a16z: What are the entrepreneurial opportunities in the Agent-Payment Transaction Blue Ocean?

Original Title: Agents will pay like locals, not tourists
Original Author: Sam Broner, a16zcrypto
Translation: Peggy, BlockBeats

Editor's Note: As AI Agents evolve from assistants to autonomous "digital doers" capable of completing tasks independently, the payment system is also undergoing changes. In the past, internet transactions mainly revolved around the retail process of "user click-checkout-payment." But in the Agent era, the subject of transactions is no longer just humans but intelligent systems capable of continuous operation and establishing long-term partnerships.

This article presents a vivid metaphor: An Agent will not make ad-hoc payments like a "tourist" but will behave more like a "local," completing transactions through stable vendor relationships, credit, and pre-negotiated business terms. In this model, the traditional card-centric payment system may only handle a portion of transactions, while programmable payment tools like stablecoins are expected to play a greater role in new payment scenarios.

The following is the original text:

Walking into a marketplace, if you are a tourist, what you often see is a bustling scene: people moving around, looking at products, comparing prices, sampling, bargaining with vendors, pulling out coins or swiping cards to complete transactions. Each interaction seems like a separate deal. An instant negotiation, trust settled instantly through cash or cards.

But in reality, most transactions do not happen this way.

If you look closer, you'll find more locals in the marketplace. They have a clear goal as they head to familiar merchants. Restaurant owners seek out familiar butchers, fishmongers, and farmers; tailors look for repairmen, weavers, and craftsmen. There is little haggling between them, and many transactions are completed directly on credit.

When we discuss how Agents will make payments, we often instinctively do so from the perspective of a "tourist." But an Agent's behavior is more akin to a local.

Differences between Agents and humans, such as infinite replication, flexible resource allocation, and near-zero startup costs, mean that a few Agents can establish dominance in specific areas. Even as the future lowers the barriers to building Agents, relationship networks, partners, and trust mechanisms will remain key determinants of experience quality.

Agents that truly dominate do not need tourist-style payment channels; they require vendor relationships, operating capital, and credit lines.
An Agent will complete transactions with the "tourists" (i.e., users).

So, what will this pattern look like?

As the Agent gradually evolves into an enterprise-like platform, its payment model will also shift from retail rails to pre-negotiated B2B terms and a credit system. The existing payment infrastructure is not well-equipped to meet this need.

This precisely provides an opportunity for the next-generation payment networks, such as stablecoins. However, the prerequisite is that entrepreneurs can build solutions around new payment scenarios, such as Agent payments, streaming payments, and high-frequency, low-value, globalized commercial transactions.

This article will explore this viewpoint from three aspects: first, what are the key differences between Agent and humans, and how these differences shape the future payment model; second, why the existing payment system struggles to meet Agent's needs; and third, what capabilities the next-generation payment infrastructure needs to have in order to succeed in future competition.

Differences Between Agent and Humans

Understanding Agent and payment requires answering two questions:

1. Is Agent's behavior more like an individual or more like an enterprise?

2. Are Agent's decisions more focused on short-term transactions or long-term partnerships?

The answers are: Agent is more like an enterprise and will establish long-term relationships.

Agents are often "lightweight instances" built on top of a larger business ecosystem. For example, a "smart tour guide Agent" supported by a large travel platform, or a franchisee fine-tuning operations based on local market demands within an existing supply chain.

Why do Agents behave like enterprises?

First, excellent experiences often come from pre-designed solutions, not on-the-spot negotiations.

Users do not want their Agent to start price comparisons, contact merchants, or renegotiate terms at checkout. An ideal Agent should have already completed this work: it knows which vendors are reliable, prices are negotiated, and can directly complete transactions.

This is a business relationship, not a tourist-style one-time transaction.

In fact, such a pattern has long existed in human society. Travel agents, literary agents, talent agents, watch dealers, real estate agents, etc., all fall under the category of "Agent." These agents establish long-term cooperative relationships with publishers, production companies, watch distributors, or financial institutions, and each transaction is customized on this basis.

Second, while an Agent can be infinitely replicated, the advantages of a scaled enterprise cannot.

The most successful Agents will leverage the benefits of scale: lower computational costs, better supplier pricing, deep system integrations, and stable technical components.

Scale begets scale. A travel agent booking one million flight tickets a year will secure better terms from airlines than an agent booking only ten tickets a year.

This trend is already evident. Only products like ChatGPT with significant user distribution can partner with platforms like Shopify, Amazon, Expedia, and others. Small startups often rely on browser automation or reverse-engineering API endpoints while operating at a retail cost structure.

That's also why Agents will ultimately centralize, or at the very least, most Agents will be built on top of large platforms.

Agents themselves are straightforward to develop, but economic realities dictate that only a few core Agents will emerge in each vertical, with deep supplier relationships and the ability to continuously optimize experiences through profits.

Meanwhile, specialized Agents in verticals can collaborate with user-side Agents to deliver more comprehensive services.

Two Payment Relationships

If an Agent's behavior leans more toward the enterprise, two payment relationships need to be designed: User → Agent; Agent (or Agent Platform) → Supplier

Users pay fees to Agents, which can take various forms: subscription fees, task-based fees, credit limits, authorization for the Agent to use the user's account

Agents, in turn, make payments to suppliers through B2B terms, such as pre-negotiated prices, bulk discounts, Net-30 invoices, sub-agent settlements

From the current enterprise expenditure perspective, Agents may still occasionally use retail payment channels, but this will only represent a small portion of overall spending.

