Data science takes the next step as it delivers the intelligence in AI. Companies deploy algorithms to turn knowledge from the past into accurate predictions about the future. For example, using sales patterns of historical purchases, organisations build algorithms to predict which clients will buy a certain product next. Modern data science also involves anomaly detection (e.g. fraud detection), recommendation engines (e.g. cross-selling the right product or service to the right client), and natural language processing (e.g. routing emails to the right agents).
Usage of data is not a one-off analytical exercise. When Netflix creates a recommendation engine to recommend series to their viewers, it should be put into production. An MLOps framework takes an experimental machine learning model and makes sure it robustly produces relevant output. In that sense, the analysis becomes a data product. Sometimes it is further integrated into a full-blown digital product, such as an API, a mobile app, or a website. In some occasions, the data product suffices (consider a bank where bankers receive the outcome of the credit evaluation), in other cases, the digital product enforces the solution (consider a bank where clients can compute the outcome of their credit evaluation through their banking app).