Wink Pings

AI Is Getting Smarter, but Enterprises Aren't Keeping Up? Databricks CEO: The Key Is Business Context

AI model capabilities are increasing, but AI adoption inside enterprises hasn't kept pace. At DATA + AI LIVE, Databricks CEO Ali Ghodsi said the problem lies in business context. Genie Ontology automatically builds a permission-aware metadata graph, so AI agents know where to find data—delivering faster answers and using fewer tokens.

AI models are getting smarter, but progress inside enterprises hasn't kept up. Databricks co-founder and CEO Ali Ghodsi said this at DATA + AI LIVE, and explained: what's missing is business context.

Enterprise knowledge is scattered across different systems, and not everyone has permission to access all data. Letting AI search through all data aimlessly is slow, expensive, and incomplete. No matter how strong the model is, if it doesn't know which data is available and which it can touch, it's like a new employee who hasn't been given the company directory and permission handbook—they can only look for answers in the wrong places.

Ali introduced Genie Ontology on stage. It automatically builds a metadata graph across enterprise data and connected systems, and it is permission-aware. Put simply, this graph knows where each data source is and who can access it. When an AI agent receives a question, it first consults this graph and only goes to relevant places where it has permission, so answers are faster and more accurate, and fewer tokens are consumed.

Databricks builds data lakehouses—put simply, it helps enterprises manage data in a centralized way. Genie Ontology is the navigation map it provides for AI agents. For enterprise users, it solves the problems of data discoverability and permission control: AI won't rummage around, and it won't make wild guesses because it can't see data.

Video:

Full talk replay: DATA + AI LIVE: Context Is Just the Beginning

发布时间: 2026-11-20 05:48