India, Aug. 29 -- The first phase of enterprise AI was relatively easy to understand. Pick a model. Identify a use case. Build a pilot. Demonstrate that generative AI can summarise a document, assist a developer, answer an employee query or predict what might happen next. Now we are talking about radical trends like autonomous enterprise.

The next phase is proving considerably harder, given the ambition and the expectation.

As AI moves from experimentation into production, enterprises are discovering that the model itself is only one part of the equation. Underneath it sit data, applications, cloud infrastructure, security, governance and costs. Around it sit people, workflows and business processes. And when AI begins acting rather tha...