Read and Describe Images with the AI Analyze Image Action

AI Analyze Image gives a workflow the ability to look at a picture and answer your question about it in text. Use it to read text from screenshots and receipts, describe photos, list what is visible, or verify that an image shows what it should. It is a premium action billed per run.

Settings

• Model: choose one of the available vision models. If unsure, run the same image and prompt through both and compare.
• Image URL: the picture to inspect, typed in or inserted from an earlier step or contact field.
• Analysis Prompt: instructions telling the AI what to look for and how to answer.
• Detail Level: Auto (default), High for fine detail, or Low for a faster, broad read.

Add it to a workflow

1. Click + and choose AI Analyze Image under AI Actions.
2. Name the action.
3. Pick a model.
4. Provide the Image URL.
5. Write a specific prompt, for example "Read the total amount on this receipt and return only the number."
6. Choose a detail level and save.

The answer becomes a variable for all later steps.

The URL must be public

The AI fetches the image itself, so the link must open without signing in. A quick check: open it in a private browser window. It fails if it needs a login or token, sits behind a restricted share link, or points to a web page instead of the image file. Images in your Media Storage, including ones made by AI Image Generation, are already public.

Prompt ideas

• "Describe this product photo in two sentences."
• "Extract all text visible in this image."
• "List every item you can see."
• "Does this image contain a person? Answer yes or no."

Tips

• Ask for a strict output format when you plan to store the answer in a field.
• Combine with AI Extract Data to turn a long description into structured values.

Good to know: If the action errors out, the first thing to check is whether the image link is publicly reachable.