Author: jessy

Translated by: Jiahua, ChainCatcher

Over the past year, I have been dedicated to building infrastructure for the Agent economy, engaging with teams from Stripe, Visa, Coinbase, Google, and dozens of startups driving Agent business. I have mapped out the entire industry, launched products, and tried to find market fit.

Currently, there is no real demand, and startups face many structural issues when entering this field.

Last month, Stripe launched 288 new products at the Sessions conference, with access to their Agent documentation accounting for nearly 40% of total document readership. Their Agent business marketplace has over 1,000 active merchants. However, the number of registered Agents transacting at the Sessions conference was in single digits.

Visa mentioned that their Agent payment tokens (linked to Agents, used for tokenized payment on behalf of users) currently require 3 to 9 months of KYC approval, and one must actually achieve a minimum revenue threshold of $250 million to qualify. Nowadays, only companies of Amazon and Walmart's caliber can complete this verification loop.

Coinbase reported that as of April, there were 69,000 active Agents on the x402 protocol and 165 million transactions. However, independent on-chain analysis indicates that the actual daily transaction volume is around $17,000, of which about half are test transactions (according to CoinDesk March 2026 report).

Agent for Merchants

We built shop.fast.xyz to directly validate the real application of agent-style commerce. It includes real products, merchants, and transactions.

For most product categories, the current user experience of AI shopping falls short compared to traditional e-commerce. When you buy clothes, electronics, or furniture, you want to see pictures, browse various options, and make comparisons.

The dialog format of chatbots is actually a step backward. You are replacing a rich visual interface with plain text conversations, while humans are essentially visual shoppers.

Agents excel in aspects we originally thought would be difficult. They can understand user needs and handle requests like "something similar but cheaper." The model layer plays a role.

However, it cannot replace the experience of browsing ten products side by side and selecting one. The chat interface can be enhanced with carousel images and interactive displays, but to that extent, you are essentially just recreating an e-commerce frontend within a chat window. For visually-driven price comparison shopping, we have not found a convincing reason to prove that a chat interface is better than a native e-commerce interface.

We see real demand from merchants, but it is a defensive demand.

Merchants want their stores to be searchable by Agents. This is not because their current customers are buying through Agents, but because they worry that if this becomes the mainstream channel, they will be left behind by the times.

This is a strategy of "Agent Engine Optimization (AEO)," but currently, it's just an embellishment rather than essential. Merchants are preparing for a wave that has yet to arrive.

Conversational commerce can indeed enhance the experience in certain scenarios: high-frequency, low-decision-cost purchases where users already know what they want. Ordering takeout is the most obvious example. The market is huge, the frequency is extremely high, and the decision is quick ("Help me order pad thai from that place we last ordered from"). Conversational Agents have a chance here.

However, major food delivery platforms do not open their APIs. The only route is "computer use": letting AI operate applications through visual navigation just like humans. This approach is slow, fragile, and cannot handle the reasoning cost for a $15 lunch order.

Another breakthrough lies in the extremely complicated UI navigation of certain stores, which is quite painful. Layered discounts, promotional codes, loyalty programs, and confusing checkout processes.

An Agent that can understand "use my coupon, deduct my reward points, find the cheapest shipping, operate in my native language" can simplify today's experiences, which are extremely poor. This is particularly important for elderly users, non-native speakers shopping from distant online stores, or in niche scenarios with very specific needs.

Both breakthroughs require vast consumer-facing (B2C) distribution channels. You are competing for user gateways with DoorDash (the largest delivery platform in the US, holding 56% of the market) and Amazon.

Consumer-scale distribution is a stronghold for giants. The supply side of agent-style commerce is ready, but the demand side is constrained by user experience and distribution channels; building more infrastructure does not solve these two problems.

Agent for API

We discussed the actual payment needs with dozens of developers. The situation is almost remarkably consistent: the current Agent use of APIs is frequent, including calculations, reasoning, and data sources. Developers already have subscription services, archived API keys, and billing relationships with core suppliers.

A typical argument for stablecoins is: on Stripe, the minimum effective cost of credit card processing is about 2.9% plus 30 cents, making API calls under one dollar unprofitable. However, for today's low-frequency transaction volume, prepaid amounts can solve this issue. Developers recharge their accounts in advance, and the problem is resolved.

