Animoca Brands has launched a live pilot with Visa that adds AI-driven commerce functions to its Minds platform. The pilot focuses on reward discovery and purchases at selected merchants in Hong Kong.
Minds has completed testing of two agent-led tools that use Visa Intelligent Commerce and a cardholder rewards and benefits directory. These tools let AI agents identify relevant Visa rewards and complete purchases on a user's behalf under preset conditions.
The first merchant in the pilot is the Bruce Lee Club eShop. The offering is aimed at users in Hong Kong, with rewards linked to eligible Visa Hong Kong and China cards.
How it works
Minds is Animoca's platform for persistent AI agents that users can deploy and manage without dealing directly with infrastructure. Through a marketplace called Bazaar, users can add tools, apps and other functions to agents, and set up multiple agents to work together on more complex tasks.
In this pilot, the new shopping tools extend AI activity beyond search and comparison into payment. The system relies on tokenised payment credentials, authentication checks, transaction controls and fraud protections within Visa's payments framework.
Users set the permissions for when an AI agent can act, how it can spend and under what conditions it can complete a transaction. This is intended to keep users in control while allowing the agent to handle parts of the buying process.
Rewards focus
A second part of the pilot focuses on rewards and card benefits. The rewards directory is designed to help an AI agent find offers relevant to a customer's eligible cards, stated preferences and intended purchase.
That could reduce the need for users to search manually through promotions or compare benefits across cards before buying. The agent can surface the most relevant options, assess potential value and incorporate them into the shopping journey before a purchase is made.
The collaboration reflects a broader effort by payments and technology groups to bring AI agents into consumer transactions rather than limiting them to advice and recommendation. It also points to growing interest in adapting existing payment rails for software agents acting under user instructions.
Chairman and Co-founder Yat Siu outlined the company's view of that shift.
"For most of the history of digital payments, innovation has meant making the same consumer journey faster or less expensive. Agentic commerce introduces a more fundamental shift by allowing an AI agent to act on a user's behalf within clearly defined permissions. Today marks a step toward that future, and reflects our view that users should have meaningful control over how AI participates in their digital lives, including how it helps them transact. Our work with Visa Intelligent Commerce acts as a crucial bridge, connecting the emerging agentic world with established human-centric commerce infrastructure," said Yat Siu, Chairman and Co-founder, Animoca Brands.
Wider strategy
For Animoca, the pilot adds a commerce use case to a business better known for digital assets, blockchain-related investments and consumer-facing Web3 projects. The group has invested in more than 600 companies and digital assets, while also building platforms including Moca Network, Open Campus, Anichess and The Sandbox.
The move also places Minds in a more competitive part of the AI market, where developers are trying to turn assistants into agents that can take actions as well as generate responses. In commerce, that means connecting recommendation systems to payment systems while building guardrails around spending authority, identity and fraud risk.
Visa's role in the pilot gives Animoca access to a payments brand with established consumer trust and merchant acceptance. For the payments sector, trials like this offer a way to test whether consumers are willing to delegate limited buying decisions to software agents when clear controls and visible protections are in place.
Hong Kong is the first market for the project, and the rollout is limited to selected merchants rather than broad retail deployment. That suggests a controlled test environment in which the companies can assess how users respond to agent-led shopping and rewards discovery before any wider launch.
The pilot centres on a simple proposition: an AI agent that can find a relevant card offer and complete the purchase if the user has already approved the terms. Whether consumers accept that model at scale will depend on how much trust they place in both the agent and the payment framework behind it.
The current pilot is built around user-defined controls and trusted payment credentials, with the AI agent acting only within pre-authorised preferences and spending rules.