AI agents’ real-world spending could boost XRP

Chandler Fang, co-founder of AI agent trust startup t54, wrote on X on Aug. 21 that AI agents shifting from buying digital services to making real-world purchases under user-defined budgets could increase use of XRP on the XRP Ledger (XRPL).

Fang outlined an agent economy with markets defined by payment size, frequency and settlement needs. He noted agents have already completed more than one million transactions across XRPL without human intervention. Fang co-founded t54 after a tenure as product lead at Ripple and earlier roles in quantitative finance and blockchain at J.P. Morgan. Investors in t54’s $5 million seed round included Ripple and Franklin Templeton.

In the near term, Fang described agents automatically buying digital resources such as APIs, computing time and model inference. Ripple’s XRPL AI Starter Kit, released June 10, implements x402 payments using XRP or Ripple USD (RLUSD) so autonomous software can pay for online services. Mastercard launched Agent Pay for Machines on June 10 with more than 30 launch partners, including Coinbase, Stripe and the Solana Foundation. Ripple joined that program through RippleX, creating a framework that links crypto settlement rails and traditional payment processors for automated purchases.

Fang outlined a later stage in which users give agents limited spending authority through mandates. Those mandates would set budgets, restrict which merchants agents may use and require human approval for specific choices. Consumers and businesses would need crypto wallets and controls that define permitted purchases and spending limits. Merchants would need to present prices, inventory and delivery terms in formats agents can read. Over time, agents with reliable identity, transaction and repayment records could receive financial capacity or credit instead of being pre-funded for each task.

Not every automated action would become an on-ledger XRPL transaction, Fang noted. Repeated payments between the same parties could use XRPL Payment Channels to exchange signed claims off-consensus and settle net obligations on the ledger. He described three adoption phases: frequent, low-value payments for digital services; fewer, higher-value consumer and business transactions; and many internal agent actions before a single settlement for embedded use cases. Platforms could aggregate activity or settle net positions rather than recording each step on the ledger.

The choice of payment asset would depend on transaction purpose and settlement needs. XRP could serve for direct payments, routing and final settlement, while RLUSD could provide dollar-denominated value for purchases that require price stability. Existing analysis treats XRP and RLUSD as complementary tools within XRPL.

Fang wrote that long-term adoption may run largely out of view of end users, with agents operating inside treasury platforms, logistics systems and financial apps and selecting blockchain infrastructure in the background. “Long-term adoption does not require every person to open an XRPL wallet or consciously choose a blockchain before completing a task,” he wrote.

He added that transaction volume alone would be an incomplete gauge of agent adoption and recommended measures such as repeat activity, merchant revenue, compliance with spending mandates, successful delivery and dispute resolution. “By then, XRPL will not be judged only by whether agents can use it. It will be judged by how much useful economic activity they choose to settle on it,” Fang wrote.

Fang compared the pattern to past rollouts of consumer-facing AI products, saying years of infrastructure work can appear sudden once a product reaches wide use. He predicted users will come to expect agents to operate within defined limits and that adoption will reveal which XRPL capabilities need to scale next.