The World Cup is a useful lens for observing how wallets are evolving.
Whether a team will qualify or how championship odds are changing — topics fans discuss every day — can become tradable, priceable events in prediction markets such as Polymarket. In the longer run, World Cup prediction activities integrated into mainstream Web3 wallets can indeed serve as a lightweight starting point for users to try on-chain interactions.
Further reading: “World Cup Fever Propels Prediction Markets: How Polymarket and Peers Are Driving Mainstream Crypto Adoption”
At the same time, another more nascent, yet highly imaginative shift is worth watching: as AI Agents begin entering wallet use cases, the way users interact with the on-chain world may also start to change.
For example, in its World Cup-related exploration, imToken has started putting AI Agents into real use cases. Its web-based and Discord Agents can help users act on specific prediction needs and complete prediction-related trades more naturally. Users no longer have to complete every step inside the wallet app. Instead, they can participate in prediction markets through Discord, web pages, and other platforms, while being smoothly guided back on-chain by the Agent.
This may well be an early form of the Agentic Wallet: the Web3 wallet of the future may not be limited to a wallet app, but it will likely take the form of an AI-powered wallet experience that is available almost everywhere.
1. The World Cup Experiment: When AI Agents Begin to Understand “Intent”
Let’s return to a familiar topic. Over the past decade, the core problems wallets solved were clear: where assets are stored, who controls the private key, and who signs transactions.
For example, when users open imToken, they mainly want to check balances, make transfers, interact with DApps, and manage multi-chain assets. In this stage, wallets have functioned more like gateways to assets and signatures: as long as users know what they want to do, the wallet helps them complete the final step.
But in the Agentic era, the change is that users may not know exactly which button to tap at the very beginning.
In a World Cup scenario, for example, an everyday user may not start by thinking, “I need to open Polymarket, find a specific market, assess the odds, and complete the trade.” More likely, they may ask: “What should I watch for in tonight’s match?” “I think Portugal will qualify. Are there any related markets?” “Are these odds already too low?” “If I only want to participate with a small amount, how does the process work?”
In the past, these questions may have been scattered across Telegram groups, social platforms, or search engines. Each user had to piece the information together and then execute each step on their own. Once an Agent is involved, however, the interaction changes significantly: users only need to express a general intent, and the Agent can proactively break down the path, while the wallet turns that path into a series of on-chain actions.
So this is certainly not as simple as adding a chat box to a wallet.
The real change is that wallets are beginning to shift from “function menus” to “intent interpreters.” In the past, wallets mainly asked users to decide whether they wanted to transfer, swap, stake, or connect to a DApp. In the future, wallets may go one step further: users may only need to describe what they want to accomplish in natural language.
This is also why a mainstream event like the World Cup is a suitable entry point for Agentic Wallets. It naturally comes with context, gives users something to talk about, and requires them to make decisions. An Agent does not need to start by managing complex portfolios for users — especially given the higher risks involved in complex asset management. It can first help users find an interaction path within a specific context, and then return final control to the wallet and the user.
The imToken web-based and Discord Agents mentioned earlier are a typical example. They bring wallet capabilities into lighter entry points, allowing users to find an interaction path through an Agent on an event page or in a World Cup scenario, without first opening the app or entering a traditional DApp browser.
This means the boundaries of wallets are expanding outward.
In the past, the wallet entry point was relatively clear: users opened the app, entered the asset page, tapped a function, and then connected to a DApp. In the future, wallet entry points may be scattered across more places: web pages, Discord, Telegram, AI chat interfaces, event pages, developer tools, or even lightweight wallet interfaces generated by users themselves.
From this perspective, World Cup prediction itself is not the main point. The real point is that it allows wallets, for the first time, to position themselves more naturally around user intent.
2. Agent Pay Shows That AI Is Entering the Payment Layer
If we look only within crypto, Agentic Wallets can easily be understood as another narrative: AI helps users read the market, find opportunities, and make trades. It sounds like a continuation of the previous wave of AI Agent hype.
But Mastercard’s launch of Agent Pay for Machines on June 10 suddenly made this more than just a Web3 story.
Mastercard’s definition of Agent Pay is clear: it allows trusted AI Agents to participate in payments under user authorization. This includes how an Agent is identified, how it is authorized, how it is verified within a payment network, and how merchants, issuers, and users can know that a transaction was completed with the assistance of an Agent.
