MoonPay has connected Kamino, the largest lending protocol on Solana, to PayBox, its AI-powered payment vault. Eligible users can now lend and borrow on Solana directly from an AI assistant, issuing the instruction through ChatGPT or Claude rather than a DeFi dashboard. CoinMarketCap flagged the integration on August 28, 2026.
The timing lands in a strong week for Solana. SOL traded at $109.46 as of August 28, 2026, up 8.3% in 24 hours and 24.1% over seven days, according to CoinMarketCap market data. That rally gives the launch an active backdrop, though the mechanics matter more than the price move.
The chat window becomes the interface
PayBox is MoonPay's attempt to package payments and, now, DeFi actions behind a conversational front end. Instead of opening Kamino, connecting a wallet, choosing a market, and confirming a transaction, an eligible user types a request into an AI chat and the vault routes it on-chain. Lending and borrowing are the first Kamino functions exposed this way.
The design leans on a pattern the industry has been circling for a year: agents and assistants that hold spending authority and execute financial instructions on a user's behalf. MoonPay is a regulated payments company with card and fiat rails, so wiring a lending protocol into that stack pushes the assistant from "answers questions" toward "moves money."
Kamino is the substance behind the feature. It is Solana's leading lending venue by deposits, offering supply and borrow markets across major assets. Plugging it into PayBox means the AI layer is not simulating yield, it is placing real positions in a live protocol.
Convenience and the risk it hides
Borrowing through a chat prompt removes friction, and removing friction from a leveraged action cuts both ways. A borrow on Kamino is a collateralized loan with a liquidation threshold. If collateral value falls far enough, the position gets liquidated on-chain, and a conversational interface does not change that math. It can make the action feel lighter than it is.
There is also a trust question. Handing an AI assistant the authority to lend and borrow assumes the routing is correct, the market selection is sound, and the user understands the position being opened. The interface abstracts away the dashboard, but the counterparty risk of the protocol, the smart-contract risk of Kamino, and the price risk of the collateral all remain. Users spending or stablecoin balances through connected wallets should treat an AI-issued borrow with the same caution as a manual one.
For readers who prefer to keep control of keys, this is a reminder of why spending from your own wallet and understanding each on-chain action still matters, even when the action is one sentence away.
The direction of travel for AI payments
MoonPay is not alone in bolting DeFi onto an assistant, but a regulated payments provider doing it changes the shape of the trend. Payment companies own the fiat on-ramp, the card, and increasingly the AI layer that decides where money goes. Kamino inside PayBox is a small feature with a large implication: the same chat that helps a user buy crypto can now put that crypto to work in a lending market.
The near-term audience is narrow. Access is gated to eligible users, and the first functions are lend and borrow rather than a full suite of DeFi actions. Solana's crypto card ecosystem already routes real spending on-chain, so an AI vault that also manages lending on the same network fits the direction of travel. The open question is how much authority users are comfortable delegating once the assistant can both hold funds and deploy them.
Overview
MoonPay integrated Kamino, Solana's largest lending protocol, into PayBox, its AI payment vault, letting eligible users lend and borrow on Solana through ChatGPT or Claude. CoinMarketCap reported the move on August 28, 2026, during a week when SOL rose 24% to $109.46. The feature collapses a multi-step DeFi flow into a single chat instruction. That convenience does not remove liquidation risk, smart-contract risk, or the price risk of collateral, and it raises a broader question about how much financial authority users will hand to an AI assistant that both holds and deploys their money.
