In the process, they ‌discover that making AI productive inside a complex organisation is proving harder than gaining access to the technology. The challenge increasingly lies not in choosing a model, but in making AI work with the software, data and business processes companies already use. "AI applications are the battleground, and that is where most value will be created," UBS said in a recent note. AI ⁠systems must work with software built up over decades, fragmented databases, customised applications and increasingly complex governance requirements. The company's acquisitions of data specialist Dremio and AI company Prior Labs underline the growing importance of making enterprise data accessible to AI applications.