The deeper issue lies in the vendor market. Most mainstream SaaS companies do not want to provide pay-as-you-go API access that costs just a fraction of a cent. Their business models are based on multi-year enterprise contracts. Companies that rely on large commitment contracts will resist pricing mechanisms that bypass their existing models.

Machine commerce is structurally a long-tail market, including smaller services, niche data sources, individual developers, and MCP servers. Protocols like MPP and x402 are very well-suited for this sub-market.

But by definition, this is a market serving advanced users with special needs, and historically, developers have often been among the groups with the lowest willingness to pay.

When Stripe Projects were launched, they partnered with 32 vendor partners like Vercel, Supabase, Cloudflare, Twilio, etc., covering most of the tools developers use to build and deploy software, all accessible through the existing billing system. The top-tier demands in the developer tech stack have been met.

The opportunities for new payment channels lie in all areas outside these top 30 services: opportunities exist, but their scale is inherently much smaller than the impressive numbers suggest.

The same pattern applies to content acquisition. Agents are constantly scraping and summarizing articles, while publishers are fighting back.

But when content monetization comes on a large scale, it will be realized through CDN providers that are already positioned between publishers and the internet (Cloudflare has already launched AI audit tools for this), or through large licensing agreements between publishers and AI labs.

This opportunity in infrastructure will ultimately flow to those giants that already possess distribution channels.

Agent for Agent

The Agent for Agent business model is a long-term vision that currently exists almost entirely at the theoretical level, with no one realizing meaningful transaction volumes. Various startups are tackling the core challenges: agent discovery, trust-building, terms negotiation, and dispute resolution.

When this transactional structure is truly realized, it will be completely different from existing payment tracks. Neither party will include a human identity. The latency will be in sub-second levels. Funds ranging from a fraction of a cent to millions of dollars will operate within the same process.

Additionally, there is a multi-party settlement mechanism, which does not conform to the bilateral buying and selling model preset by existing payment tracks. Once this occurs, we believe it will come quickly and at large scale.

This is a long-term bet on dedicated settlement infrastructure, and it exists in reality. But "real long-term bet" and "current market" are two different matters.

For months, we have also been among those promoting this market and have built complete infrastructure around it over the past few years. With our distributed network, it can theoretically scale to over 1 billion TPS, with latency under 50 milliseconds and average consistency of 10 milliseconds. But we must align with the actual positioning of the market today.

Agent for Finance

This can be said to be the only category with existing demand. The customer base already exists and has a willingness to pay. Today, fund managers, finance teams, and DeFi users are paying for financial tools. Integrating AI into existing workflows is a natural product evolution.

Agent finance also creates entirely new behavioral patterns. An Agent that can autonomously monitor and rebalance hundreds of positions in real-time operates in a way that humans cannot replicate manually. This is not just automation, but a substantial enhancement of capability.

The challenge lies in the competitive landscape. The financial industry is heavily regulated and relies significantly on existing business relationships. Established institutions have licenses, compliance infrastructure, and customer relationships. Startups can seek a foothold in lightly regulated areas (like DeFi), domains where giants act slowly, or areas where AI can create capabilities that giants do not possess.

But compared to the other three categories, the competition dynamics here are more favorable to mature enterprises because layering AI onto existing products and customer bases is far easier than reverse engineering.

The Real Game Changer

So, why is everyone still building these things? There are two reasons.

First is the motivation. Industry giants have ample cash flow to bet on a future that may take years to manifest. For them, entering five years early is merely a rounding error, while entering a year late could be catastrophic. So they must build.

Second is the cognitive blind spot. When your core business is payments, every problem looks like a payment problem. If the Agent economy needs a payment layer, then build that payment layer.

But payment is just a part of a larger problem. The real challenge is not how to transfer funds between Agents, but how to coordinate work between Agents and humans, validate outcomes, and settle results. Payment is just a part of settlement. Settlement is just a part of coordination. And coordination is the real big pie.

Massive coordination will naturally spawn settlement mechanisms as a necessary demand. Payment is just one instrument in this symphony, not the whole composition. Companies solving coordination issues will subsume payment businesses, not the other way around.

Most legacy firms are engaging in defensive construction to prepare for scenarios of large-scale machine transactions in the future. Their funding runway is infinite, making timelines irrelevant to them.

But startups do not have that luxury. We must search for the true location of the market and cannot afford to wait for the wave to crash ashore.

A year of building has led us in an unexpected direction. There, market activity exists in reality, is growing rapidly, and has not been sufficiently serviced. It exists outside the four categories we have described.

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。