> ## Documentation Index
> Fetch the complete documentation index at: https://elephas.app/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# How to Ask Questions About Your Documents

> Ask Elephas a question about your documents and get an answer that cites them. Choose Brains, send your question, follow up, and stop an answer at any time.

Open Ask, choose the Brains you want answers from, type your question and press Return (Enter on Windows). Elephas searches your documents on your computer, writes an answer from what it finds, and cites the exact passages it used. It answers from your documents, not the internet.

## Before you start

* Add a few documents to a Brain and let them finish getting ready. See [Add files and folders](/docs/brains-and-documents/add-files-and-folders).
* Ask needs a way to answer: Elephas AI, your own API key or Local AI, set in **Settings › AI**. See [Choose how Elephas answers](/docs/ai-options/choose-how-elephas-answers).
* The first time you ask, Elephas offers a free trial. See [Quick start](/docs/getting-started/quick-start).

## Steps

<Steps>
  <Step title="Open Ask">
    Click **Ask** in the sidebar, or press ⌘2 on Mac or Ctrl+2 on Windows. To start from a Brain instead, open the Brain and click **Ask This Brain**. Elephas opens a new chat with that Brain already chosen.
  </Step>

  <Step title="Choose what to ask about">
    The box at the bottom shows what the answer may come from as chips, one per Brain. To change it, click **Choose Brains & files** and tick the Brains you want. You can choose up to 8 Brains for one question.

    To narrow the question to particular files, pick them from the same menu. See [Ask about specific documents only](/docs/ask/ask-specific-documents).

    If nothing is chosen, the box says **Nothing chosen yet. Choose a Brain or file to ask about.** and you can't send.
  </Step>

  <Step title="Type your question and send it">
    Type in **Ask anything about your documents…** and press Return (Enter on Windows) to send. Press Shift+Return (Shift+Enter) for a new line.

    On a new chat you can also click one of the suggestions under **Try one of these**, such as **What are the key dates?** or **Summarize these documents**. Clicking one asks it straight away.
  </Step>

  <Step title="Read the answer and check the sources">
    While Elephas works, a short line under its mark says it is busy, for example **Searching** followed by your Brain's name. When the answer appears, each claim carries a small number. Click a number to see the exact passage it came from. See [Check every answer with citations](/docs/ask/citations-and-sources).
  </Step>

  <Step title="Follow up">
    Type your next question in the same chat. Elephas keeps the last few messages of the conversation in mind, so "And what about the renewal date?" works after a question about a contract. Start a new chat with ⌘N on Mac or Ctrl+N on Windows when you change topic.
  </Step>
</Steps>

The steps are the same on Mac and Windows. Only the shortcut keys differ.

## Stop an answer

Any of these stops the answer that is being prepared:

* Click **Stop** in the question box.
* Press Escape while the cursor is in the question box.
* Press ⌘. (Command and period) on Mac or Ctrl+. on Windows, or choose **Chat › Stop Answer**.

Your question stays in the chat. Ask it again with **Try again** when you are ready.

## How it works

* **Answers come only from what you chose.** Ask searches the Brains and files in the chips and nothing else. It never browses the web and never reads Brains you did not choose. See [Why Elephas says an answer isn't in your documents](/docs/ask/answer-not-in-your-documents).
* **Searching happens on your computer.** Elephas finds the matching passages in its local index. Indexing happens on your computer. Always.
* **What goes to the AI depends on your setting.** With Elephas AI or your own API key, your question, a few recent chat messages and the matching passages are sent to get the answer. With Local AI, nothing is sent to an AI provider. See [What data is sent to AI](/docs/privacy-and-security/what-data-is-sent-to-ai).
* **Answers are checked before you see them.** With Elephas AI or your own API key, Elephas reads further into your documents when it needs to, then checks the answer against its sources before showing it. The line under the answer reads **Answer checked** when this is done. With Local AI, the local model answers in a single pass from the passages Elephas found.
* **The chat bar names the AI.** The top of the chat shows which AI is answering, for example **Answered by: Elephas AI** or **Answered by: This computer · Qwen 4B**.
* **One question at a time per workspace.** If another chat in the same workspace is still being answered, Ask says **Another question is still being answered in this workspace.** Wait for it, or stop it.
* **Questions can be up to 8,000 characters long.**

Under each finished answer you will find the sources button (for example **3 sources**), **Try again**, and a **⋯** menu with **Copy**. Your own questions have a **⋯** menu with **Copy** too.

## Troubleshooting

<AccordionGroup>
  <Accordion title="I can't send my question">
    Check three things. The question box must have text in it. At least one Brain or file must be chosen. If you use Local AI, it must be ready: Send stays off while the model is downloading or starting. See [Local AI troubleshooting](/docs/troubleshooting/local-ai-troubleshooting).
  </Accordion>

  <Accordion title="Ask says Configure AI to start asking">
    Elephas does not have a way to answer yet. Click **Open Settings** and choose an option under **How Elephas answers**. See [Choose how Elephas answers](/docs/ai-options/choose-how-elephas-answers).
  </Accordion>

  <Accordion title="The answer says it didn't find anything">
    The passages Elephas found don't support an answer. Check that the right Brains are chosen and that your files finished getting ready. See [Why Elephas says an answer isn't in your documents](/docs/ask/answer-not-in-your-documents).
  </Accordion>
</AccordionGroup>

## Related articles

<CardGroup cols={2}>
  <Card title="Check every answer with citations" href="/docs/ask/citations-and-sources" />

  <Card title="Ask about specific documents only" href="/docs/ask/ask-specific-documents" />

  <Card title="Tips for better answers" href="/docs/ask/tips-for-better-answers" />

  <Card title="Keyboard shortcuts" href="/docs/ask/keyboard-shortcuts" />
</CardGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.