Oobit is expanding its stablecoin payments business into agentic commerce with the launch of AI Payments, a service designed to give artificial-intelligence assistants the ability to complete real-world purchases while limiting their access to users’ underlying funds and payment credentials.

The product, reported on October 9, allows an Oobit customer to connect an account to a compatible AI assistant and authorize that assistant to complete a specific transaction. Rather than exposing a conventional reusable card or giving the software direct control of a digital-asset wallet, Oobit generates a one-time payment card after the user approves the proposed purchase. The card is intended for that transaction and cannot function as an unrestricted payment credential.

The structure places Oobit at the intersection of two rapidly developing areas of financial technology: stablecoin-based consumer payments and AI agents capable of acting on behalf of individuals. AI assistants are increasingly able to search for products, compare prices, make reservations and navigate online checkout processes. What has remained more difficult is giving those systems the authority to move money in a way that is sufficiently controlled for consumers, merchants and financial-service providers.

Oobit’s model attempts to address that problem by separating payment authorization from account access. The AI assistant can receive the payment capability required to complete an approved checkout, but it does not receive the customer’s primary card number, unrestricted wallet access or general authority over the account balance. The consumer remains part of the approval process before money is committed.

Under the payment flow described by Oobit, a user first connects a compatible assistant to an Oobit account. When the assistant needs to make a purchase, it sends information about what it intends to buy, the merchant and the expected amount. The customer reviews and approves the request. Oobit then reserves the necessary funds and generates a single-use card credential for the transaction.

Once the payment is completed, the credential is no longer reusable. Transactions also appear in the customer’s Oobit account, giving users a record of AI-initiated spending alongside other activity. Oobit’s technical documentation says spending limits can be applied, users can revoke an agent’s access, and funds are set aside when a single-use card is issued. The system may also reserve a limited additional amount to accommodate expenses such as taxes or shipping.

Those controls are important because payment authorization represents one of the more consequential challenges in the development of autonomous AI services. A software agent that can locate a flight or choose a hotel has relatively limited financial risk if it only makes recommendations. The risk changes substantially once the agent can place an order, reserve inventory or transfer purchasing power.

A reusable card number or permanently connected wallet could expose a consumer to repeated transactions if an assistant makes an error, receives incorrect instructions or interacts with a malicious external service. Transaction-specific credentials offer one method of reducing that exposure because the payment instrument expires after its intended use.

Oobit is also using the launch to position stablecoins as a funding mechanism for AI-driven purchases. The company says supported digital assets include stablecoins such as Tether’s USDT. In markets where other funding options are available, customers may also be able to add money through local banking and payment systems, including bank transfers, mobile-money services and Brazil’s Pix network.

The distinction is important for the stablecoin industry. Much of the sector’s growth has historically centered on cryptocurrency trading, exchange settlement, remittances and cross-border transfers. Consumer spending has been more difficult because merchants generally do not want to integrate multiple blockchain networks, maintain crypto wallets or manage the volatility and compliance considerations associated with accepting digital assets directly.

Card-based conversion models seek to remove that requirement. The consumer can hold or fund an account using digital assets, while the merchant receives payment through familiar checkout infrastructure. In Oobit’s AI Payments model, the AI assistant effectively becomes another interface through which the customer instructs the payment service to spend those funds.

An AI payment interface shows a user approving a stablecoin-funded purchase completed through a one-time virtual card.

That could reduce the relevance of whether an individual online merchant directly accepts stablecoins. Instead, the critical question becomes whether the payment intermediary can transform a stablecoin-funded balance into a credential that works through the merchant’s existing payment setup.

For Oobit, the move also expands its role beyond a consumer application operated directly by account holders. The company is increasingly presenting its payment infrastructure as something that can be called by external software, including AI assistants. That shift resembles broader developments in embedded finance, in which payments, accounts and identity services are delivered through application-programming interfaces and integrated into third-party experiences.

Oobit says AI Payments can reduce the amount of financial infrastructure an AI developer must construct independently. Building a payment system for autonomous agents can require user identification, authorization controls, wallet or account connectivity, transaction monitoring, card issuance and merchant acceptance. By packaging those functions into a connection to Oobit, the company is attempting to become a financial execution layer for AI services.

The company says its integration can serve eligible users across more than 75 countries, subject to product availability and regional requirements. That geographic reach could be particularly relevant for developers trying to deploy the same AI assistant across markets with different banking systems and payment habits.

Oobit has named Muse, Grok, Dots and Instinct in connection with the AI Payments initiative. The Paypers reported that the rollout is beginning with selected AI assistants and partners, with broader integrations planned later. Oobit’s own product pages similarly indicate that availability can differ by assistant and that some integrations remain in rollout or coming-soon stages.

