Capture what your enterprise knows,
make it answer your questions
A modern enterprise knowledge base for SMEs — structured spaces, hybrid semantic + keyword search, and an Ask-the-Workspace AI palette that answers in natural language with citations to the source page. Self-host or bring your own LLM.
What X-Note does
Pragmatic features built for real-world teams.
Rich-text + Markdown
Block-based editor with code, callouts, tables and image uploads.
Spaces & Permissions
Group docs into spaces with read/write controls per team.
Fast Search
Full-text search across every page, with locale-aware ranking.
Embeddable Widget
Drop a single script tag in any app to surface your KB inside it.
Page History & Diff
Every save snapshotted; revert or diff against any prior version.
AI that knows your docs
Three real surfaces shipping today — not a roadmap.
Ask your workspace · Live
Cmd+J anywhere → ask in natural language → get a streamed answer with citations to the source pages. Hybrid retrieval blends keyword matching with semantic embeddings, so questions don't need to use the exact words on the page.
Per-page assistant · Live
Toolbar dropdown on every article: Summarise · Improve · Outline · Translate (EN ↔ 中文). Output streams to a side panel — copy what you like; your original is never touched.
Bring your own LLM · Live
Two env vars point the platform at any OpenAI-compatible endpoint — OpenAI, Anthropic, DeepSeek, or a local Ollama / llama-cpp. Embeddings server is separately configurable. Run the whole stack air-gapped.
Your knowledge stays on your servers
Set LLM_URL and EMBEDDINGS_URL to localhost (Ollama default), or any cloud OpenAI-compatible API. Embeddings stored as JSON in Postgres — no pgvector required for sites under ~100k pages. Search still works if the LLM is offline (graceful keyword-only fallback).