The World Cup gives the crypto industry a practical setting for watching wallets evolve. Topics that football fans discuss every day, such as whether a team will qualify or how championship odds are changing, become priced trading behavior inside prediction markets such as Polymarket. When mainstream Web3 wallets integrate World Cup prediction activities, the result is more than a sports-themed campaign. Over a longer timeline, it functions as a lightweight entry point for users to begin interacting on-chain.
imToken tests AI Agents in a World Cup prediction flow
In this World Cup-related exploration, imToken has started putting AI Agents into actual user scenarios. The agents it launched on the web and in Discord are designed around specific prediction needs, helping users complete betting-related transactions in a more natural way. Instead of only opening a wallet app to act, users can participate in prediction markets through Discord or a webpage, while the Agent guides them back into the on-chain process.
This is presented as an early form of the Agentic Wallet. The future Web3 wallet, in this framing, is not limited to a standalone wallet application. It becomes a broader AI wallet form that can appear across different contexts. For the past decade, the core questions that wallets handled were relatively clear: where assets are stored, who controls the private key, and who signs the transaction. A user opening imToken mainly wanted to check balances, transfer assets, interact with protocols, and manage assets across multiple chains. In that stage, the wallet was primarily an asset entrance and a signing entrance.
From a feature menu to an interpreter of intent
The Agentic shift changes the starting point. Users do not always know which button they should click. In a World Cup scenario, an ordinary user may not begin with a precise plan such as opening Polymarket, finding a specific market, evaluating the odds, and then completing a transaction. The user is more likely to ask questions in natural language: what is worth watching in tonight’s match; whether there is a market related to Portugal qualifying; whether the odds are already too low; and how the process works if the user only wants to participate with a small amount.
Previously, those questions were scattered across WeChat groups, social platforms, or search engines. Each user had to piece together the information and then execute the process step by step. Once an Agent enters the flow, the interaction changes. The user only needs to express a broad intention. The Agent breaks down the path, and the wallet turns that path into a series of on-chain actions. That is why an Agentic Wallet is not simply a wallet with an added chat box. The deeper change is that the wallet begins to move from being a feature menu to being an interpreter of user intent.
The World Cup is suitable as an entry point because it naturally carries context, encourages expression, and requires judgment. The Agent does not have to start by managing a complex asset portfolio for the user, especially because complex asset management carries higher risk. It can first help the user find an interaction path in a concrete scenario, while the final control remains with the wallet and the user. The web agent and Discord agent mentioned in connection with imToken are examples of this lighter entry model: wallet capabilities are brought into an event page and a community scene rather than staying only inside the app or a traditional DApp browser.
Wallet boundaries extend beyond the app
This also means the boundary of the wallet is expanding outward. In the older model, the entrance was relatively clear: users opened the app, entered the asset page, clicked a feature, and connected to a DApp. In the future wallet model described here, entrances can be distributed across more places, including webpages, Discord, Telegram, AI chat boxes, activity pages, developer tools, and even lightweight wallet interfaces generated by users themselves. From that angle, the World Cup prediction campaign is not the central point. The more important shift is that it lets the wallet stand in front of user intent in a more natural way.
If the discussion stays inside crypto, Agentic Wallet can be understood too narrowly as a narrative in which AI watches markets, finds opportunities, and makes trades for users. Mastercard’s launch of Agent Pay for Machines on June 10 moves the issue beyond Web3. Mastercard defines Agent Pay as a way for trusted AI Agents to participate in payments under user authorization. That includes how the Agent is identified, how it is authorized, how it is verified inside the payment network, and how merchants, issuers, and users know that a transaction was completed with Agent assistance.
The questions are highly similar to those faced by Web3 wallets. When AI is only helping someone write text, the cost of a mistake is usually manageable. Once AI begins to participate in asset interactions, the questions change: whether the AI has permission, whether it accurately understands the user’s intent, whether the service it calls is trustworthy, whether the transaction it initiates exceeds its boundary, and who handles the outcome if it differs from the user’s expectation. Mastercard’s answer is to redesign trusted Agent identity, tokens, authorization, risk control, and dispute handling inside the payment network.
Trusted agents and on-chain control
Domestic payment companies are also moving in the same direction. WeChat Pay is working with Tencent’s agent product WorkBuddy to test AI payment functions, including an AI-only card inside WeChat Wallet. Based on disclosed information, the core of these products is not to let AI spend freely. Instead, they set boundaries for agent payments through recharge limits, payment authorization scopes, and password verification confirmations. This follows the same logic as Mastercard Agent Pay: AI can participate in payment, but it has to be identified, authorized, limited, and audited.
Web3 wallets face the on-chain version of the same problem. The difference is that traditional payment systems emphasize networks, merchants, issuers, and compliance responsibilities, while on-chain wallets emphasize private keys, signatures, authorization, contract calls, and user self-custody. Agentic Wallet therefore cannot simply copy the path of traditional payments. In traditional payments, users can rely on banks, card networks, merchants, dispute processes, and risk-control systems. In the on-chain world, once a transaction is recorded, there is often no button for regret.
The more efficiency an AI Agent brings, the more important the wallet becomes as the final security boundary. The future wallet should not merely allow an Agent to do things; it must ensure that the Agent can only act within the range permitted by the user. This is why Web3 is a fitting environment for discussing Agentic Wallets. Many discussions of AI Agents move quickly toward extreme automation: AI trading automatically, managing wealth automatically, finding airdrops automatically, and arbitraging automatically. For wallets, however, the harder issue is not automation itself but the sense of boundary.
A wallet is not an ordinary application. If AI recommends the wrong song, writes the wrong paragraph, or opens the wrong page, the result is mostly an experience problem. In a wallet, if AI misunderstands an instruction, calls the wrong contract, or grants an oversized authorization, it can become an actual asset risk. The first-principles question of the Agentic Wallet is therefore not how many tasks AI can perform for the user, but how the user can know what the AI is doing.
This connects with imToken’s long-standing emphasis on control. From self-custody wallets to multi-chain asset management, and now to AI co-creation and Agent exploration, the continuous thread is not simply that more functions are added. It is that users should still be able to understand, confirm, and control their digital world. In the Agentic era, that thread becomes more concrete: the wallet needs to show who the Agent is, which capabilities it can call, how long the authorization lasts, whether it can operate across DApps, when the user must confirm again, and whether the user can pause or revoke it with one action.
The strength of an AI Agent comes from its ability to simplify complex processes. A single user sentence can be broken down into more than ten execution steps. That improves experience, while also creating a security challenge. The longer the path and the more intermediate links it contains, the more the wallet needs to bring critical nodes back in front of the user. The World Cup prediction activity is an event-based entrance, while imToken’s web agent and Discord agent are community-based entrances. Together, they show how small scenarios are becoming early tests for the Agentic Wallet form.