Zhipu AI Open-Sources AutoGLM, an AI Agent for Autonomous Phone Operation
Zhipu AI has officially open-sourced AutoGLM, a groundbreaking AI agent model recognized as the first to demonstrate stable, on-device capabilities for autonomous smartphone operation. According to reports from Blue Whale Finance and TechNode, this advanced model can interpret on-screen content and simulate human-like interactions, such as taps, swipes, and text input, to execute complex multi-step tasks. These tasks include ordering food delivery and booking flights without direct human intervention. Currently, AutoGLM supports core workflows across more than 50 high-frequency Chinese applications, including major platforms like WeChat, Taobao, Douyin, and Meituan. The open-source release is designed for hardware manufacturers, smartphone vendors, and developers, providing them with trained models, a phone-use framework, toolchains, runnable demos, Android adaptation layers, and comprehensive documentation. Zhipu AI emphasizes that the project supports both local and cloud deployment options, allowing users to maintain strict control over their data and privacy. This development marks a significant step forward in the integration of artificial intelligence into daily mobile device usage, offering a fully functional phone-control assistant that can be reproduced on various systems.
Wire timeline
Zhipu AI Open-Sources AutoGLM, an AI Agent for Autonomous Phone Operation
Zhipu AI has officially open-sourced AutoGLM, a groundbreaking AI agent model recognized as the first to demonstrate stable, on-device capabilities for autonomous smartphone operation. According to reports from Blue Whale Finance and TechNode, this advanced model can interpret on-screen content and simulate human-like interactions, such as taps, swipes, and text input, to execute complex multi-step tasks. These tasks include ordering food delivery and booking flights without direct human intervention. Currently, AutoGLM supports core workflows across more than 50 high-frequency Chinese applications, including major platforms like WeChat, Taobao, Douyin, and Meituan. The open-source release is designed for hardware manufacturers, smartphone vendors, and developers, providing them with trained models, a phone-use framework, toolchains, runnable demos, Android adaptation layers, and comprehensive documentation. Zhipu AI emphasizes that the project supports both local and cloud deployment options, allowing users to maintain strict control over their data and privacy. This development marks a significant step forward in the integration of artificial intelligence into daily mobile device usage, offering a fully functional phone-control assistant that can be reproduced on various systems.
TechNode