To The Moon

Western Securities has released a research report stating that as foundation model capabilities continue to improve, competition among Agents is expected to gradually shift from single-turn question-and-answer performance toward task execution stability, cross-application collaboration efficiency, and long-term service capabilities.

The firm believes that platforms with advantages in user entry points, application ecosystems, and accumulated work context are likely to build stronger user stickiness in Agent competition. In China, investors should watch Tencent Holdings (00700) and Alibaba-W (09988), as the two companies respectively possess ecosystem strengths in social networking, office collaboration, e-commerce, and cloud computing, and are expected to drive Agents into high-frequency use cases.

Agents Are Upgrading from Information Q&A to Continuous Task Execution, and AI Applications May Usher in a New Round of Development Opportunities

A new generation of Agents represented by Meta Muse can complete tasks such as shopping, travel planning, schedule management, and small business operations under user authorization through independent cloud-based operating environments, long-term memory, and tool connectivity, while continuously following up as external information changes. Compared with traditional AI assistants, the core change is the shift from providing one-off answers to undertaking affairs over the long term, reducing the cost of repeated communication and manual operations for users.

From the Perspective of Product Form, Major Vendors Are Exploring Different Agent Development Paths

Muse organizes life and work affairs around users' long-term goals, Grok Bot achieves team division of labor through multiple digital roles with independent responsibilities, and ChatGPT dot coordinates complex cross-tool and cross-scenario tasks with a unified assistant. Although the three differ in task organization and interaction methods, all emphasize long-term memory, tool invocation, and continuous background execution capabilities. The firm believes that as foundation model capabilities continue to improve, competition among Agents is expected to gradually shift from single-turn question-and-answer performance toward task execution stability, cross-application collaboration efficiency, and long-term service capabilities.

As Agent Execution Capabilities Continue to Expand, the Product Boundaries Between Personal and Office Scenarios Are Gradually Blurring

Codex, Claude Code, and WorkBuddy have expanded from software development to research analysis, data processing, document creation, and business process automation, forming overlaps with the capabilities of products such as Muse, Grok Bot, and dot. In the future, the same task may be completed by different types of Agents, and when choosing products, users will place greater emphasis on whether existing materials can be smoothly connected, whether tasks can be executed reliably, and whether results can be conveniently used going forward. The firm believes that platforms with advantages in user entry points, application ecosystems, and accumulated work context are likely to build stronger user stickiness in Agent competition.

Risk Warnings: Model iteration progress falling short of expectations; Agent product release pace falling short of expectations; privacy and authorization risks, third-party platform restrictions, and function availability and commercialization progress falling short of expectations, among others.