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

# Prompt Assistant

> Use the AI-powered Prompt Assistant to plan generations, enhance prompts, and get model-aware suggestions tailored to each AI model.

## What is the Prompt Assistant?

The Prompt Assistant is an AI chat interface built into every session. It goes beyond simple prompt enhancement -- it can plan multi-step generation workflows, recommend models, and help you iterate on your creative vision through conversation.

## How it works

<Steps>
  <Step title="Open the assistant">
    The Prompt Assistant is accessible from the chat panel in your session. Click the chat icon or use the keyboard shortcut to open it.
  </Step>

  <Step title="Describe what you want">
    Tell the assistant what you're trying to create. You can be as vague or specific as you like:

    * "I need a hero image for a fitness app landing page"
    * "Create a series of product shots for a coffee brand"
    * "Help me generate a cinematic sci-fi scene"
  </Step>

  <Step title="Review the plan">
    The assistant generates a **plan** with recommended models, aspect ratios, and refined prompts. You can review and adjust before executing.
  </Step>

  <Step title="Execute and iterate">
    Apply the plan to generate images directly, or continue the conversation to refine your direction.
  </Step>
</Steps>

## Key features

### Plan tool

The Prompt Assistant's primary capability is the **Plan tool** -- it creates structured generation plans that include:

* Recommended AI models based on your use case
* Optimized prompts with style, composition, and technical details
* Suggested aspect ratios and settings
* Support for both image and video generation

Plans can be applied directly to the generation bar with one click.

### Model-aware suggestions

The assistant understands the strengths and limitations of each available model. When you describe your goal, it recommends the most appropriate models:

* Photorealistic needs → Flux 2 Max, Imagen 4
* Text-heavy designs → Ideogram V3
* Cinematic video → Kling V3 Pro, Veo 3.1
* Fast exploration → Flux Schnell, Nano Banana

### Reference image support

You can attach reference images directly in the chat using drag-and-drop or paste. The assistant will factor these references into its plan, using them to guide model selection and prompt refinement.

<Info>
  Reference images attached in chat are tagged with `@image1`, `@image2`, etc., and can be referenced in follow-up messages.
</Info>

### Streaming responses

The assistant streams responses in real time, so you see suggestions as they're generated rather than waiting for the full response.

## Tips for effective use

<Tip>
  **Be specific about your end use**: Telling the assistant "this is for a social media ad" vs. "this is for a print poster" leads to very different recommendations.
</Tip>

<Tip>
  **Iterate through conversation**: Don't expect the perfect plan on the first try. Respond with feedback like "make it more dramatic" or "try a wider aspect ratio" to refine.
</Tip>

<Tip>
  **Attach references**: Drag and drop reference images into the chat to give the assistant visual context alongside your text description.
</Tip>

## Related guides

<CardGroup cols={2}>
  <Card title="Writing Prompts" icon="pen-fancy" href="/image-generation/prompts">
    Learn manual prompt writing techniques.
  </Card>

  <Card title="AI Models" icon="sliders" href="/image-generation/models">
    Browse available models and their strengths.
  </Card>

  <Card title="Reference Images" icon="image" href="/image-generation/reference-images">
    Upload references to guide AI output.
  </Card>

  <Card title="Best Practices" icon="star" href="/features/best-practices">
    Tips for getting great results.
  </Card>
</CardGroup>