> ## Documentation Index
> Fetch the complete documentation index at: https://docs.e6data.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Use chat

> Ask questions of your data, attach tables and files, follow streamed progress, and keep the results.

export const DocMeta = ({type, applies, owner}) => {
  const LABELS = {
    tutorial: "Tutorial",
    "how-to": "How-to guide",
    explanation: "Explanation",
    reference: "Reference"
  };
  return <div className="e6-meta">
      <span className={type === "how-to" || type === "tutorial" ? "e6-eyebrow" : "e6-eyebrow e6-eyebrow--quiet"}>{LABELS[type] || type}</span>
      {applies ? <span>Applies to: {applies}</span> : null}
      {owner ? <span>Owner: {owner}</span> : null}
    </div>;
};

<DocMeta type="how-to" applies="All deployments" />

Chat is the **Ask** surface. Start with a question, an investigation, or an outcome you want. A conversation can carry a selected model, attached tables and files, streamed progress, structured answers and durable artifacts.

## Start a chat

1. Open a ready project and select **Chat**.
2. Choose **New chat**. The conversation is saved when you send the first message.
3. If more than one model is available, select one. Your choice applies to this conversation; otherwise it follows the current default.
4. Optionally attach context:
   * Browse and attach relevant tables.
   * Upload files, when attachments are enabled. Supported data files: CSV, TSV, JSON, JSONL, Parquet, Arrow, Feather, XLSX and XLS.
5. Enter your question or instruction and send it.
6. Follow the streamed response and tool progress.
7. Stop the turn if needed. AgentFlow keeps the partial result.
8. Open generated files in the artifact browser.
9. Ask follow-up questions or attach more context.
10. Rename the conversation once it's saved.

## Leave and come back

If you leave a live conversation and return, AgentFlow reattaches to its event stream, so you don't lose progress.

## Answer format

With a [response contract](/agentflow/ask/structured-answers), the final answer is validated and shown as the configured fields: commentary, tables, charts, follow-up questions or your own fields. Without one, answers are free-form Markdown.

## Run an existing workflow from chat

Type `/workflow` to select and run an **existing** workflow. To create a new workflow, describe it in chat instead. See [Build a workflow in chat](/agentflow/automate/build-in-chat).


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