CoPaw
Officially listedSelf-hosted personal AI agent platform from Alibaba Cloud's Tongyi team, deployable with three commands.
CoPaw
CoPaw (Collaborative Personal Agent Workspace) is a self-hosted personal AI assistant from Alibaba Cloud's Tongyi team, built on AgentScope and deployable locally with three pip commands. It is not a Q&A tool but a "growing" agent that proactively remembers, reports on schedule, and acts on its own — built for Chinese technical users who value data privacy.
Core Capabilities
- Multi-Channel Integration: Native support for DingTalk, Feishu, QQ, Discord, and iMessage — interact directly in the IM apps you already use
- Long-Term Memory System: Hybrid semantic vector + BM25 retrieval accumulates user preferences across sessions, understanding you better the more you use it
- Scheduled Tasks and Heartbeat: Configure cron expressions for daily task pushes; the agent proactively checks in with reports
- Autonomous Operation: Edit documents, organize files, browse the web, and run scripts — genuinely getting work done on your behalf
- Extensible Skills Architecture: Generate new skills through conversation, with built-in modules for document processing, stock monitoring, browser automation, and more
- Localized Privacy: Data stored entirely locally, never uploaded to any third-party server
Use Cases Professionals who live in DingTalk/Feishu, productivity enthusiasts who need scheduled reports and bulk document processing, and tech hobbyists who want to customize a personal AI assistant. Not suited to ordinary users seeking a zero-configuration, out-of-the-box experience.
Unique Advantages While OpenClaw focuses on Telegram/WhatsApp, CoPaw is the only personal agent with native DingTalk, Feishu, and QQ integration — a decisive advantage for Chinese workplace users. Built on the mature AgentScope framework, its technical foundation is more reliable than standalone open-source projects.
Editor's Review It drew broad attention from the Chinese technical community within two weeks of release. The minimalist installation and privacy-first design earned high user praise, with the Skills dynamic generation capability a particular highlight. The main risk is how new the product is — the open-source GitHub release is not yet complete. Technical users should try it now; enterprise users should wait for the community to mature before adopting.
Pricing
### 定价模式：完全免费（开源） **起步价**：免费（工具本身零费用） #### 主要方案 - **本地部署**：完全免费，仅需承担硬件成本 - **魔搭创空间**：免费一键云端部署 - **Zeabur 云端**：按 Zeabur 平台计费（第三方费用） #### 试用/其他信息 工具本身开源免费，需用户自备 LLM API Key（支持通义、OpenAI、DeepSeek 等），API 调用费用由用户直接支付给 LLM 服务商。
FAQ
Is CoPaw free?
The CoPaw tool itself is completely free and open source, with no software subscription fees. You need your own LLM API key and pay only the API call costs charged by providers such as Tongyi, OpenAI, or DeepSeek.
What can CoPaw be used for?
CoPaw is a self-hosted personal AI agent that supports long-term memory accumulation, scheduled task notifications, document processing, web browsing, file management, stock monitoring, and more, and you can interact with it directly in IM apps such as DingTalk, Feishu, and QQ.
How do you install and deploy CoPaw?
Local deployment takes just three commands: pip install copaw, copaw init --defaults, and copaw app — about 30 minutes total. You can also deploy to the cloud in one click via ModelScope Studio, with no local environment setup required.
Which communication platforms does CoPaw support?
DingTalk, Feishu, QQ, Discord, iMessage, and a native console. In particular, the native integration with DingTalk, Feishu, and QQ is a unique advantage among comparable open-source products, designed specifically for Chinese workplace users.
How does CoPaw differ from OpenClaw?
CoPaw focuses on Chinese IM platforms (DingTalk/Feishu/QQ) and is maintained by Alibaba Cloud's Tongyi team, with a more mature technical foundation; OpenClaw covers Telegram/WhatsApp/Signal and has a larger international community (100,000+ stars). The two are complementary in positioning.
Does CoPaw upload data to the cloud?
In local deployment mode, all data (memory files, configuration files) is stored entirely locally and never uploaded to any third-party server, giving you true ownership of your data privacy. In cloud deployment mode, data is stored on the user's own cloud server.
Is CoPaw suitable for non-technical users?
There is a real barrier. Users with Python basics can complete deployment within 30 minutes; users with no technical background can deploy in one click via ModelScope Studio, but configuring the LLM API key still requires basic understanding. Not suited to ordinary consumers seeking a zero-configuration experience.