Nexmo, a company, used Facebook Workplace to share photos of their pets and noticed that some people were asking about the breed of dogs in the photos. To solve this problem, Nexmo built a machine learning model using Keras to classify dog breeds from images. The model was trained on a dataset of over 10,000 images of dogs from the Dog Breed Identification Challenge on Kaggle. The model used transfer learning with Xception as the pre-trained base model and added custom layers for classification. After training the model for 10 epochs, it achieved an accuracy of 99% in classifying 12 breeds. A function was created to make predictions on new images, which took a photo from the internet, formatted it to the expected size, made a prediction using the model's `predict()` method, and returned the breed name sorted alphabetically.