
Zorv AI — Open source AI Agent assistant for Android. Multi-model dialogue, personality system, memory, full-duplex TTS/STT, scheduled tasks, extensible toolchain, integration with Feishu/QQ/WeChat. Developed with Kotlin + Jetpack Compose.
Zorv AI is an on-device AI Agent for Android. It runs on the phone and can operate the device through accessibility, Shizuku, and ROOT channels. It includes 235 built-in tools and runs MNN / llama.cpp offline large models, with terminal and Linux sandbox, MCP, knowledge base, and speech synthesis/recognition. It also connects via Feishu, QQ, and WeChat. The project addresses three things: letting AI act, letting AI work offline, and letting AI work across apps. Higher-privilege capabilities are tiered L1–L5 and require explicit authorization. Offline operation is supported by local inference engines, local STT, local TTS, local RAG, and an in-app Ubuntu 24.04 Linux sandbox. Cross-app operation uses ACI, a same-device, AIDL Binder-based, no-root local protocol that lets any app expose itself as an AI-callable capability. The design is Tool-first: every capability is expressed as a QuroTool registered in QuroToolRegistry. It is also a self-extensible Agent runtime with an APK-level plugin framework, where independent APKs can register extension points to add new tools, ACI capabilities, interfaces, or commands without changing the host code.
Teams use Zorv AI to operate Android devices through accessibility, Shizuku, or ROOT channels.
Teams use Zorv AI to run offline LLM inference with MNN / llama.cpp.
Teams use Zorv AI to run a Linux sandbox with proot and Ubuntu 24.04.
Teams use Zorv AI to connect via Feishu, QQ, and WeChat.
Teams use Zorv AI to build vector semantic RAG knowledge bases and memory.
Teams use Zorv AI to extend capabilities with APK-level plugins.
Teams use Zorv AI to generate native Compose interfaces with GenUI.
Teams use Zorv AI to use MCP client and local MCP service.
Lynote combines AI-draft rewriting and AI-text detection with note-taking, transcription, video summaries, and study tools.
LMSYS Chatbot Arena is a crowdsourced open platform for LLM evals. Collected over 1,000,000 human pairwise comparisons to rank LLMs with the Bradley-Terry model and display the model ratings in Elo-scale.
Zhipu Qingyan is a GLM-based AI assistant whose official description emphasizes understanding goals, breaking down tasks, and using tools.
字节跳动AI助手