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

# Chat

> Use natural language to search, analyze, and create research outputs.

The chat interface is the primary way you interact with Purna. Ask questions in natural language, attach files, run tools, and receive evidence-backed answers with citations.

## Starting a conversation

Click the **+** button in the top bar to start a new chat. The home screen displays suggested prompts to help you begin.

Click a suggestion or type your own request directly into the chat input.

![Chat home screen](https://storage.googleapis.com/health_reports/web_images/docs/chat-home.png)

<Tip>
  Start with a focused question. Include the gene, pathway, organism, dataset type, file, disease area, or target output when possible.
</Tip>

## Choosing a research mode

Use the **Research Mode** dropdown in the chat input before sending your first message.

| Mode                    | Use it for                                                                                                 |
| ----------------------- | ---------------------------------------------------------------------------------------------------------- |
| **Biomedical Research** | Literature, pathways, disease mechanisms, genes, proteins, variants, drugs, and broad biological reasoning |
| **Omics Analysis**      | RNA-seq, single-cell, proteomics, metabolomics, multi-omics, and dataset analysis                          |

Research mode is locked after the first message in a chat. Start a new chat to change modes.

## Using Preferences

Click **Preferences** in the chat input to adjust the current chat.

<AccordionGroup>
  <Accordion title="Workflow modes" icon="sliders">
    Turn on focused workflows such as **Deep Research**, **Consilium**, or **Paper Writing** when you need more structured research, debate, or drafting behavior.
  </Accordion>

  <Accordion title="Capabilities" icon="wand-magic-sparkles">
    Enable or disable tool families such as web search, code execution, image generation, and background jobs. Locked capabilities remain controlled by your plan.
  </Accordion>

  <Accordion title="Skills and knowledge" icon="book-open">
    Choose which knowledge resources and source-aware guidance should be active for the conversation.
  </Accordion>
</AccordionGroup>

## Types of questions you can ask

<AccordionGroup>
  <Accordion title="Gene, variant, and disease evidence" icon="dna">
    Ask about genes, variants, disease associations, pathway effects, clinical evidence, allele frequencies, and literature. Example: "What evidence links PTEN loss to immune checkpoint response?"
  </Accordion>

  <Accordion title="Literature synthesis" icon="book-medical">
    Ask Purna to summarize a field, find contradictions, extract evidence tables, or compare papers. Example: "Summarize recent work on senolytics in idiopathic pulmonary fibrosis."
  </Accordion>

  <Accordion title="Dataset analysis" icon="chart-bar">
    Attach or reference a file and ask Purna to inspect, summarize, transform, or visualize it. Example: "Use @counts.csv to identify the strongest treatment-response genes."
  </Accordion>

  <Accordion title="Reports, drafts, and structured outputs" icon="file-lines">
    Ask for literature briefs, research plans, tables, code files, manuscript sections, or downloadable spreadsheets. Example: "Create an evidence table comparing EGFR, MET, and HER2 resistance mechanisms."
  </Accordion>
</AccordionGroup>

## Citations and sources

When Purna uses external evidence, responses include inline citation badges and source controls. Click a citation to open the original source record in a new tab.

![Chat conversation with citations](https://storage.googleapis.com/health_reports/web_images/docs/chat-conversation.png)

<Note>
  Verify important scientific or clinical claims against the cited source records before using them in decisions, publications, or regulated work.
</Note>

## Attaching files

Click the paperclip icon next to the send button to attach files to your message. You can upload datasets, documents, images, and other research files for Purna to inspect.

Type `@` in the chat input to reference a file that is already in your Files library.

## Working with artifacts

During conversations, Purna may create artifacts that appear in a side panel next to chat.

Common artifact types include:

* **Documents** for summaries, briefs, protocols, drafts, and reports
* **Code files** for reusable Python or R scripts
* **Spreadsheets** for tabular results and exports
* **Images and figures** for generated visual outputs

Use follow-up prompts to revise artifacts, change formatting, add sections, or export results.

## Managing chat history

The sidebar displays past conversations organized by **Today**, **Yesterday**, and **Last 7 days**. Use the search bar to find conversations by keyword.

Click any conversation to resume it. Hover over a conversation to reveal a menu with actions such as **Rename**, **Share**, and **Delete**.

## Chat visibility

The lock icon with a dropdown arrow in the top bar controls conversation visibility. Click it to toggle between private and shared visibility for your current conversation.

<Tip>
  Keep conversations private by default when they include confidential data, unpublished results, or sensitive research plans.
</Tip>

## Providing feedback

Each response includes feedback options:

* **Copy**: Copy the response text to your clipboard
* **Thumbs up**: Mark the response as helpful
* **Thumbs down**: Mark the response as unhelpful