Most operational systems are full of signals, but the signals are not yet knowledge. They sit inside tools, messages, workflows, exceptions, and half-remembered decisions.
The hard part is not just collecting more data. It is giving the organization a way to describe itself clearly enough that software can reason about what is happening.
Prediction comes later. First, the world has to be represented with enough structure that a model can stop guessing in the dark.