In fact, this mirrors the credit card system today to a large extent. Credit card issuers establish retail relationships with consumers, taking on risks and providing credit and rewards; whereas acquirers form business relationships with merchants, completing transactions through negotiated fees, scaled settlements, and operational funding arrangements.

Agents and Credit Cards: A Seemingly Good Match

Many people believe that credit cards are actually a payment tool that is quite suitable for Agents.

Reasons include: globally accepted, suitable for transactions ranging from $20 to $1000, built-in arbitration and refund mechanism, providing monthly statements, with the monthly statement being particularly important as it helps users understand their spending.

In the future, when Agents replace children and iPads as the primary source of "unexpected bills," this may become even more important.

However, there are two problems in reality: 1. Credit card technology is not well-suited for Agent scenarios; 2. The fee structure of credit cards has led the industry into the typical "innovator's dilemma."

Challenges of Credit Card Technology Upgrade

Almost all credit card systems default to human involvement: human approval, user interface interaction, traditional payment types (one-time or subscription-based).

Virtual card technologies like Stripe Link, Visa 3D, etc., have taken over 15 years to mature. But the development pace of Agents is much faster than the upgrade of payment infrastructure. Thousands of PSPs, POS systems, merchant backends, and client interfaces cannot be adapted in a short time.

Credit Cards Unable to Cover Extreme Payment Scenarios

For example: real-time streaming payments from Agents to computing power service providers, Agents paying tiny fees for API calls, these transactions are difficult to complete with credit cards.

The reason is simple: Visa does not support transactions below 1 cent, and the credit card economic model relies on about a 30-cent fixed fee.

Technically, Visa could support micropayments, but this would directly impact its business model. What's more complex is that the payment scenarios of Agents often exceed the traditional amount range of credit cards. For example, many early Agent scenarios involve API service fees, which are both difficult to refund and difficult to resell. Credit cards can still play a role, but the innovator's dilemma often limits the speed of change in existing systems.

Traditional Payments Still Have Their Role

As the Agent platform gradually evolves into an enterprise-like system, a large amount of high-frequency spending will be completed through B2B terms: invoices, Net-30, discounts, credit limits.

In this mode, the "payment network" itself is not critical. Settlements may be completed via wire transfer, ACH, or batch transfers. Traditional payments still work in mature business relationships. However, Agents will not only exist in this environment.

Agents are rapidly emerging, often operating in the least efficient traditional payment scenarios: initial partnership setup, cross-border payments, complex reconciliation, new Agent-Vendor models, real-time payments, microloans

In these scenarios, stablecoins present a superior payment tool. More importantly, building new functionalities on programmable money is likely much easier than on traditional payment infrastructure.

Once new business relationships are established on stablecoins, they tend to persist in this form long-term. Over time, the proportion of stablecoins in the payment system is likely to continue to rise.

Opportunities in New Payment Technologies

Stablecoins are essentially a new financial platform.

They have the following characteristics: faster, lower cost, globally accessible, backed 1:1 by high-quality liquid assets

Most importantly, stablecoins are programmable. Functions like arbitration, invoicing, credit, custody, and conditional payments can all be flexibly achieved within the same system.

Compared to banks or credit cards, stablecoin payments are easier to embed: API, database, Agent settlement processes

This significantly simplifies reconciliation, approval, and system integration processes, which is especially important for entrepreneurs building Agent business ecosystems.

In terms of economic models, stablecoins also address the efficiency issues on both ends of credit card transactions: no 30-cent minimum fees, large transfers not eroded by interchange fees

Therefore, whether it's an Agent paying 0.001 USD per second in computational fees or a business settling a $50,000 vendor invoice, they can all use the same payment network.

Building More Stablecoin Infrastructure

A common criticism is the high onboarding and offboarding costs of stablecoins.

For "tourists," this is indeed an issue. However, when users are guided by an Agent acting as a "tour guide," this friction quickly diminishes.

Agents can assist users in fund exchanges, only executing necessary transactions to save costs. If we layer on billing and arbitration mechanisms, we are approaching a complete system.

Imagine a scenario where a user browses multiple brands in a department store, selects items, and only needs to check out once. The mall backend will distribute the funds to each merchant, and Agents would need a similar pattern. What the user sees is, "Your Agent wants to book flights, hotels, and rent a car for you," not three separate checkout processes.

The Agent platform is responsible for supplier relationships, while the user only needs to confirm the transaction intent.

Conclusion

An Agent doesn't pay like a tourist. They pay like a local, through relationships, trust, and long-term collaboration. This means that the true scale of future payments will flow through pre-negotiated B2B terms rather than card swipes.

But we are currently at a critical juncture. Agents are emerging, entrepreneurs are building new commercial systems, and they need payment tools they can use today.

Credit cards aren't ready yet: microtransaction costs are high, reconciliation is complex, technical debt is heavy, and they rely on manual risk management.

Stablecoins, however, are ready. They are programmable, global, easy to integrate, and can support Agent payments from day one.

Payment relationships exhibit strong path dependence. Once new business relationships are built on stablecoins, they tend to persist in the long term. In the coming years, as the ecosystem matures and fund transfers become smoother, a wave of startups will build new capabilities around this infrastructure: billing systems, arbitration mechanisms, credit systems, batch approvals, and cross-system interoperability.

Perhaps the new era of payments is starting right here.

[Original Article Link]