> ## 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.

# Ask your first question

> Create a project, give it a semantic layer, and ask your data a question in chat.

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="tutorial" applies="All deployments" />

In this tutorial you create a project, connect it to your data through a semantic layer, and get your first answer in chat. This is the **Ask** part of AgentFlow.

**Before you start:** an administrator has completed [Before you begin](/agentflow/get-started/before-you-begin).

<Steps>
  <Step title="Create a project">
    Open AgentFlow and start the project wizard.

    1. **Name**: enter a project name.
    2. **Context**: add a description and business context, such as your domain terms and rules.
    3. **Review**: confirm and create the project.

    Creating a project doesn't connect any data yet. You do that next.
  </Step>

  <Step title="Create a semantic layer">
    1. Open **Semantic Layers** and choose **Create semantic layer**.
    2. Enter a name. You can't change it later.
    3. Select the active e6data cluster if you're asked to.
    4. Browse the catalog and schema, then pick at least one table and the columns AgentFlow should understand.
    5. Add business context: domain language, measures, known relationships and important rules.
    6. Create the layer. AgentFlow starts its first build automatically.
    7. Wait until the version shows **Built**.
  </Step>

  <Step title="Pin the semantic layer to the project">
    1. Open the project and go to **Settings**.
    2. Select the semantic layer and the exact built version.
    3. Save.

    The project is now ready. A later rebuild of the layer won't change this project until you pin the new version.
  </Step>

  <Step title="Ask a question">
    1. Open the project and select **Chat**, then **New chat**.
    2. Type a question about your data, for example "What were last week's defect rates per line?", and send it.
    3. Watch the streamed answer and the tool progress.
    4. Ask a follow-up, or attach a table for more context.
  </Step>
</Steps>

You have a project with governed data context and a working chat.

## Next

<Columns cols={2}>
  <Card title="Turn the answer into a workflow" icon="workflow" href="/agentflow/get-started/first-workflow">Make it repeatable and run it on a schedule.</Card>
  <Card title="Use chat" icon="message-square" href="/agentflow/ask/chat">Attach files, stop and resume turns, open artifacts.</Card>
</Columns>


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