Skip to main content

Query

The Query tab is where your Knowledge Base pays off. Ask any question and Lira will search across everything you've indexed — documents, connected sources, and crawled web pages — and generate a grounded, source-cited answer.

Path: Sidebar → Workspace → Knowledge Base → Query

How it works​

Query uses Retrieval-Augmented Generation (RAG):

  1. You type a question
  2. Lira runs a semantic search across all indexed content — finding the chunks most relevant to your question, even if they don't use the exact same words
  3. The retrieved chunks are assembled into a context block and sent to the AI model alongside your question
  4. The AI generates an answer using only the content you've provided — it does not guess or use general knowledge beyond your data
  5. Source citations are shown below each answer so you know exactly where the information came from

Asking questions​

Type your question in the input box at the bottom of the Query tab and press Enter (or click the send button).

The conversation is multi-turn — Lira remembers the previous messages in the session, so you can ask follow-up questions naturally:

You: What is our refund policy?
Lira: [finds and explains the refund policy from your uploaded policy doc]
You: Does that apply to enterprise customers too?
Lira: [follows up using the same context]

Source citations​

Every AI response shows source badges below the answer text. Each badge shows:

  • Whether the source is a Document (uploaded file or connected source import) or a Web Page (crawled URL)
  • The name of the document or page title

Click through to the original source in the Documents tab or on the web.

Suggested questions​

When no messages have been sent yet, the Query tab shows a few suggested question prompts. Click any suggestion to pre-fill it in the input — then edit or send as-is.

"Not enough information" responses​

If Lira cannot find relevant content in your Knowledge Base for a question, it will say so rather than make something up:

"I don't have enough information in the knowledge base to answer this question accurately."

This means one of two things:

  1. The content hasn't been added yet — go to Documents, Connected Sources, or Web Sources and add the relevant material
  2. The question is too specific — try rephrasing or asking a broader version first

Query tab is empty / disabled​

The Query tab requires at least one indexed source to work. If you haven't uploaded any documents, connected any sources, or crawled any web pages, you will see a message directing you to add content first.

Good questions to ask​

About your organisation:

  • "What is our return and refund policy?"
  • "What are the key features of [Product Name]?"
  • "Summarise our Q3 OKRs"

About processes:

  • "What is the onboarding process for new engineers?"
  • "What are the steps to submit an expense claim?"
  • "How do we handle customer escalations?"

Research and reference:

  • "Find everything we have about competitor X"
  • "What did we learn from last quarter's customer interviews?"
  • "What does our compliance policy say about data retention?"

Limitations​

  • Session-only memory — the conversation history is only kept for the current browser session. Refreshing the page starts a fresh conversation.
  • No cross-session continuity — past query sessions are not saved.
  • Only your data — Lira will not use general AI knowledge to fill in gaps. If the information is not in your Knowledge Base, it will say so.
  • Language and format — Lira performs best on English-language content that is well-structured and clearly written.