Databricks targets AI operations bottlenecks with ZeroOps
Databricks is pitching a fix for what it sees as the growing operations mess in enterprise AI. With the launch of Genie ZeroOps, unveiled at its Data + AI Summit, the company is targeting a problem many data teams know too well: it’s no longer building pipelines and models that hurts, it’s keeping them running. As data estates sprawl and AI workloads multiply, engineering time is increasingly eaten up by maintenance. Meanwhile, AI coding tools are accelerating development, churning out even more assets that need oversight, widening the gap between how fast teams can build and how much they have to manage. Databricks Genie ZeroOps is a new agentic operations capability that is designed to automate the monitoring, investigation, and remediation of issues across data and AI workloads. Currently in private preview, ZeroOps uses an AI agent to identify anomalies, trace root causes using metadata and lineage information via Unity Catalog, generate proposed fixes, and then test those fixes in