OKX 钱包对话 Vitalik Buterin：人工智能代理自主交易后钱包、L2 和身份系统将如何演变？ OKX Wallet遵守 |此前，OKX钱包推出OnchainOS，不断迭代开放Agent能力，同时推出全新的Agentic钱包。当人工智能成为新的“用户界面”，当智能体开始作为链上参与者自主交易、参与治理时，以太坊等区块链的角色将发生怎样的变化？这就是《OKX钱包之友》系列——嘉年华版。本系列通过与不同建设者的对话，记录他们在行业关键时刻的判断和反思。本期OKX钱包VP Paul Wan将与Vitalik Buterin对话，共同探讨AI和Web3的长期趋势和底层结构，探讨区块链在Agents时代将扮演什么角色。AI时代以太坊和区块链的角色演变问题一：面对Agents时代，链上操作系统必须提供哪些原语？今天以太坊还缺少什么？ Vitalik Buterin：这就是我的看法。以太坊
主要为应用程序提供两大功能。其中之一是公告板，这意味着任何人都可以在链上发布数据，然后各种应用程序可以以多种不同的方式解释这些数据。另外一个是链上计算和链下计算，包括金融应用、DeFi以及其他各种应用。在AI时代，从根本上来说，这些用途本质上是一样的。相同的用例仍然存在，并且这些功能可以继续保持其重要性。然而，人工智能将不可避免地导致我们与区块链和其他工具交互的方式发生巨大转变。其中一个重要的区别是，在前区块链时代，用户通过特定的界面进行交互，而该界面对应于特定的任务。但在
OKX Wallet in Conversation with Vitalik Buterin: How Will Wallets, L2s, and Identity Systems Evolve After AI Agents Trade Autonomously?
Author | OKX Wallet
Complied by | WuBlockchain
Previously, OKX Wallet launched OnchainOS, continuously iterating to open up its Agent capabilities, and simultaneously introduced the brand-new Agentic Wallet. As AI becomes the new “user interface,” when Agents begin to act as on-chain participants to trade and participate in governance autonomously, how will the role of blockchains like Ethereum change?
This is the “Friends of OKX Wallet” series — Carnival Edition. Through conversations with different builders, this series records their judgments and reflections at critical industry junctures. In this episode, OKX Wallet VP Paul Wan will speak with Vitalik Buterin to discuss the long-term trends and underlying structures of AI and Web3, seeking to understand: what role will blockchain play in the era of Agents.
The Evolution of the Roles of Ethereum and Blockchain in the AI Era
Question 1: Facing the era of Agents, what primitives must an on-chain operating system provide? What is Ethereum still missing today?
Vitalik Buterin:
This is how I see it. Ethereum primarily provides two major functions for applications. One of them is a bulletin board, meaning anyone can publish data on-chain, and then various applications can interpret this data in many different ways. The other is on-chain computation and off-chain computation, which includes financial applications, DeFi, and various other applications.
In the AI era, fundamentally speaking, these uses essentially remain the same. The same use cases still exist, and these functions can continue to maintain their importance. However, AI will inevitably cause a massive shift in how we interact with blockchains and other tools.
One of the important differences is that in the pre-blockchain era, users interacted through a specific interface, and that interface corresponded to a specific task. But in the world of AI, especially the form of AI we see now and in the coming years, you can have a client-side AI invoke these different skills, combine tasks all at once, and simultaneously interact with many different objects.
This also significantly increases the number of workflows interacting with Ethereum and other systems. So, in my opinion, the metaphor of an “operating system” is not entirely apt from certain perspectives. Operating systems will still continue to exist, but they will become smaller and simpler. At the same time, we will have a bunch of different tools and skills, with AI helping users to utilize them and get things done. Blockchain is a natural choice; it allows multi-party collaborative applications to exist and enables different participants to collaborate effectively over the long term, without the need to establish trust or reach a consensus in advance.
Another key point is the Economic Layer. If AI becomes more decentralized, there will be many different AI entities built and controlled by different people, and they will need to interact with each other. To make this interaction possible, an economic layer is required. This is because cooperation is either based on economic incentives and rules, or based on centralized control; fundamentally, these are the two paths.
If we can build this economic system, it will better enable AIs to interact with each other in a decentralized manner. Just like a complete operating system has a runtime as well as software and infrastructure built on top of it. In a new economic system centered around Agents, we need a mechanism to discover, define, and match suitable Agents. Meanwhile, users and their Agents can also participate, interact with each other, and build their own Skills, MCP, CLI, and strategies.
Question 2: In high-frequency Agent trading scenarios, how should we view L2?
Vitalik Buterin:
L2 is very important, but we need to be more imaginative in the way we build L2s. Past approaches often involved simply replicating the EVM and scaling it, but this method is not ideal. A better approach is to start from application needs and supplement the capabilities that L1 does not provide. In an ideal scenario, different functions should be distributed across different layers: accounts can be placed on L1, while high-frequency trading and matching can be placed on L2.
Furthermore, L2 can also take on privacy functions, such as Tornado Cash, Railgun, Privacy Pools, etc. In a sense, these can be viewed as “privacy-focused L2s.” In the future, more L2 solutions developing in different directions will emerge.
