This is an iOS Shortcut to interact with the new vision API for live prompting via the camera.

Flow:

Execute shortcut <Prompted for voice input, this will be your prompt> Photo is then automatically taken with the back camera. Photo is uploaded to imgur (this can be changed to send it base64 encoded, i jsut preferred the imgur urls) Prompt and the photo is passed to GPT vision API Response is dictated back via audio.

Enter OpenAI API key on import at the dialogue. Have fun!