How to Describe Uploaded Images with a Make Webhook and GPT Vision
Receive an image URL through a Make webhook and get an alt-text description back from a vision-capable model.
Vision models can look at a photo and write a description, which is perfect for generating accessible alt text at scale. In this guide a Make webhook receives an image URL, sends it to a vision model, and returns clean alt text you can store with the asset.
What you need
- A Make account
- An OpenAI API key with access to a vision-capable model
- A publicly reachable image URL to test with
Step 1: Receive the image URL
Add a Custom webhook module and send a sample POST with an imageUrl field so Make learns the structure. The image must be reachable over the public internet for the model to fetch it.
Step 2: Build the vision request
Add an HTTP Make a request module pointing at https://api.openai.com/v1/chat/completions. The user message holds an array with a text part and an image_url part.
{
"model": "gpt-5",
"messages": [
{
"role": "user",
"content": [
{ "type": "text", "text": "Write concise alt text under 120 characters." },
{ "type": "image_url", "image_url": { "url": "{{1.imageUrl}}" } }
]
}
],
"max_tokens": 80
}Step 3: Save the description
The alt text lives in choices[0].message.content. Map it into a later module, for example to update the image record in Airtable or a CMS so the alt attribute is filled automatically.
Example: a product photo URL comes back as Black leather backpack standing upright on a wooden desk, which you write straight into the alt field of your store.
Watch related tutorials
05:00
13:00
30:00
22:00
25:00
19:30