Image Handling

The Image Handling tool allows your bot to understand images from users, including any text within them via OCR (optical character recognition), and send relevant images from articles in response.

You can navigate to Bots → Select your bot → Tools → Preset → Image Handling

---

Configuration

Click View Details on the Image Handling card to open its settings.

How should the bot respond to images? (Optional)

This field lets you guide the bot on what to notice or how it should react when it receives an image. It is optional — if left blank, the bot will read everything it can find in the image and use its own judgment on what's relevant.

You can use this field to set conditional actions based on what the bot reads in the image, such as escalating to a human agent when a certain condition is met.

Example:
If the image is a receipt and the total amount is more than $10,000, call the human agent.

Use this field when you want to:

• Direct the bot to extract specific information from images (e.g. order numbers, totals, dates)
• Set expectations for what the bot should do next when it detects a particular type of image
• Trigger a specific action, such as human escalation, when a value read from the image meets a condition
• Keep the bot's responses consistent across images that share similar content

Once you're done, click Save to apply your changes.

---

Best Practices

• Enable this tool if customers commonly report visual issues or errors, share screenshots for clarification, or ask product-related questions that involve images
• Enable this tool if customers often share a photo or screenshot of a receipt or invoice, so the bot can read the text inside the image (e.g. order numbers, totals)
• Use the prompt field when your business deals with predictable image types (e.g. receipts, IDs, error screenshots) so the bot consistently extracts the same key details
• Leave the prompt blank if your image use cases vary widely and you're fine with the bot relying on its default judgment