Ripple is making a direct push into AI-agent payments, aiming to place XRP and RLUSD into a market that is still largely centered on USDC. Earlier this week, the company released the XRPL AI Starter Kit, a developer toolkit designed for building AI agents that can send payments on the XRP Ledger.
According to release details shared with CoinDesk, the package includes access to XRPL documentation through an MCP server, which links AI tools to external data sources. It also comes with Claude skills for wallet creation, balance checks and payments, plus support for x402 payments using XRP and Ripple USD.
Ripple is targeting machine-to-machine payment flows
The company’s pitch is simple. If AI agents are going to buy API access, pay for model inference, settle invoices or move value across services, they need payment rails that are cheap, fast and easy to trigger without a human approving every transaction. Ripple says XRPL can offer three-to-five-second settlement, predictable fees, native payments, escrow, multisig support and a built-in decentralized exchange.
Those features matter most in a world of many small payments. If an agent is sending large numbers of low-value transactions, stable costs can determine whether the model works at all. Speed matters too. Once a payment is confirmed, the agent may need to move immediately to the next step. Ripple also points to XRPL’s native DEX as a possible advantage, since it could let one side send RLUSD while the recipient takes XRP, or the reverse, without routing through an external swap contract.
x402 has traction, but stablecoins still dominate
Ripple is entering a market that is still early, though the leading payment asset is already clear. The x402 protocol was created by Coinbase and is now stewarded by the Linux Foundation’s x402 Foundation. It reuses the HTTP 402 “Payment Required” code so machines can pay for online resources inside normal web requests. An agent requests a paid service, receives a payment demand, submits an on-chain payment and then resends the request with proof of payment.
That design makes payments feel closer to API calls, but actual usage remains fairly narrow. Most of the activity so far has been concentrated in stablecoins. A Chainalysis report published in early June said x402 activity on Base climbed from near zero in mid-2025 to more than 100 million cumulative transactions through the first quarter of 2026. Part of the late-2025 spike, though, was tied to PING, a pay-to-mint meme coin experiment that turned x402 payments into a speculative loop.
Public data from Web3 Trackers shows the protocol has now processed more than 120 million cumulative transactions, with over $41 million in settled USDC volume across 14 supported chains. The average payment size is about $0.05. Base accounts for roughly 70 million transactions and $21.5 million in volume, while Solana represents about 45 million transactions and $16.4 million.
Protocol-level payment functions may appeal to developers, but x402 adds its own risk layer
Ripple also highlighted what it described as an absence of smart-contract execution risk. On XRPL, payment functions are built into the protocol instead of depending on arbitrary contract code. That could matter for institutions and developers that want agent payments without relying on a fresh smart contract for every payment path.
The trade-off is that x402 introduces a different set of risks. A recent academic paper argued that the protocol creates new web-and-chain failure points around authorization and payment-service synchronization. A service might accept the wrong proof of payment, fail to match a payment with the correct request, or allow an old payment to be reused.
In practical terms, the website and the blockchain have to stay aligned on three points: who paid, what the payment was for and whether that payment is still valid. That is the market Ripple is now trying to enter, one where USDC holds the lead, transaction volume is already meaningful and the underlying coordination model is still being tested.