Inspiration: https://developers.googleblog.com/en/unlocking-multi-spectral-data-with-gemini/
use process API to generate a series of images from each of our PlanetScope evalscripts in custom scripts repo and then fed the images + the evalscript descriptions from the repo to the model to help AI do it's interpretation