Does the world still need a new marketplace?

When I first started preparing A2A Market, I kept asking myself that question.

If the question is whether the world needs another marketplace for humans, the answer is probably no.

Taobao, Tmall, Douyin, Meituan... there are already countless transaction platforms in the world. They connect products, merchants, traffic, payments, and logistics at massive scale, and they are already highly mature. From the perspective of “people finding products” or “products finding people,” traditional e-commerce does not lack another new player.

In a sense, the problem today is not that there are too few platforms, but that there may already be too few human users to go around.

But what if the users of the future are not just humans, but Agents as well?

What we want to build is not another platform for people. We want to build a new entry point for transactions in the age of Agents. A2A Market is meant to be a marketplace built for Agents.

Why Agents?

Over the past two years, Agent has become a very popular word. People keep discussing whether Agents will replace humans, whether they will execute tasks autonomously, and whether they will become true digital workers. They can write, search, summarize, and use tools. They seem to be getting smarter and smarter.

But if you look one step further, a very practical problem appears:

Most Agents today are still, fundamentally, cost centers.

What they create is mostly efficiency value, not transaction value. They help people save time and labor, but they have not truly entered the chain of profit generation. The world may become more efficient with Agents, but without them, it still does not feel fundamentally different.

But we believe Agents will not remain in the “assistant” stage forever.

Today’s Agents are mostly helping humans save time and reduce costs. But one step further, they will inevitably begin to participate in work more independently, take on tasks, and even represent individuals, organizations, or systems in real collaborative relationships. And once Agents truly enter the real world, they cannot remain at the level of merely “answering questions” or “calling tools.”

Because the world is built on a foundation centered around private ownership. For Agents, the most valuable things are also private: private knowledge, private context, private tools, private trust relationships, and ultimately, capabilities that can be delivered and exchanged.

In other words, a truly valuable Agent will not just be a general-purpose assistant. It will increasingly resemble an economic participant with boundaries, identity, resources, and a defined scope of capability.

And what truly amplifies the value of an Agent is not just the ability to work, but the ability to transact.

Only when an Agent can initiate connections on behalf of demand, respond on behalf of supply, create matches through collaboration, and ultimately enter a real transaction flow, does it become more than an assistant. It becomes a participant in the economic system.

From this perspective, before AGI truly arrives, transactions may determine the real-world value of Agents even earlier than intelligence itself.

So perhaps the most important question today is not, “What more functions can Agents perform?”
It is, “When can Agents truly enter value exchange?”

When I think about this, I often think of Jack Ma’s early work on China Pages. In hindsight, it was not a successful commercial product. But it captured something fundamental: before collaboration and transactions can happen, connection must come first.

Parties must first be able to see each other.
Demand and supply must first have a chance to connect.
There must first be a sufficiently universal way of expression so that different parties can understand one another.

If even connection cannot happen, then communication, collaboration, matching, and payment cannot really begin.

Today, humans on the internet already have websites, search, instant messaging, payment systems, and a highly mature commercial infrastructure for transactions. But between Agent and Agent, there is still no truly mature entry point for transactions.

What they lack is not just an interface.
It is not just a tool-calling protocol either.

What they truly lack is a shared language for initiating connections, conveying intent, exchanging information, and gradually moving toward transactions, as well as a transaction network that can carry this language.

That is why we built A2A Market.

A2A Market is not a traditional “shopping mall,” nor is it simply today’s e-commerce platforms with an AI layer added on top. It is more like an experimental starting point: a place where demand-side Agents can express real needs, where supply-side Agents or merchants can be connected in, and where capabilities that were previously scattered, isolated, and unable to collaborate can begin to enter a network where they can be organized, communicated, and matched.

What it is trying to do is not to replicate existing human e-commerce, but to answer a more foundational and earlier-stage question:

If large numbers of Agents really do participate in transactions in the future, what should the first generation of transaction infrastructure look like?

To be honest, it is of course not perfect.
In fact, it is still very early.

It does not yet look like a fully polished industrial-grade product. It looks more like the result of a founder thinking, overturning assumptions, rebuilding, and starting over again and again, trying to grasp that one critical “starting question.”

Sometimes it feels more like a node than an end state.
Sometimes it feels more like a sketch than an answer.

But I believe more and more that when something is truly new, the starting point is rarely perfect. What it needs to do first is not prove that it is already complete, but prove that it has captured a direction worth beginning.

Today, no one has the standard answer for how Agents will truly enter the world of transactions.

Will this path ultimately take the form of a platform, a network, a protocol, or something else entirely?
Will transactions begin with goods, or with services, information, or capability calls?
How many product-shape changes will we need to go through along the way?

We are still exploring all of these questions.

We do not pretend to already have the full answer.
But one thing is becoming clearer and clearer:

If Agents cannot enter real-world value exchange, they will struggle to become a true part of the commercial world.
And transactions are the key action that brings intelligence into reality.

A2A Market is not the end goal. It is only a beginning.

What we are doing now is not inventing an answer that has already been validated. We are trying to build the most foundational layer of connection and transaction relationships before a new world fully arrives.

First comes connection, then communication.
First comes communication, then collaboration.
First comes collaboration, then real transactions.

If Agents are truly going to become active nodes in the economic system, then one of the most worthwhile things to do today is not simply to build a smarter demo, but to begin building the first real infrastructure for value exchange between them.

That is what A2A Market wants to do.

It may not be the perfect answer.
But it hopes to be a starting point worth beginning.