Databricks opened Genie ZeroOps to public beta on November 19, 2026, after private-preview testing of its autonomous data-pipeline agent. The corporate says ZeroOps noticed greater than 80,000 actual buyer points throughout that restricted preview.
ZeroOps detects failures, investigates their causes and prepares fixes. The beta provides shared inboxes, notifications that summarize every problem and suggestion, customized directions for company-specific workflows, and one-click pull-request creation for Git-backed jobs.
Non-public-preview testing coated a whole bunch of corporations, and Databricks studies that greater than half of the problems required organization-specific information to triage, rank and resolve correctly. That discovering drove options that permit groups determine identified flaky sources, most popular fixes and naming conventions, or filter anticipated failures and lower-priority notifications.
From Detection to Reviewed Fixes
ZeroOps watches for specific job and pipeline failures in addition to tables that cease updating or arrive with unexpectedly few rows, issues that will not set off a traditional job failure. It gathers logs, supply code, desk metadata and lineage, runs SQL queries to determine root causes, and prioritizes findings by downstream affect. Proposed fixes are examined with unit checks and dry runs in an remoted sandbox earlier than anybody applies them.
Investigations begin robotically when a difficulty is detected. Constructed-in entry to run historical past, data-quality indicators and Unity Catalog lineage, which maps dependencies between property, connects code diagnostics to the state of the information. Alerts can go to Slack, Microsoft Groups, PagerDuty or generic webhooks; e-mail locations are unsupported.
Approval and Knowledge Controls
ZeroOps doesn’t change manufacturing code or information with out approval. The documented guardrails require an specific person instruction and affirmation earlier than it pushes a Git-backed repair; the push provides a commit with out force-pushing or rewriting department historical past. Naming a repository’s default department can bypass pull-request evaluation until department safety blocks direct pushes.
Databricks says the agent makes use of Anthropic and OpenAI fashions on Databricks-hosted endpoints inside its safety perimeter. Deciding on property authorizes steady, autonomous large-language-model processing till the function is disabled, together with repository contents for Git-backed jobs. Earlier than an exterior notification features a discovering’s title or abstract, the agent checks for delicate info akin to names and e-mail addresses and masks what it finds. Copies saved in exterior techniques fall exterior Databricks’ retention and deletion controls.
A configuration will be shared with as much as 25 particular person workspace customers, however not with teams or service principals. Collaborators can work on its findings, but shared customers see code excerpts, instructed fixes and pattern information rows even with out entry to the underlying property. Sharing doesn’t grant entry to these property.
Setup Necessities and Beta Scope
An administrator should allow Genie ZeroOps from the workspace Previews web page. Customers then choose property and want OWNER, CAN MANAGE or CAN VIEW permission on them. The setup process backfills findings from the earlier 24 hours earlier than reporting new points. Investigating a Git repository requires a credential with learn entry; making a pull-request department requires write entry.
Databricks introduced ZeroOps on June 16, 2026 with deliberate private-preview assist for jobs, pipelines, tables and machine-learning workloads. Present documentation states that monitoring covers jobs, pipelines and tables, not fashions or apps, and that efficiency and price regressions are exterior its detection scope. The beta operates at workspace reasonably than account degree, just isn’t HIPAA compliant, and requires individually billed data-quality monitoring to detect late-arriving or incomplete desk information.
ZeroOps automates investigation and repair preparation whereas leaving manufacturing authority with the client. The beta’s central enterprise trade-off is much less guide prognosis in trade for steady AI processing of chosen operational information, making sharing visibility, exterior notifications and Git evaluation a part of deployment reasonably than separate administrative particulars.

