PandaWiki

PandaWiki fills a specific niche: an open-source, AI-native knowledge base that you control fully. If your team values data sovereignty and has DevOps chops, it's a solid bet. The built-in embedding and reranker models mean you don't need to cobble together a RAG stack — just bring your chat model API key and go. For a self-hosted wiki, the integration with Chinese enterprise messengers (DingTalk, Feishu, WeCom) is a rare plus. But it's not for everyone. You need Linux, Docker, and comfort with AI model configuration. There's no real-time collaboration — concurrent editing isn't supported, so teams used to Google Docs-style co-authoring will hit friction. The free edition lacks SSO and advanced permissions; those require the paid version. Compared to Outline, PandaWiki is heavier on AI features (Outline's search is keyword-based, not semantic) but lighter on integrations (Outline connects to Slack, Jira, etc.). If you're outside China, the bot integrations for DingTalk/Feishu may be irrelevant. Docusaurus is simpler if you just need static docs without AI. Where PandaWiki shines is as an internal knowledge base for technical teams that want AI Q&A on controlled data — think support docs, product manuals, or internal wikis. The web widget makes it easy to embed a chatbot on a customer-facing site. For the price (free self-hosted), it's hard to beat if you can handle the setup. One caveat: the project is backed by Changting Technology (a Chinese security company), so consider data residency regulations. The paid pricing isn't public; you must contact sales. Support for DeepSeek and Baizhi Cloud suggests a China-first approach, but OpenAI compatibility keeps it global.