As AI agents continue to grow in popularity, there’s one question everyone is wondering: When will they achieve full autonomy? Bets are being made soon on the recently launched foundation.
In September, Cloudflare and Coinbase introduced the x402 Foundation, which supports the open payments standard x402. The protocol, named after the HTTP 402 Payment Required Status Code, is a payment standard for online-based exchanges, paving the way for the digital dollar to become a viable currency and opening the possibility for agents to be equipped with wallets so that they can make autonomous payments.
The open standard also enables individuals or organizations with online-based services, such as API developers or content creators, where the revenue paradigm has been turned upside down by features such as large language models and AI observation, the opportunity to charge for their products.
“Businesses are starting to adopt x402 now because (agents) are scraping content,” said Dan Kim, vice president of digital asset listings and services at Coinbase. He said that having an agent as part of the standard is beneficial to content creators, but it also allows users to purchase products online without revealing their credit card numbers.
Options currently exist that enable generic AI systems to make purchases on behalf of users such as ChatGPT Shopping or ChatGPT Instant Checkout, but Kim pointed out that a human would have to be in the loop and these systems are built around credit cards, which could be compromised.
Digital dollars or stablecoins, on the other hand, offer a payment form that currently prevents payment fees and chargebacks.
“It’s not just, ‘Hey, I can do this with a credit card.’ It’s like, ‘Okay, what can you not do with a credit card and what can you do with a digital dollar that is going to make that experience much more interesting, much more powerful, much more engaging, much more scalable, in a world where people are already going to search engines far less, and by default to ChatGPT or Gemini,” Kim said. “It’s not just to be able to buy things, but to be able to allow agents to become autonomous buyers and sellers in a world where humans don’t actually have to be in a loop and don’t have to be in a loop.”
Kim gave an example of travel planning. If he interfaces with a larger language model to develop an itinerary, he will have to take additional steps to book a flight or hotel and may miss out on any deals the AI encountered. If Kim could instruct an agent to make payments on her behalf, the experience would be completely different, he said.
However, even without a credit card, the risk of agents working on the human side is high, which is why there is a need for a trusted agent registry and platforms like AWS where developers are building agents they can trust, Kim said.
“What this agentic economy really needs to thrive is not just the existing technology, but also this idea of an AI registry, where as an operator or owner operator of a site or owner of content, you say, ‘Look, I’ll serve agents, either sell my stuff or my data, but they need to be trusted,'” he said. “It’s growing, but does that still exist today? No, not yet. But it is happening.”
