Chinese artificial intelligence firm Zhipu AI has announced the open-source release of AutoGLM, a pioneering AI agent model capable of autonomously operating smartphones. The model interprets on-screen content to simulate user actions, effectively turning a smartphone into a hands-free device for complex tasks.
How AutoGLM Works
AutoGLM functions by analyzing the visual information on a phone’s screen and executing a series of human-like actions, including taps, swipes, and text input. This allows it to navigate different app interfaces and perform multi-step tasks from start to finish.
The model can handle sophisticated workflows such as ordering food delivery, booking flights, or navigating complex social media apps. It currently supports core functionalities across more than 50 widely used Chinese applications, including WeChat, Taobao, Douyin, and Meituan.
An Open-Source Toolkit for Developers
By open-sourcing the project, Zhipu AI is empowering hardware manufacturers, smartphone vendors, and developers to build and customize their own phone-control assistants. The comprehensive release includes pre-trained models, a dedicated phone-use framework and toolchain, runnable demos, and Android adaptation layers.
A key feature of the release is its support for both local and cloud deployment. This flexibility allows developers to build solutions that maintain user control over data and privacy, a critical consideration for personal device automation.
Relevance for the MENA Tech Ecosystem
The launch of an open-source phone agent like AutoGLM presents significant opportunities for the MENA region’s tech scene. Startups and developers can leverage this foundational technology to build hyper-localized AI assistants tailored for the Arabic language and popular regional apps such as Careem, Talabat, and Noon.
This could accelerate the development of innovative accessibility tools, enhance enterprise automation for mobile-first workflows, and pave the way for a new generation of smart assistants that deeply understand regional user behaviors and application ecosystems without the need to build the core technology from scratch.
About Zhipu AI
Zhipu AI is a China-based artificial intelligence company that specializes in developing large language models (LLMs). Spun out of the Knowledge Engineering Group of Tsinghua University, the company is recognized as a key player in China’s AI landscape, focused on creating new-generation cognitive intelligence models and applications.
Source: Tech in Asia
