I had a moment to sit down and gather my thoughts on Databricks and their Data + AI Summit 2026 in SF last week. The deeper read starts with what they shipped together, not what they shipped individually.
Most enterprise AI conversations focus on which model is smartest. Business leaders should be asking a different question. Which platform actually holds your company’s knowledge in a way that AI can use, and which platform controls who gets to see what.
Databricks positions itself to handle both sides of that question at once. Genie Ontology organizes everything your business knows, from the formal data in your warehouse to the working knowledge sitting inside every connected enterprise application, and Omnigent is the agent platform that puts the knowledge to work. Unity AI Gateway decides which AI agents are allowed to read which pieces of that knowledge, when, and under whose permissions, with every access logged.
The architectural pattern at work here is Glass-Box Governance. That encapsulates my broader point. The same kind of accountability that companies have always demanded from human employees is now applied to AI agents. The Model Context Protocol becomes something the enterprise can govern with confidence. Open standards become safer to adopt when the controls around them are real.
Ali Ghodsi described the competitive moment at the analyst session this week as a bottom-up versus top-down one. The frontier labs are betting on smarter models. Databricks is betting on owning the place where your enterprise knowledge lives. Mind you, the bet only works if knowledge and control ship together. The company that organizes all its knowledge for AI without governance ends up handing that knowledge out the front door to whichever model walks in.
The procurement question for CIOs reshapes accordingly. Skip the question about which platform has the smartest agents. Ask which platform can show you exactly what knowledge each agent is reading, who authorized it, and what was done with it. One question is the marketing question. The other is the architectural one.
Special shout out to one of my Day Ones, Sonya Vargas (Ruzheva) and her team for having me and putting on a great analyst program. See you, Kayli Berlin. Thanks again, y’all.