This is highly similar to the challenge Web3 wallets are facing.
When AI only helps you write copy, the cost of mistakes is usually manageable. But once AI begins to participate in asset interactions, the problem changes: does it really have permission? Has it understood the user’s intent correctly? Is the service it calls trustworthy? Does the transaction it initiates go beyond the allowed boundary? If the result differs from what the user expected, who is responsible?
Mastercard’s answer is to redesign the identity, tokens, authorization, risk controls, and dispute handling for “trusted Agents” within the payment network.
This signal matters. If Agentic ideas in Web3 still carry a hint of geek imagination, then a traditional financial giant designing payment infrastructure for Agentic Commerce shows that this shift has already entered a more practical business context.
A closer example can also be found among Chinese payment giants. WeChat Pay is working with Tencent’s AI agent product WorkBuddy to test AI payment features, such as an “AI-only card” in WeChat Wallet. Based on the information disclosed so far, the core idea is not to let AI spend money freely, but to set boundaries for agent payments through prepaid limits, payment authorization limits, and password or PIN confirmation.
This follows the same logic as Mastercard Agent Pay: AI can participate in payments, but it must be identified, authorized, restricted, and auditable.
Web3 wallets are facing the on-chain version of the same problem. The difference is that traditional payment systems place more emphasis on networks, merchants, issuers, and compliance responsibilities, while on-chain wallets place more emphasis on private keys, signatures, approvals, contract calls, and user self-custody.
Precisely because of this, Agentic Wallets cannot simply copy the traditional payment path. In traditional payments, users can rely on banks, card networks, merchants, dispute handling, and risk control systems. But in the on-chain world, once a transaction is recorded on-chain, there is often no undo button.
The more efficiency AI Agents bring, the more important the wallet becomes as the final safety perimeter. The wallet of the future should not only allow Agents to “do things”; it must also ensure that Agents can only act within the scope users have allowed.
This is also why Web3 may actually be a more suitable place to discuss Agentic Wallets.
3. In the Agentic Era, How Should Wallets Redefine On-Chain Interaction?
When people discuss AI Agents, they often instinctively move toward an extreme vision: in the future, AI will automatically trade for me, manage my assets for me, find airdrops for me, and arbitrage for me.
This direction is certainly appealing. But for wallets, the truly difficult part is not “automation”; it is maintaining clear boundaries.
A wallet is not an ordinary app. In an ordinary app, if AI recommends the wrong song, writes the wrong paragraph, or taps the wrong page, it is mostly an experience issue. But in a wallet, if AI misunderstands an instruction, calls the wrong contract, or grants excessive permissions, it may lead to actual financial loss.
Further reading: “Sign Is More Than a Signature: When an AI Agent Signs for You, Who Is Still in Control?”
So the first-principles question for Agentic Wallets is not “How much can AI help users do?” It is “How can users know what AI is doing?”
This is also why imToken has emphasized control over the past few years. From self-custodial wallets to multi-chain asset management, and now to AI co-creation and Agent exploration, the continuous thread has never been simply “more features.” It has been that users should always be able to understand, confirm, and control their own digital world.
In the Agentic era, this thread becomes more concrete.
Wallets need to help users understand who an Agent is, what capabilities it can call, how long the authorization lasts, whether it can operate across DApps, when the user needs to confirm again, and whether the user can pause or revoke it with one tap. These questions may sound tedious, but they are the foundation that determines whether Agentic Wallets can truly work.
The power of AI Agents comes precisely from their ability to simplify complex processes. A user says one sentence, and the Agent may break it down into more than a dozen execution steps. This is good for user experience, but it is also a security challenge. The longer the path and the more intermediate steps involved, the more wallets need to bring key checkpoints back in front of the user.
A good wallet interaction in the future may not be about showing users more technical details. Instead, it may be about translating complex transactions into language users can understand:
- You are authorizing an Agent to call a specific contract within the next 24 hours.
- This action can use up to a specified amount of USDC.
- It can only access World Cup-related prediction markets and cannot touch your other assets.
- Any transaction above a certain amount must be confirmed again.
- The authorization will automatically expire when the time limit ends.
- You can pause this Agent in the wallet at any time.
This may sound like a distant future, but it is already starting from small scenarios. World Cup prediction activities are an event-based entry point, while the web-based Agent and Discord Agent that imToken is testing are community-based entry points.
At moments like this, wallets should not retreat into the background.