Artificial intelligence is beginning to move beyond answering questions and generating content toward performing practical tasks on behalf of users.

One emerging application involves automated payments, allowing software agents to access digital services and complete small transactions without requiring a person to handle every step manually.

How the Payment Experiment Works

The test brought together three principal components: HSBC's Tokenised Deposit Service, Ant Digital Technologies' Anvita Flow network and the Jovay Testnet blockchain environment.

Each component performed a distinct function. HSBC supplied tokenized deposit capabilities and settlement infrastructure, including real-time risk checks through its interface for connecting software agents with external tools. Anvita Flow enabled artificial intelligence agents to discover digital services, interact with them and coordinate micropayment settlement. Jovay Testnet provided the blockchain environment in which the transactions were tested.

Micropayments are generally small transactions, often defined as payments below US$2 in this context. Such transactions could become relevant when digital services charge individually for specific actions rather than requiring subscriptions or larger bundled payments.

Why Tokenized Deposits Matter

Tokenized deposits are digital representations of claims on commercial bank deposits recorded through a token-based technological system. Unlike a generic digital token, a tokenized deposit remains a bank liability and operates within the relevant banking arrangements.

Tokenization can make financial assets or claims easier to integrate with programmable platforms. Transactions can be linked to software instructions, enabling certain operations to occur automatically when predefined conditions are satisfied.

For example, a digital service could potentially charge a small amount whenever an authorized software agent retrieves information or completes a defined task. With suitable infrastructure, service access and payment settlement could be coordinated within one automated workflow.

This capability may reduce the need for separate manual steps and make small-value transactions more practical. However, the actual benefits depend on transaction costs, system compatibility, processing reliability and the safeguards surrounding each payment. Tokenization alone does not guarantee instant settlement, eliminate fees or remove the need for conventional banking controls.

Potential Applications for AI-Driven Payments

Automated micropayments could support several emerging digital business models. One possibility involves software agents purchasing access to specialized information. An AI system performing research might retrieve individual datasets or use a paid digital tool, with payments processed according to predefined permissions and pricing rules.

Another application could involve business software that purchases computing resources, accesses external services or pays for individual processing tasks. Instead of requiring a human operator to approve every small transaction, the system could execute authorized payments within established limits.

Digital marketplaces could also benefit from more flexible pricing arrangements. Service providers might charge according to actual usage, allowing customers to pay for specific functions rather than purchasing an entire subscription.

Security, Authorization and Operational Challenges

Allowing software agents to initiate payments introduces important questions about control and accountability. A payment system must distinguish between an agent that has legitimate authorization and one attempting to perform an unintended or unauthorized transaction. Establishing spending limits, verifying service identities and monitoring unusual activity can help reduce the risk of misuse.

Real-time risk checks are therefore an important part of the architecture. Nevertheless, automated screening should not be treated as a guarantee against fraud, software errors or compromised credentials.

Clear rules are also necessary for handling failed transactions, duplicate payments, disputed charges and interruptions in service. When an automated agent acts incorrectly, responsibility must be clearly defined among the customer, service provider, technology operator and financial institution.

Expert Insight

Pablo Hernández de Cos, General Manager of the Bank for International Settlements, highlighted the potential advantages of tokenization in a speech delivered in August 2026: “Tokenisation offers real gains: programmability, atomic settlement and around-the-clock operations.”

What Comes Next for Automated Payments?

The HSBC and Ant Digital demonstration provides a technical foundation for examining how AI agents might interact with digital services and payment infrastructure. Further development would need to address customer authorization, operational resilience, dispute handling, transaction economics and integration with existing financial systems.

HSBC and Ant Digital Technologies have demonstrated how artificial intelligence agents, tokenized deposits and blockchain infrastructure can be combined to support automated micropayments. The demonstration is therefore best understood as an early technical step toward more programmable payments, rather than evidence that fully autonomous financial transactions are already ready for widespread customer use.