Why Your AI Ambitions Are Hitting a Data Wall

Everyone’s talking about AI in the boardroom, but the dirty secret is that most companies aren’t ready for it. Not because they lack the latest models or GPUs, but because their data is a mess.

Consumer AI tools are fast and flashy. But when you try to deploy AI at scale inside an enterprise, you quickly run into a problem: the data is fragmented, siloed, and often garbage. Bavesh Patel from Databricks puts it bluntly: “the quality of that AI and how effective that AI is, is really dependent on information in your organization.” If your data is locked away in legacy systems, SaaS silos, and disconnected formats, your AI will be terrible. That’s his word, not mine.

Patel argues that the real competitive differentiator for most organizations is their own data, plus whatever third-party data they can layer on top. But to get there, you need to consolidate everything into open formats, govern it properly, and make it accessible across teams. Without that foundation, you’re building on sand.

Rajan Padmanabhan from Infosys adds that enterprises need to tie AI initiatives directly to business metrics. Don’t treat AI as a science project. Use governance frameworks to figure out what’s actually delivering results and kill the rest quickly. That’s harder than it sounds, because everyone wants to play with the new toy.

One thing I found interesting: Patel talks about AI literacy with business users. They’re eager to understand how to think about AI, but they need help peeling back the layers. What are the building blocks? What does it take from a technology, training, and enablement standpoint? That’s a practical gap that vendors often gloss over.

Padmanabhan also notes a shift from systems of execution or engagement to “systems of action.” AI agents are evolving from copilots into autonomous operators that can manage workflows and transactions. That’s a big deal, but it only works if the data underneath is solid.

This episode was produced in partnership with Infosys Topaz, but the message is clear: the future of enterprise AI depends on whether companies can turn fragmented information into a strategic asset. If your data stack isn’t ready, no amount of AI hype will save you.