Meta News

By MSB

A recent experiment by Anthropic could mark the start of a profound transformation in the global economy: a market where people, but artificial intelligences, no longer do the buying.

According to the TechCrunch article, the company created an experimental marketplace where AI agents acted as buyers and sellers, closing real transactions with real money. What seemed like science fiction until recently is beginning to take shape as a new economic system: “agent-to-agent commerce” or agent commerce.

The project—known internally as Project Deal—was more than a simulation:

• Dozens of human users participated, but did not interact directly
• Each person had an AI agent with its own budget
• The agents negotiated, bought, and sold on their own
• Hundreds of real transactions worth thousands of dollars took place

In other words, humans defined the goals, but the machines executed the entire economy. This represents a radical change from traditional e-commerce.

This model is part of a broader concept known as “agentic commerce.”

Instead of manually searching for products, comparing prices, or reading reviews, an AI agent does all that on behalf of the user. But here, something more advanced happens: the agents don't just buy; they also negotiate among themselves.

Agents can coordinate, negotiate, and complete tasks without direct human intervention. There are even specific protocols for them to interact with each other, which turns the market into an automated, dynamic, and autonomous system.

In the classic model, the user enters a platform, searches for a product, compares options, and makes a decision. In the emerging model, the user simply defines a need, and their agent analyzes thousands of options, negotiates, and purchases automatically.

The most advanced scenario, already tested experimentally, is one in which buying agents negotiate directly with selling agents, adjusting prices and conditions, and executing the transaction without human intervention.

This model is not just a technological improvement, but a structural change in the economy.

Firstly, the consumer ceases to be the direct actor. The user delegates decisions to automated systems, transforming the role of the client in the market.

Secondly, competition becomes algorithmic. Companies will no longer compete only to attract people, but to optimize their products and prices to be selected by AI agents.

Furthermore, the speed of the market increases dramatically. Agents can compare thousands of products in seconds, negotiate in milliseconds, and operate continuously.

Finally, consumption becomes more rational. By eliminating emotional factors, decisions are based on data, preferences, and efficiency.

One of the most complex points of this new model is the payment system.

In this economy, the buyer is an agent acting on behalf of the user. There is no direct human interaction in every transaction, which forces a rethinking of authorization and trust systems.

To solve this problem, new approaches are emerging:

• Payment protocols designed for agents
• Delegated authorization systems
• Verifiable digital identity

The central question is how to ensure that a machine acts securely and aligned with the user's interests.

This model opens up opportunities, but also significant risks.

One of the main risks is automated fraud. Agents could manipulate prices, create fake markets, or execute malicious purchases on a large scale.

There is also the risk of identity spoofing, where an agent could impersonate another user or company.

Another potential problem are systemic failures. An error in an automated system could generate thousands of incorrect transactions in a matter of seconds.

Finally, there is the question of human control. Delegating economic decisions implies a greater technological dependency and a potential loss of direct control.

This model will not impact all sectors at the same time. The first to transform will be those that are the most digitized.

The technology and digital services sector will be the quickest to adopt this model, especially in SaaS tools and APIs.

E-commerce will also undergo significant changes, with marketplaces adapting to agent interaction.

Other sectors such as travel, booking, logistics, and supply chains will also be impacted, especially in automated purchasing and replenishment processes.

One of the most relevant aspects is that commerce becomes invisible to the user.

Agents will act as permanent intermediaries, managing purchases, payments, and decisions without the need for constant intervention. The user will limit themselves to defining objectives, while the system executes the necessary actions.

This implies that the economy will continue to function, but with less direct interaction between humans and markets.

The Anthropic experiment is not a mere technological curiosity, but a clear signal of where the economic system is heading.

We are entering a stage where machines do not only assist humans, but actively participate as economic actors.

This new paradigm redefines fundamental concepts such as consumption, decision-making, and competition. And it poses a central question for the future: whether humans will remain the protagonists of the economy, or if they will move into a more supervisory role in an increasingly automated system.