On most smartphones, artificial intelligence is often limited to a chatbot application, without deep access to other aspects of the operating system. The emergence of DroiClaw (Chinese name: Zhuge) is trying to break this limitation. It has created a native AI operating system that allows artificial intelligence to perform almost all tasks on the device, from information search and content generation to cross-application operations, without relying on third-party apps.
Chinese manufacturer develops “DroiClaw”, a native AI operating system for mobile phones
Traditional chatbots can only answer questions and cannot operate devices;China Zhuoyi TechnologyThe positioning of the developed DroiClaw is completely different. It allows AI to understand speech, images, files, screen content, and perform operations directly at the system level. It provides customizable AI avatars and voices, memory functions and personalized settings, installable pre-made skills and user-created skills, and cross-application automation capabilities. In other words, DroiClaw is an AI system that can “understand you, understand your screen, and operate for you”; on supported devices, DroiClaw can perform various tasks on behalf of the user.
DroiClaw not only understands your needs, but also automatically completes all operations within the app. In addition, it can assist with shopping, summarize group chats, translate conversations, schedule repetitive tasks, and more. Its core capabilities include:
- Cross-application automation: understands your instructions and operates the app on your behalf (user authorization required).
- Multi-modal interaction: supports text, voice, images, files, camera and screen content.
- Personalized extensions: AI avatars, memories, model selection, skill shop, etc.
- Productivity tools: translation, meeting summaries, scheduling reminders, etc.
Another highlight of DroiClaw is its hybrid AI architecture. It uses both:
- Cloud Large Language Model (LLM): Provides faster and smarter responses.
- Device-side model: Starts when offline or when cloud quota is exhausted to ensure that basic functions are not interrupted.
This design allows DroiClaw to maintain basic AI functions when there is no network, and supports multiple cloud model providers and self-built models, allowing manufacturers to quickly build AI infrastructure without having to build an entire agent system by themselves. For manufacturers, DroiClaw provides complete components such as agents, models, skills, permissions and automation, allowing them to focus on hardware and product differentiation instead of spending a lot of resources on AI architecture.
DroiClaw emphasizes “user control”. As AI agents become more capable, the Droi team wants to ensure that all operations are transparent, safe, and overseen. The system has complete user authorization and rights management, clear task scheduling and data protection mechanisms, clearly defined AI operation boundaries and observability. Users can see all steps executed by DroiClaw in the background, so that every action of AI is under the user’s control.
DroiClaw hopes to create a native AI operating system that does not rely on apps, so that AI can be popularized on all devices, instead of only existing on high-end flagship machines. Currently, Coolpad and Philips have launched two smartphones equipped with DroiClaw, priced at about US$160, which provide a complete AI experience while reducing costs.
Source: KOCPC Chinese