Restructuring the Relationship Between Humans and Agents
Question 3: When Agents can autonomously trade, hold assets, and even participate in governance, how should we redefine the “user”? Particularly from the perspective of on-chain governance, how do you think mechanisms should be redesigned for these non-human participants?
Vitalik Buterin:
I personally still view humans as the users, and view AI as a replacement for the UI; it is a new way for humans to interact with the chain.
It can be understood this way: before AI emerged, you might have needed to obtain information through multiple tools like Google, Wikipedia, and Stack Overflow. But now, you can directly ask an AI a question, and it will complete the operation and provide the result. This kind of change will also soon happen in blockchain interactions.
This means that our perspective on the attributes of infrastructure, such as latency, will change. In human interaction, low latency is usually very important; but for Agents, some scenarios require extremely low latency, while in others, latency is actually not that important — for instance, complex problems can wait for a longer time.
This difference will change the way we think about the interface layer. Take wallets as an example: what remains unchanged is the SDK, which is the API layer (such as transfers, queries, and privacy operations); but the product formats surrounding it may change. Users might even no longer use wallets directly, but instead interact with the SDK.
Therefore, we will see a series of carefully polished software packages that possess strong security and formal verification capabilities, equipped with Skills files, to be invoked by AI. So I believe that the layer between the blockchain and the user will undergo a very significant change.
Question 4: If the actor in a transaction could be a human, an Agent, or a combination of both, how should we rethink on-chain identity? Is it possible to build a unified framework that applies simultaneously to humans, Agents, and mixed scenarios?
Vitalik Buterin:
The key to establishing on-chain identity is disaggregating identity, only proving the necessary information required to complete a certain interaction. In most cases, fully exposing an identity is meaningless; a more reasonable approach is to reveal only partial information, such as proving reputation or the source of funds through zero-knowledge proofs.
At the same time, it is necessary to make on-chain behaviors and assets easier to be proven via zero-knowledge proofs, and to let wallets better assist users in managing private data. Different applications should adopt different implementations to meet needs while protecting user information as much as possible.
The construction of a unified framework applicable to both humans and Agents is feasible. Agents can judge for themselves how to allocate tasks across different L2s. Currently, their reasoning methods have no essential difference from humans; therefore, the market will gradually find more reasonable allocation methods.
The Evolution Path of Agent Products and Native Standards
Question 5: What is a good Agent product experience?
Vitalik Buterin:
Speaking of a good Agent product, it should be intuitive and easy to use, while operating as part of a larger ecosystem, rather than trying to take over the user’s entire life. At the same time, it must attach high importance to privacy and security, an area where many current systems are still lacking.
Ideally, we need more AIs that are aligned with user interests, not belonging to a single company or application, but acting on behalf of the user. This can reduce attacks and exploitation. Meanwhile, applications need to be compatible with users’ existing configurations, support personalization, and be able to interact with other tools, ensuring security on this basis. If all these can be achieved, that is the ideal state.
Question 6: In the Agent era, what is the development direction for wallets?
Vitalik Buterin:
On one hand, AI can be used to build Ethereum itself, such as enhancing security through formal verification, which will even become a necessary capability in the future. On the other hand, it can also act as an Agent wallet, integrating various capabilities under the premise of ensuring privacy and security.
If this experience relies on third-party servers, true decentralization and privacy cannot be realized. At the same time, restrictions need to be placed on AI behaviors, which is also the wallet’s responsibility in risk control. More importantly, Ethereum should not be viewed as an isolated system; instead, it should be integrated into a global AI, helping users complete various tasks at the operating system level, including on-chain interactions, internet searches, and local data management, thinking from a full-stack perspective.
Question 7: In the Agent economy, how should public goods mechanisms evolve? What will the Native Agent Standard look like in the future?
Vitalik Buterin:
Public goods financing is essentially a governance issue, and governance requires defining stakeholders. Behind every Agent is still the human running it, so the human is always the core. AI and ZK provide new possibilities for governance, but AI also lowers the cost of attacks, making many mechanisms more susceptible to automated attacks. Therefore, this is a direction that requires continuous exploration and continuous iteration.
Regarding the native standard for Agents, there is no fully determined form yet, but an important direction is ZK Payments and ZK API. The core objective is: no matter what type of API request is initiated, each request itself is private, and completely isolated from each other.
This is extremely crucial because in AI scenarios, even if pseudonymous or anonymous identities are used, as long as this identity persists, it will ultimately be re-identified as information continuously accumulates, thereby losing privacy. Therefore, it is necessary to systematically guarantee that there is no correlation between each request.
The key to achieving this lies in utilizing zero-knowledge proofs while avoiding putting every request on-chain; otherwise, the cost and latency would become unacceptable. Although latency can be tolerated in some scenarios, high costs remain an issue that must be solved. On this basis, it can also be combined with bonding/staking mechanisms to prevent abuse on both the user and application sides, without undermining privacy. There is still a lot of work advancing in this direction.
Follow us
Twitter: https://twitter.com/WuBlockchain
Telegram: https://t.me/wublockchainenglish