The distinction between a product announcement and broad production availability will be important for evaluating the commercial impact. Agentic payments remain a relatively early market, and the ability of AI systems to execute transactions at scale depends not only on payment technology but also on merchant compatibility, regulatory obligations, user adoption and confidence in automated purchasing.

Oobit’s Agent Pay documentation says the infrastructure uses the Model Context Protocol, or MCP, an open approach for connecting AI assistants to external tools and services. Compatible agents can use such connections to request services outside the underlying language model, allowing payments to become another callable capability alongside activities such as retrieving information or interacting with software.

This architecture reflects a broader change in how fintech companies may distribute financial products. Instead of expecting consumers to open a dedicated payment application every time they want to perform an action, financial functions could increasingly appear inside conversational interfaces. A customer could ask an assistant to identify a suitable hotel, compare available rooms, select one meeting stated requirements and then request approval for payment.

The user’s financial application would remain responsible for the controlled movement of money, but the AI assistant would handle more of the steps surrounding the transaction. That arrangement could compress product discovery, decision-making and checkout into a single conversational workflow.

For payments companies, however, the transition creates new questions around authorization. Traditional digital commerce assumes that a person manually enters card details, presses a purchase button or authenticates through a wallet. Agentic commerce introduces another participant between the customer and merchant: software acting under delegated authority.

Fintech providers therefore need mechanisms for distinguishing what an agent is allowed to do from what the user is allowed to do. Transaction limits, merchant restrictions, temporary credentials, confirmation requirements and revocation controls are likely to become central components of that infrastructure.

An AI payment interface shows a user approving a stablecoin-funded purchase completed through a one-time virtual card.

Oobit’s single-use-card approach provides one possible implementation. The agent does not receive a general payment instrument that can be used repeatedly. Instead, payment authority is generated when needed and constrained to an approved transaction. The structure does not eliminate every risk associated with AI commerce, but it reduces the financial exposure associated with handing a software agent a persistent card credential.

The launch also illustrates how virtual-card technology could gain a new use case beyond conventional expense management and online security. Virtual cards have long allowed companies and consumers to isolate payments, impose spending controls and replace reusable account numbers with temporary credentials. AI agents create a potentially much larger class of software-driven spenders for which those features are especially relevant.

Stablecoins add another layer to the model. Their relatively stable value compared with assets such as Bitcoin makes them more practical as a source of transactional funds, particularly when the objective is to purchase goods priced in conventional currencies. The use of USDT and other supported digital balances allows Oobit to market AI Payments to cryptocurrency users without requiring merchants themselves to redesign checkout systems around blockchain payments.

The potential market extends beyond retail shopping. AI assistants capable of purchasing could eventually handle travel bookings, software subscriptions, logistics services, recurring business expenses and other transactions where digital systems already perform much of the research and administrative work before payment occurs.

Business use cases could prove especially significant because companies already use virtual cards and policy-based controls to manage employee expenses, procurement and vendor payments. Similar controls could be adapted to AI agents, enabling a company to assign a defined budget to software and require human approval only for transactions above specified thresholds or outside approved categories.

Oobit has previously worked on programmable card infrastructure for AI-related spending, making the new consumer-oriented launch part of a broader effort to connect digital money with automated software. The October rollout moves that strategy closer to everyday consumer purchasing by combining stablecoin funding with one-time transaction credentials and user authorization.

The competitive landscape is likely to become increasingly crowded. Payment processors, card networks, crypto companies and AI-platform developers are all exploring ways to support transactions initiated by software agents. The strategic question is which companies will control the authorization layer between the assistant making a request and the regulated financial infrastructure moving the money.

Oobit is betting that stablecoin connectivity, geographic reach and transaction-specific cards can give it a position in that emerging layer. Its immediate challenge will be demonstrating that customers want to delegate purchases to AI assistants and that the authorization experience can remain simple without weakening financial controls.

The launch does not mean AI assistants have unrestricted authority to spend a user’s crypto. Oobit’s design emphasizes the opposite: the assistant can request a transaction, but the customer retains approval authority and the payment credential is created for a specific purchase. That distinction is central to the product’s positioning and to the broader debate over how autonomous software should interact with financial accounts.

If agentic commerce develops as AI companies expect, payment providers may increasingly compete not just for consumers or merchants but also to become the financial infrastructure used by software agents. Oobit’s AI Payments launch is an early attempt to define that relationship, using stablecoins as the source of funds and tightly scoped virtual cards as the mechanism through which AI gains permission to spend.