AI developers may not prioritize cryptocurrency, yet insiders claim stablecoins are the key to empowering financial autonomy.

According to stablecoin professionals, the innovative realm of autonomous, micro-transacting AI agents is where programmable cryptocurrencies will excel.

Stablecoins are well-suited for agentic finance, according to industry insiders. (Gabriele Malaspina, Unsplash)

Key points:

• Utilizing stablecoins for millions of nano-payments significantly differs from employing ChatGPT as a front-end for e-commerce and connecting a credit card.
• Reconciling regulated money transfers with a multitude of agents and bots lacking financial identities necessitates programmable currency.

To grasp the potential impact of AI-driven commerce on the cryptocurrency sector, one should consult entrepreneurs and developers engaged in digital assets, especially stablecoins. They will readily assert that blockchain-based currency is an ideal match, a critical component in the ecosystem, and so forth.

Their reasoning is straightforward. In recent years, stablecoins—primarily digital representations of the dollar on public blockchains like Ethereum—have started to penetrate the global payments landscape. While they have demonstrated greater speed and lower costs compared to traditional banking transfers, it is within the emerging domain of autonomous, micro-transacting AI agents that they are expected to excel.

This perspective is held by companies such as Circle Internet (CRCL), the issuer of the second-largest stablecoin, and engineers at cryptocurrency exchange Coinbase (COIN), which has spearheaded the development of x402, a payments protocol intended for use by autonomous AI agents within a sector referred to as agentic finance.

As 24/7, seamless, global payment solutions have emerged as a growth sector for stablecoins, agentic commerce possesses specific requirements that the dollar-pegged tokens are designed to meet, according to Dante Disparte, Circle’s chief strategy officer and head of global policy. These include the capacity to program the coins for transfers that activate only when certain conditions are satisfied and to chain together, or compose, a series of actions triggered by the receipt of a token.

“Firstly, you must leverage the otherwise seemingly innocuous characteristics of stablecoins, which are programmability and composability,” Disparte stated in an interview. “Secondly, the location of the stablecoin, the actual blockchain ledgers, serves as the common reference point for the agents.”

Nevertheless, the cryptocurrency sector is often regarded with, if not outright skepticism, then at least caution by some AI developers. For instance, Peter Steinberger, the creator of the AI agent OpenClaw, is openly critical of cryptocurrency, to the extent that he refuses to provide further commentary on the topic and opted not to comment for this piece.

While the optimistic stance on AI from the crypto realm represents one end of the spectrum, the opposite viewpoint is also worth considering, said Sean Neville, co-founder of Catana Labs, a company developing infrastructure for agentic finance that raised $18 million in seed funding last year, led by a16z.

“I’ve collaborated with individuals from the AI development and engineering community who hold a rather low opinion of cryptocurrency,” Neville, who is also a co-founder of Circle, remarked in an interview. “I believe stablecoins have gained significant traction, but the AI developer community, in particular, tends to view crypto negatively due to factors like memecoins and Ponzi schemes, among other issues.”

Untouched by human hands

A significant aspect of agentic finance is that it encompasses micro-transactions, or nano-payments, some of which occur between AI agents while humans remain in the background.

This contrasts sharply with using ChatGPT as a front-end for e-commerce and linking a credit card, although in the short term, agentic systems will utilize both cryptocurrencies and cards, Neville noted. Agentic payments are likely to be frequent transactions in the fractions-of-a-cent range, which credit card networks may struggle to accommodate.

“Over time, I believe that stablecoins and blockchain infrastructure will offer considerable advantages for agentic flows beyond merely the retail commerce scenario,” Neville stated. “If AI is engaged in activities like utilizing 24/7, programmable systems to transfer various types of money globally and across borders, it becomes challenging to accomplish that with anything other than stablecoins.”

With clear regulatory guidance for stablecoins finally emerging in the U.S., there are potentially more urgent issues for AI agents regarding fragmentation and competing protocols vying for dominance, Neville mentioned.

“There are numerous methods for agents to facilitate payments among themselves, but if they cannot reach a consensus on how payments should function, it becomes difficult to establish marketplaces, regardless of whether they involve micro-payments,” he explained. “I would like to see something akin to an SSL equivalent arise for agents, and it would be beneficial to establish a standard that is not owned by anyone, enabling us to develop upon the same interoperable framework.”

SSL, or Secure Sockets Layer, is a standard technology that secures the connection between a web server and a browser.

The stablecoin-friendly option x402, frequently referenced in discussions, has led some to focus excessively on the protocol’s transaction volume fluctuations from month to month, noted Erik Reppel, head of engineering for Coinbase Developer Platform and an x402 co-founder. He emphasized his commitment to looking forward to an entire category of commerce that is set to significantly disrupt the existing online advertising market.

“I think what people haven’t fully grasped is that we are on the verge of altering the fundamental economic model of the internet, transitioning from browsers and visiting the websites of content publishers to consuming content through your agents and chat interfaces,” Reppel stated in an interview.

The modest amounts paid by an agent navigating a website, equivalent to the value of an advertisement displayed to a human viewer, could theoretically be achieved by generating numerous virtual cards, provided a developer has a partnership with, for instance, Visa, Rappel noted.

“However, anyone can program stablecoins,” he remarked. “Anyone globally can create as many wallets as they wish, using them to fully isolate funds for an agent. Our goal is to enable agents to have isolated, programmable funds, where your agent cannot exceed your credit card limit or access your credit card.”

Catena’s Neville stated that the company is working to align regulated money transfers with a multitude of agents and bots that lack financial identities. The objective is to filter out undesirable bots while identifying and permitting the ones you want, providing them with specific guidelines and policies they must adhere to.

“The approach to manage this is through programmable currency, as we can utilize cryptography to ensure verification, auditability, and so on,” Neville explained. “It effectively establishes identity and policy controls, allowing agents to function within the parameters, regardless of the protocol or wallet or account infrastructure employed.”