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

# Build your first AI image generation pipeline

> Step-by-step walkthrough for creating an image generation workflow on the Knouds canvas — from adding nodes to running your first test.

The Knouds canvas lets you wire AI models together visually. In this guide you will build a simple image generation pipeline: a text prompt goes in, a generated image comes out. By the end you will have a saved, runnable workflow and a clear mental model for building more complex pipelines.

<Tip>
  Choose your workflow name carefully before you create it. The name is converted into a URL slug at creation time — for example, "Product Banner Generator" becomes `product-banner-generator` — and **the slug cannot be changed later**. A descriptive name makes your API calls readable and your workflow easy to find.
</Tip>

<Steps>
  <Step title="Open the canvas">
    Go to [knouds.ai](https://knouds.ai) and click **New Workflow** in the top bar or on the homepage.

    In the dialog that appears, choose **From Scratch**, then type a descriptive name for your workflow (for example, "Product Image Generator"). Click **Create** to open the blank canvas.
  </Step>

  <Step title="Add a Request Input node">
    The left sidebar lists all available node categories. Open the **Inputs** category and drag a **Request Input** node onto the canvas.

    In the node's settings panel, click **Add field** and configure it:

    * **Name:** `prompt`
    * **Type:** Text

    This field is what callers will provide when they run the workflow — either through the canvas form or via the API.
  </Step>

  <Step title="Add an Image Generator node and connect it">
    Open the **Image Models** category in the sidebar and drag an **Image Generator** node onto the canvas next to your Request Input node.

    Connect the nodes:

    1. Hover over the **text** output handle (orange dot) on the right edge of the Request Input node.
    2. Click and drag to the **prompt** input handle on the left edge of the Image Generator node.
    3. Release to create the wire.

    With the Image Generator node selected, configure it in the panel on the right:

    * **Model:** choose your preferred image model from the dropdown.
    * **Aspect ratio:** set to your desired output dimensions (for example, `16:9`).

    <Note>
      Only nodes that are connected to the flow will execute. If you place a node on the canvas but leave it disconnected, it is skipped automatically when the workflow runs.
    </Note>
  </Step>

  <Step title="Add a Response Output node and connect it">
    Open the **Outputs** category in the sidebar and drag a **Response Output** node onto the canvas.

    Connect the **image** output handle (pink dot) on the right edge of the Image Generator node to the input handle of the Response Output node.

    The Response Output node is what the API returns and what the canvas displays after a run.
  </Step>

  <Step title="Run the workflow on the canvas">
    Click the **Run** button on the Image Generator node (the play icon in the node's top-right corner). A prompt input form appears.

    Type a test prompt — for example: `a golden retriever sitting in a sunlit park` — and click **Run**.

    Knouds executes the connected nodes in order. You will see a progress indicator on each node as it runs.
  </Step>

  <Step title="View the result inline">
    Once execution completes, the generated image appears directly inside the Image Generator node and in the Response Output node. Click the image to expand it.

    If the result is not what you expected, adjust the prompt or model settings and click **Run** again — each run is independent.
  </Step>

  <Step title="Save the workflow">
    Click **Save** in the top bar (or press `Cmd+S` / `Ctrl+S`). Knouds writes the current node graph, settings, and last result to your account.

    Your workflow is now saved and callable via the REST API. See [Deploy a workflow as a REST API](/guides/deploy-as-api) for the next step.
  </Step>
</Steps>