Many enterprise AI ambitions run into the same obstacle: the data an agent needs is spread across systems it cannot easily reach.
Amit Sharma, Founder and CEO of CData Software, discusses how that challenge has shifted from a reporting and analytics problem to an AI one. CData originally focused on helping companies connect fragmented SaaS systems for decision-making. Now the same issue shapes how useful AI can be inside an enterprise. For many use cases, Sharma argues, it is more effective for an AI system to reach into live systems than to rely only on data that has already been moved into a warehouse.
He breaks that requirement into three parts: connectivity, context, and control. An AI system needs a way into the source system, an understanding of what that data means inside a specific business, and clear guardrails around who can access what. That becomes especially important in areas like sales, where answering one question may require live access to call transcripts, correspondence, ticketing data, and CRM records.
As enterprises push further into AI agents and automation, Sharma makes the case that data connectivity is becoming a much more important part of the software stack.