Descriptions:
In this Bloomberg Technology interview recorded at the Databricks Data + AI Summit in San Francisco, CEO Ali Ghodsi argues that AGI-level productivity is already within reach — but the bottleneck is not intelligence, it’s context. Despite AI systems that audiences overwhelmingly rate as smarter than many colleagues, agents haven’t flooded the workplace because they lack the organizational knowledge graph that humans take for granted: who opened what document, what that document means, how it connects to everything else.
To close that gap, Databricks is launching Genie Ontology — an enterprise-wide knowledge index computed behind the scenes that lets AI agents tap pre-built context rather than crawling documents one at a time. Ghodsi compares current agents to a search engine that reads one website at a time before Google invented the index. Genie 1, built on this ontology, “computes rather than recites” — it can answer questions about live data that don’t exist in any document. Novo Nordisk is cited as an early customer, using Genie 1 to accelerate analysis of obesity drug trial results across control groups in real time.
Ghodsi also unveils LakeBase, a unified database designed for agentic workloads — combining transactional and analytical capabilities that today require two separate systems. Prada is named as a customer storing and querying KPIs through LakeBase-powered agents. On Databricks’ own IPO timeline, Ghodsi reiterates he wants to wait for calmer market conditions before listing, despite the current wave of major tech offerings.
📺 Source: Bloomberg Technology · Published June 16, 2026
🏷️ Format: Interview
