Workflows · 2 min read
AI Analyze Image Action
The AI Analyze Image action reads an image and returns a text answer to your prompt. Use it to pull text from receipts and documents, describe product photos, spot objects or details, or check what an image shows.
Set up the action
- Go to Automation > Workflows and open your workflow.
- Click +, open AI Actions, and choose AI Analyze Image. Give it a clear name.
- Choose a Model. If you're not sure which one to pick, test the same image and prompt with each and compare the results.
- Enter the Image URL. It can come from an earlier step, such as an AI image generation action or a customer's attachment, or from a custom value.
- Write the Analysis Prompt. Be specific, for example:
- "Extract all text visible in this image."
- "Describe this product image in detail."
- "Read the total amount from this receipt and return only the number."
- Set the Detail Level. Auto is the default. Use High for fine detail or Low for a quicker, broader look.
- Save the action.
Image URL requirements
The image must be public, because the AI loads it straight from the URL. The URL can't:
- Need a login, password, or token.
- Sit behind a private share link, restricted drive, or internal network.
- Point to a web page that contains the image instead of the image file itself.
To check, open the URL in a private browser window. If the image loads without signing in, the action can reach it. Images in your Media Storage work.
Use the answer
The answer is plain text. Find it under AI Analyze Image in the custom value picker of later steps. You can save it to a custom field, add it to an email or SMS, send it by webhook, or branch on it with If/Else.
Example
- Trigger: Customer Replied, with an image attached.
- AI Analyze Image: image URL from the attachment, prompt "Extract all text visible in this image", detail level High.
- Update Contact: save the text to a custom field for review.