X4 - integrating the model

• often the goal is to integrate the model into an app

• to easily do this, endpoints can be used

• the trained model is exposed as a service that apps can call

• the app sends a request to the end point, the hosted model generates a prediction, and the response is instantly returned

• simple and consistent

• eg: a mobile banking app detecting potential fraud

  • user submits a transaction
  • app calls a real-time Azure Machine Learning endpoint with the transaction details
  • model evaluates the data and immediately returns a risk score
  • app blocks, approves or flags the transaction in seconds

image: AI-300, Microsoft Learn