypto payments or card networks become compelling. While everyone is citing technical arguments for why crypto payments are superior, I believe the reason they may ultimately prevail is something else—permissionless.

The reality today is that neither payment method is technically optimized for agent commerce. While blockchain theoretically offers better unit economics for micro-payments, it lacks identity verification and risk scoring mechanisms—critical factors that may become particularly important in the future agent era. Additionally, while instant settlement is often mentioned, it merely means that fraudulent transactions will settle on-chain immediately. In contrast, card organizations have complex fraud patterns and tokenized credentials that agents can inherit, but these tools are trained on human behavior patterns and cannot be directly mapped to autonomous agent transactions. Moreover, for cross-border transactions, agents will also be subject to the settlement times imposed by card organizations.

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

This is its ultimate structural advantage. While I believe existing card organizations like Visa and Mastercard will continue to adjust through initiatives like Visa Intelligence Commerce and Mastercard's AgentPay, they are ultimately public companies that must comply with regulatory obligations, meet customer access requirements, and collaborate with institutional trading counterparts. Blockchain has none of these restrictions. Anyone can develop on the blockchain, any agent can transact, and no approval is needed.

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

The Bottleneck is Not Infrastructure, But Ourselves

However, the longer-term question is how the speed of this experimental development can ultimately create a greater impact. The bottom-up agent economy will only truly become popular when autonomous agent organizations are significantly superior to human organizations enhanced by agents; this advantage will not be slight but significant enough that top-down human restrictions on agents become a competitive disadvantage. At that point, agents will no longer merely be automated executors of human tasks in closed environments but will become the organizations themselves.

However, we may be far from this future. The bottleneck will not lie in the technology itself. Moreover, what may truly be "not suitable for machines" is not the payment systems themselves but everything else that is not designed for an autonomous agent economy: regulatory frameworks, institutional bureaucratic styles, legal structures, and the social inertia surrounding human decision-making. These constraints have far more profound implications than any technical detail in 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.