Over the past six months, AI agent products like Claude Code, Claude Cowork, and OpenAI’s Codex have defined a new industry trajectory, prompting Chinese tech giants and AI startups to launch competing coding and work agents. Tencent leads with WorkBuddy and CodeBuddy; WorkBuddy leverages Tencent’s ecosystem of WeChat, Tencent Docs, Tencent Meeting, and ima knowledge base, while CodeBuddy now covers over 95% of Tencent’s own engineers. Alibaba’s Qoder family includes Qoder Desktop for developers and QoderWork for general productivity tasks. ByteDance’s TRAE Work offers web, desktop, and mobile access. Among AI-native companies, Moonshot AI’s Kimi Code and Kimi Work use long-context capabilities to handle extended task chains. MiniMax Code uses a Producer-Verifier adversarial loop with the MiniMax M3 model. Zhipu AI’s ZCode 3.0, powered by the GLM-5.2 model, focuses on developer depth. Chinese AI agents have a key advantage in adapting to local workflows, integrating with WeChat, DingTalk, Feishu, and domestic models, while offering greater flexibility in model choice, pricing, and enterprise customization. TRAE Work has open-sourced its Trae Agent framework.