What Is SHU

On April 21, during the roundtable discussion 'Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions', MuleRun CTO Shu Junliang discussed the topic of 'Agent Moat'. He stated that the traditional AI technology moat is being rapidly weakened, primarily due to the accelerated convergence of model capabilities and exponential improvements in development efficiency. He pointed out that the performance gap between mainstream large models is narrowing quickly, especially over the past year, where the capability gap between domestic and international models has significantly converged. Additionally, with the explosive enhancement of code generation capabilities, software development efficiency has greatly increased—features that previously took weeks or even months to complete can now be achieved in just a few days. This means that both Agent frameworks and specific functional modules can be quickly reused or replicated through open-source solutions, making the 'functional moat' at the product level increasingly fragile. In this context, Shu Junliang believes that the core competitiveness of Agents in the future will mainly be reflected in two aspects: first, the ability for continuous high-frequency iteration, meaning whether the team can maintain a leading product update speed over the long term; and second, data advantages, including exclusive data resources and user accumulated data. On one hand, platforms with unique data acquisition capabilities (such as data from specific industries or regions) will form natural barriers; on the other hand, the behavioral and memory data accumulated from long-term user engagement on the platform will become key assets that are difficult to migrate, further enhancing user stickiness and product competitiveness. He concluded that as the trend of 'equalization' in models and technologies continues, the Agent moat is shifting from 'technical capability' to a comprehensive competition of 'data assets and execution efficiency'.

On April 21, Shu Junliang, the Chief Technology Officer of MuleRun, the world's first self-evolving personal AI project, shared at the offline event themed "Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions" that, from a product definition perspective, AI Agents should essentially be viewed as "personal assistants," with the core goal of continuously reducing usage costs and barriers through technological means. Based on this positioning, the capabilities of the Agent can be abstracted into a multi-dimensional structure, including modules for "mouth (interaction capability)," "eyes and ears (perception capability)," "brain (reasoning and decision-making)," and "memory and knowledge (long-term learning)," with different capabilities corresponding to different underlying technological systems. In terms of interaction, he pointed out that AI Agents are gradually expanding from traditional text dialogues within web pages or apps to multi-channel communication, including mainstream platforms such as Telegram, Discord, Feishu, DingTalk, and WeChat, achieving a "no-interface" natural interaction experience, thereby significantly lowering the user entry barrier. Regarding core on-chain scenarios, MuleRun proposed an infrastructure solution centered on "funds permission security," which includes sandbox isolation, cloud execution, and a full-chain traceability mechanism, creating a trustless operating environment to address potential security issues during the Agent's automatic execution process. In terms of capability evolution, the Agent will feature a self-evolving decision-making model that can continuously learn users' trading strategies and risk preferences, forming a personalized investment research and execution system. Additionally, through a knowledge network mechanism, it will enable strategy accumulation and sharing, promoting the reuse and diffusion of on-chain cognition and capabilities. Shu Junliang further noted that as the capabilities of AI Agents improve, the division of labor in on-chain transactions will be restructured: Agents will gradually take over information processing and execution, while humans will focus on higher-level strategy formulation and key decision-making.

On April 20, Shu Junliang, Chief Technology Officer of MuleRun, the world's first self-evolving personal AI project, confirmed his attendance at the offline event "Decoding Web 4.0: When AI Agents Take Over On-chain Permissions," co-hosted by Rhythm BlockBeats, Dongcha Beating, and Zhihu, which will be held in Hong Kong on April 21. In his keynote speech, Shu will focus on the following areas: · Security Infrastructure: Creating a 'trustless' environment around fund permissions with sandbox isolation, cloud execution, and full-chain traceability. · Reconstructing Entry Barriers: Moving from developer tools to mass adoption, achieving an 'instant use' interaction experience with Agents. · Self-evolving Decision Models: Agents learning user strategies and risk preferences to form personalized investment research and execution systems. · Knowledge Network Mechanisms: Strategy accumulation and sharing, driving on-chain cognition and capability reuse. · Restructuring Trading Division of Labor: Agents taking over information and execution, allowing humans to focus on core decision-making. Additionally, his speech will explore the application paths of integrating digital finance and traditional finance under the empowerment of AI, in conjunction with MuleRun's latest access to trading infrastructure and multi-source data capabilities.