Export image from chat to external service

I have create a GPT to extract data from scanned document. Data are in json format. Then I create openapi description that allow me to export the json via POST.
I case the use take a picture from his smartphone, I would like to export the image too.
I used the code interpreter to convert image in Base64 but the system always exprt only a part of the Base64 code (first 100 char.)
What would be the best way to export the image?

The image can’t be sent in a post request by the model. And the model won’t give you a download link.

What is the benifit of sending the image? Is it for audit purpose?

I have parking receipts and I’m trying to automate the workflow.
These receipts have a special font, and the contents are recognized by more than 99%.
I made a GPTs, I upload a receipt to the chat and it sends me details in json format, and I created a custom action that can send the json via POST.
In real life, the idea was to take a photo of the receipt using the phone’s applications, and then send the recognized data + the image for the processing process, archiving and auditing.
When we create a custom GPT, is it possible to interact with the API?

A better way to do this is to use automation software on phone.

Example: For IOS/Macos you can create “Shortcuts”. It can send the image to the Vision API, get the JSON, and then send it to Google Sheets or something like that.

For Android, you can use Tasker or similar apps to do the same.

I have created several shortcuts that make life easy for me using GPT-4/Whisper and Vision API. You can search on youtube to get some boilerplate shortcuts for the same.

Thank you, I appreciate your answer.
I also thought about using Nextcloud to take the photos and then do some scripting to automate the workflow.

1 Like