User Guide

Reference documentation for every feature

Guide navigation

Knowledge Chat

Knowledge Chat at /chat lets you ask natural-language questions across your entire knowledge base. The system uses AI with retrieval-augmented generation (RAG) to search relevant notes and compose answers with citations.

Starting a Conversation

Type a question in the message input at the bottom and press Ctrl + Enter (or click the send button). Your message appears on the right, and the AI response appears on the left once processing is complete.

Citations and Sources

Each answer includes source badges showing which notes were used. Badges display the note title and a relevance score. Click a badge to jump directly to the source note in the editor.

Inline citations appear as wikilinks in the response text, linking back to the notes that informed the answer.

Follow-Up Questions

Below each response, the AI suggests follow-up questions as clickable buttons. Click one to automatically insert and send it, keeping the conversation flowing without manual typing.

Chat Sessions

Conversations are organised into sessions listed in a sidebar. Each session auto-titles itself from the first question. Click a session name to reload its full history.

Toggle the session sidebar with the hamburger icon in the top left.

Save as Summary

When a conversation has messages, a Save as Summary button appears. Clicking it creates a background job that synthesises the conversation into a new note. The generated summary appears in your note list once processing completes.

Knowledge Chat vs. Note Chat

OikoNotes has two related but distinct AI chat experiences:

  • Knowledge Chat at /chat searches across your whole knowledge base and cites multiple notes
  • Note Chat inside the editor stays focused on the single note you currently have open

Use Knowledge Chat for broad questions and synthesis. Use Note Chat when you want help refining or interrogating one note in depth.

How It Works

Knowledge Chat uses retrieval-augmented generation:

  1. Your question is converted into a vector embedding
  2. The most semantically relevant note chunks are retrieved
  3. These excerpts are sent to the AI along with your question
  4. The AI composes an answer grounded in your actual notes
Tip: Knowledge Chat works best when you have 50 or more well-tagged, well-developed notes. The more content in your knowledge base, the richer the answers.