%201.webp)
Oobit has launched AI Payments for supported AI assistants, allowing users to approve purchases funded through their Oobit account with stablecoins.
The product launches with assistants including Muse, Grok, Dots, and Instinct. Eligible users in more than 75 countries can connect accounts through a single integration.
The system is designed to let an AI assistant complete a specific purchase without gaining ongoing access to a user's account balance or primary payment details. When a user asks an assistant to make a purchase, booking or payment, the assistant prepares the transaction and the user approves it, including a spending limit.
Oobit then generates a one-time virtual card for that transaction. The assistant uses the card to complete checkout, while the user's underlying funds and core payment credentials remain outside the assistant's control.
How it works
The launch comes as AI assistants move beyond search and recommendation tools into services that can take direct action for consumers. That shift has raised questions about how an assistant can pay for goods or services without being given unrestricted access to a bank card, wallet or account.
Oobit's model uses transaction-specific payment credentials rather than a persistent payment method tied to an AI agent. The aim is to reduce the risk of unauthorised spending, overspending and reuse of sensitive payment information by limiting each approval to a single checkout.
Funding can come from supported stablecoins and digital assets, including USDT. In some markets, users can also add funds through local payment methods such as bank transfers, mobile money and Pix.
"AI can already understand what you want and figure out how to get it. The missing piece is the ability to pay," said Amram Adar, Co-Founder and CEO of Oobit.
"That shouldn't depend on where you live or which bank card you have. Whether your money starts as USDT, Pix or mobile money, it should work with AI," Adar said.
Control and access
Oobit is positioning the product around user control over authorisation. Under the model, the assistant can request payment approval, but Oobit handles identity, authorisation, and payment execution.
That approach reflects a wider debate about how consumer safeguards should operate when software acts on a user's behalf in online commerce. Traditional online payments usually require direct access to a payment card, bank account or digital wallet, but AI-led transactions are creating pressure for more limited, specific permissions.
"The question isn't whether AI will be able to make purchases. It's how we make sure people remain in control when it does," Adar said.
"An assistant shouldn't need access to your entire wallet just to buy something you've already approved. We're separating the permission to complete a payment from access to the money behind it," he said.
Adar described the system as a way to tie one approved purchase to one payment instrument. "If you ask an AI to book a hotel, you want it to complete that booking, not have the ability to spend your money somewhere else tomorrow," he said.
"One transaction, one card, and a limit approved by the user. That's the principle behind AI Payments," Adar said.
Stablecoin use
The rollout also reflects efforts by digital asset companies to expand the use of stablecoins beyond transfers and direct wallet-to-wallet payments. By linking stablecoin balances to AI-assisted consumer purchases, Oobit aims to place digital assets into more mainstream buying flows.
"Stablecoins have already made it possible to move money digitally across borders. The next step is making that money useful in the experiences people actually use," Adar said.
"If someone holds USDT and asks an AI assistant to make a purchase, they shouldn't have to manually move money between different services just to complete a checkout," he said.
Oobit also said the system is aimed at developers building AI-led commerce services, where payment functions often require separate links to wallets, identity checks, transaction approval systems and card infrastructure. Its proposition is that a single integration can handle those functions for supported users and markets.
"For AI developers, the goal should be to build an assistant that solves a problem, not to become a payments company in the process," Adar said.
"We're providing the infrastructure that connects a user's money to an approved action, so developers can focus on what their assistants are actually built to do," he said.
Availability will depend on the supported assistant and the market. "AI commerce will only become mainstream if people trust it with real transactions," Adar said.
"That trust starts with clear authorization, limited access, and knowing that the assistant can only spend what you've agreed to. The user controls the money while the AI does the work," he said.