For two consecutive days, two groups of people arrived at the home of Sam Altman, the father of GPT. One group threw Molotov cocktails at his house, while the other group opened fire directly into the house.

No one was injured. However, the timing of this incident is noteworthy—just a few days after OpenAI released its latest model, just after the world became aware of Anthropic's strongest model, Mythos, just after Hermes 3 recently surpassed OpenClaw in multiple benchmark tests, and during the week when the entire AI industry began to seriously discuss whether "the speed of model iteration has already surpassed human understanding."

This is not a simple piece of social news. It is a signal: when the speed of technological change begins to feel out of control, that feeling of loss of control will find specific outlets.

From any dimension, the AI industry has been in an accelerating state from the beginning of the year until now. The release cycle for new models has been compressed from quarterly to weekly, and sometimes even daily. A product that was regarded a month ago as the strongest reasoning model may have already been surpassed by three competitors today. Hermes replaced OpenClaw—not in the distant future, but last week.

This speed itself has already become a structural variable.

On April 21, Rhythm BlockBeats joined Zhihu to hold an event in Hong Kong titled "Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions." The purpose is to capture, at the first moment, what kind of financial scenario we will face soon as AI Agents combine with on-chain operations.

The theme of this event is not "What is Web 4.0," but "When Agents become the main participants on-chain, in what dimensions do the game rules shift." This event is not meant to define anything, but to find some solutions.

This event is co-hosted by Cobo and MuleRun, with thanks to Noxcat, Lutos, SmallWOD, and others as event supporters, who are the first to run in this intersectional space.

Shu Junliang, CTO of MuleRun

When the speed of model iteration has compressed to a weekly basis, the bottleneck is no longer whether AI is smart enough, but who can use it quickly and safely. MuleRun, as the world's first self-evolving personal AI, has replaced the technical barrier with natural language. Users can open their browser to describe their trading logic, and the Agent can automatically monitor the market, build workflows, and generate institutional-level investment research signals 24/7. Rather than making decisions for users, it empowers professionals to drive AI with their own judgment.

Shu Junliang has been in cybersecurity for over a decade, and from day one, MuleRun has written security into its core. The question he wants to discuss at the event is straightforward: how to liberate human judgment from execution friction.

Brad Bao, Head of AI Growth at Cobo

When Agents start to actively manage on-chain assets, how can we ensure that their autonomous execution abilities never overstep boundaries? The roles of digital asset custody and wallets also need to be redefined in the age of AI. This is not an upgrade of product functionality, but the construction of an infrastructure.

Since its establishment in 2017, Cobo has provided digital asset custody and wallet infrastructure for over 500 institutions worldwide, managing assets worth trillions of dollars, with zero security incidents to date. In the face of the rise of AI Agents, Cobo is exploring the construction of a trusted infrastructure for the agentic economy, integrating comprehensive wallet solutions with advanced risk control systems and developer tools, empowering various institutions and AI Agent users to efficiently build, operate, and achieve scalable expansion in an ever-changing digital financial world. Brad Bao is mainly responsible for implementing Cobo's strategy in the AI field.

NOXCAT

NOXCAT is building a Web3 infrastructure platform aimed at next-generation digital financial scenarios—integrating on-chain guarantee mechanisms, social relationship networks, and multi-asset trading systems, allowing users to complete value exchanges and asset allocation without needing to trust intermediaries.

The core judgment in this direction is that the next threshold in digital finance is not technology, but trust costs. What NOXCAT is doing is compressing trust costs into the protocol layer.

Xu Jialong, Associate Vice President of Hong Kong University of Science and Technology

The Hong Kong University of Science and Technology has long been at the forefront of research in digital finance. Professor Xu Jialong brings a structural analytical perspective beyond the narrative intersections of technology and capital.

Xingpt, All-Asset Trader

He once had a post titled "AI Optimizing Workflows" on X platform that garnered over 2 million views.

Chriz, Founder of Donut AI

Formerly employed at ByteDance, he founded Donut AI to build trading infrastructure for the AI era, with the first product—an active agent D0 for all-asset trading—already launched.

Christian, Founder of Infini

As the founder of Infini, Christian has focused on stablecoin payments, cross-border capital flows, and global financial infrastructure. He is committed to promoting the integration of fiat currency systems, crypto networks, and AI capabilities, helping businesses efficiently complete collection, payment, fund management, and financial collaboration, creating an AI Financial OS for global commerce to serve the next generation of global enterprises such as cross-border e-commerce, SaaS, and digital entertainment.

He believes that "true global finance does not belong to the old system but to the next-generation infrastructure driven by stablecoins and AI."

Box, Developer Relations at Monad Foundation for Greater China

As currently the high-performance public chain in the crypto industry, Monad pays special attention to AI Agent + on-chain applications. The "Monad AI Blueprint" program and AI Hackathon tailored for Agents have attracted quite a number of quality teams and projects to participate.

We believe that the sharing by these guests on the new financial models brought by AI Agents will provide everyone with different perspectives. The event on April 21 in Hong Kong, hosted by Rhythm X Zhihu, is still open for registration. For the registration link, please click: https://luma.com/r3h4eqk

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。