Zhipu AI Launches AutoClaw, a One-Click Local OpenClaw That Turns PCs Into 24/7 AI Agents

Zhipu AI Launches AutoClaw, a One-Click Local OpenClaw That Turns PCs Into 24/7 AI Agents

Hương Trương March 11, 2026 3 phút đọc
Zhipu AI launched AutoClaw, a one-click local OpenClaw installation that turns personal computers into 24/7 AI agents capable of automating complex workflows.

Chinese AI company Zhipu AI officially launched AutoClaw on March 10, introducing what it describes as China’s first “true one-click installation” local version of OpenClaw, aimed at dramatically lowering the barrier to using AI agents.

The tool comes preloaded with more than 50 popular Skills and supports one-click integration with enterprise messaging tools such as Feishu. With a single download, users can effectively turn their computer into a 24/7 AI agent capable of autonomously executing tasks.

While AI agents have become increasingly capable as underlying models improve, their high technical threshold has kept most potential users from adopting them. AutoClaw seeks to address this problem by making OpenClaw installation as simple as downloading a standard application. The platform supports both macOS and Windows, allowing users to install and run the full native OpenClaw system with a single click.

AutoClaw also integrates directly with messaging platforms such as Feishu, allowing users to assign tasks to the AI agent directly from a chat window, seamlessly embedding the agent into everyday workflows.

To encourage adoption, Zhipu AI provides a certain amount of free usage credits, enabling users to try the system at no initial cost. In contrast, many cloud-based AI agent services typically require upfront subscription payments. The platform is also model-agnostic, supporting the use of external models via Coding Plan or API integrations. Recommended compatible models include DeepSeek, Kimi, MiniMax, and GLM.

For more advanced use cases, AutoClaw includes a dedicated model optimized specifically for OpenClaw workflows, internally codenamed Pony-Alpha-2. Designed for real task execution, the model improves tool-calling stability, task progression, and response speed, making it particularly suited for Skill-based automation, scheduled tasks, and continuous agent operations. The current beta version of Pony-Alpha-2 is available to AutoClaw users and select GLM Coding Plan participants, with a full release expected soon.

AutoClaw also integrates Zhipu AI’s proprietary AutoGLM Browser-Use capability, enabling the agent to directly operate a user’s default web browser. By leveraging existing login sessions, the system can complete complex, multi-step workflows across different websites without switching to a dedicated browser environment. This enables significantly more sophisticated automation tasks that go far beyond simple commands.

1280X1280.PNG

At launch, the platform’s 50+ built-in Skills cover a wide range of high-frequency scenarios, including content creation, office productivity, programming, marketing, and financial research. Users no longer need to manually configure APIs for web search or image generation; the system is designed to work almost immediately after installation. Tasks such as generating and publishing social media content, building websites directly from PRDs, processing documents, writing structured code, or conducting financial analysis can all be handled by the agent.

The release reflects a broader vision within the AI industry: lowering the technical barriers to autonomous agents and bringing advanced AI capabilities to everyday users. According to the company, tools like AutoClaw could enable the rise of the “one-person company”, where individuals rely on AI agents to perform a wide range of operational tasks.

Source: IPOzaozhidao

H
Hương TrươngBiên tập viên Doanh nghiệp

Chuyên gia tư vấn chuyển đổi số với 10 năm kinh nghiệm giúp doanh nghiệp Việt Nam ứng dụng công nghệ. Hương viết về Enterprise AI, case study ứng dụng AI trong doanh nghiệp, và phân tích ROI. Nội d...

Bài viết hay? Ấn để tương tác

Bình luận (0)

Hãy là người đầu tiên bình luận trong bài

Đọc
-00:00