Asking Questions

The Query tab is where you chat with your documents. You type a question in plain English, and the AI searches through your uploaded files to find the answer.

Choosing What to Search

Before you start asking questions, you can narrow down which documents the AI looks through. At the top of the Query tab, you'll see two dropdowns:

• Select Individual File — Pick a single document to search within.
• Select File Set — Pick a group of documents you've organized together (like "Q4 Reports" or "Research Papers").

You only need to use one of these. Selecting a file clears the file set selection, and vice versa.

If you don't select anything, or if you click General Query, the AI will search across all of your uploaded documents.

Asking a Question

Type your question in the text box at the bottom of the chat area and press Enter (or click the send button). The AI will:

1. Search your selected documents for the most relevant passages
2. Read those passages
3. Write a response based on what it found

You'll see the response appear in real time as the AI generates it — word by word — so you don't have to wait for the full answer before you start reading.

What kinds of questions work well?

Archivist works best with questions that have answers in your documents:

• Factual lookups — "What is the contract termination clause?"
• Summaries — "Summarize the main findings of this report."
• Comparisons — "How do the two proposals differ on pricing?"
• Searches — "Which documents mention data retention policies?"

It's less suited for opinions, creative writing, or questions that require knowledge beyond your uploaded files.

Citations

When the AI bases its answer on specific passages from your documents, it can show you exactly where that information came from. These are called citations.

Citations appear below the AI's response and include the document name and the passage that was used. You can use this to verify the answer or read the full context yourself.

Citations can be turned on or off in AI Settings.

Voice Input

If you prefer to speak your questions rather than type them, Archivist includes a built-in voice input feature powered by a local speech-to-text model.

To use voice input:

1. Click the microphone button in the voice input area.
2. Speak your question. You'll see a live audio waveform while recording.
3. Click the button again to stop recording.
4. Archivist transcribes your speech into text.

You can choose what happens after transcription:

• Auto-send on — Your question is sent to the AI immediately after transcription.
• Auto-send off — The transcribed text is placed in the input box so you can review or edit it before sending.

Exporting Your Conversation

If you want to save a chat session for your records, click the Export Session button at the top of the Query tab. This saves the entire conversation — your questions and the AI's responses — as a Markdown text file.

A save dialog will appear so you can choose where to save it.

Starting Fresh

Your chat history is preserved while you work, but if you want to start a new conversation, simply navigate away and back, or begin asking questions on a different topic. The AI treats each question independently based on your document search — it doesn't rely on previous messages for context.