The action Generate an image with AI

Purpose  

Generates an image (PNG) from text using an OpenAI image model. Use this action to create illustrations, thumbnails, diagrams or visual assets automatically from text coming from a file, an email, an extracted text action, a template or any other text source in the scenario.

Where to add  

Insert this action in any scenario step that has text available (Extract Text, AI summaries/translations, templates, or plain .txt inputs). The action produces a PNG image that becomes the current item and can be saved, attached or passed to downstream actions.

Main settings

• AI model
  Select the OpenAI image model from the AI model dropdown. Check the UI for available models and updates.
• API key (required)
  Enter your OpenAI API key. You must have an account and API access with OpenAI.
  Note: Get your OpenAI key here >
• Prompt
  The textual prompt that describes the image to generate. Prompts can be typed directly or filled from fields/tokens (use Insert fields to include dynamic content such as extracted text, filenames or metadata).
• Quality
  Choose image quality level (for example Standard, High). Higher quality may increase generation time or cost.
• Size
  Choose the output dimensions (for example Square-1024x1024). The action returns a PNG at the selected size.
• Style
  Choose an overall style preset (for example Vivid, Photorealistic, Painterly) if the UI exposes style options for the selected model.

Output

• The action returns a PNG image file. The generated PNG becomes the current item in the pipeline and can be:
  - Saved to disk
  - Uploaded to a server
  - Attached to an email
  - Used as input for other image-processing actions

Supported inputs & prerequisites

1. Any text input in the scenario can be used as the prompt: extracted text, summaries, translation output, text templates or direct user input.  
2. If your prompt comes from documents that require extraction (scanned PDFs, images), run Extract Text or OCR AI before running this action.

Privacy and compliance

• The action sends the prompt (and any included text content) to OpenAI for processing. Confirm that transmitting this data to a third‑party service is allowed by your organization’s policies and applicable regulations. Do not include sensitive secrets in prompts.

How to configure — quick steps

1. Add Generate an Image with AI to your scenario where text is available.  
2. Choose the OpenAI model from the AI model dropdown.  
3. Enter your OpenAI API key in the Api key field.  
4. Enter or map the Prompt (type text or insert fields).  
5. Select Quality, Size and Style options as needed.  
6. Save the action and test with sample prompts.

Behavior notes and tips

• Prompt design: be explicit about composition, color, perspective, and subject. Short descriptive prompts work, but targeted details (lighting, camera angle, mood) produce more predictable results. Example: “Photorealistic 3/4 view portrait of an orange tabby cat wearing a blue scarf, soft natural light.”  
• Use fields for dynamic images: combine static instruction with inserted fields (e.g., “Create a product thumbnail for {FileName} with white background”).  
• Large or complex prompts: keep them focused and concise to reduce ambiguity. Use multiple runs with small variations to find the best result.  
• Post-processing: subsequent actions (resize, convert, watermark) can refine the generated image for final use.

Error handling & troubleshooting

• Authentication errors: verify the API key is valid and has image-generation permissions.  
• Rate limits / quotas: monitor OpenAI usage and quotas; generation requests may be throttled if you exceed limits.  

Security note

• Do not include personal data or confidential content in prompts unless permitted by policy.

That’s done, choose the OpenAI model, supply your API key, provide a prompt (static or from an extracted source), pick quality/size/style and the action returns a PNG image ready for downstream processing.