Log entry Nº 375 · the stack watch
Bearing SE 135° · November 17, 2026 · 4 min read
Automation & WorkflowsSE 135°Watch log
Databricks Makes Genie Always-On, Adds Ontology and Search
Databricks made Genie Ontology generally available and added always-on agents plus governed Genie Search, but omitted pricing and rollout details as analysts called for stronger action controls.
By James Calloway4 min read856 words

Databricks unveiled 3 Genie changes on Wednesday: always-on agents, general availability for Genie Ontology, and the new governed Genie Search. The announcement disclosed no prices and no availability dates for always-on Genie or Search.
Always-on Genie changes the unit of work from a question to a goal. Users can hand over an assignment, and the agent takes ownership, including delegating tasks to subagents when needed. That is a bigger operational promise than a better answer box, and it removes the human request as a natural speed bump.
Databricks has made Genie Ontology generally available after its June 2024 introduction. The knowledge layer captures company definitions, metrics, relationships and context from structured and unstructured data, wherever that data lives, then makes that context available to agents. Genie Search uses the ontology to ground governed search across the data estate for humans and agents.
Mike Leone, an analyst at Moor Insights & Strategy, put the practical promise plainly:
“Taken together, I'd call [the new capabilities] significant because Genie stops waiting to be asked.”
He said users previously needed the right question and the right moment; handing Genie a goal can put answers before people who would never open a dashboard.
What changes for practitioners?
Databricks has expanded Genie continuously since its June 2024 introduction. The current family covers 5 named areas:
- Genie One provides natural-language data analysis.
- Genie Agents helps users build domain-specific agents that reason from curated datasets and business logic.
- Genie Code helps data engineers and scientists write and debug code for building pipelines.
- Genie Ontology supplies governed business context and definitions.
- Genie Search adds unified search across the data estate.
Databricks said many features were already planned, but customer requests shaped 2 additions. Ontology Manager lets data teams govern business context and recommends refinements to ontologies used most often to inform agent actions. Databricks also extended the sources for business logic to external business-intelligence platforms, where analyst Sanjeev Mohan says much of that logic still lives.
Ken Wong, senior director of product management, said customers increasingly asked about context outside Databricks and traditional data systems: documents, tickets and chats. The company had federated connectors at launch, but built Genie Search to index those systems as well because it wanted better performance.
Since Ontology entered preview, Databricks also added 3 capabilities:
- Live dashboards and interactive reports rooted in governed data.
- Agents that work across data sources alongside governed tables.
- Access to Genie agents from any workspace, including web browsers and mobile applications.
For practitioners, the combination lets teams hand off recurring analytical work rather than repeat it by hand. The catch is familiar: an agent can act on a missing definition faster than a person can spot the omission.
Mohan, founder and principal of SanjMo, called Ontology perhaps the most valuable new capability because always-on Genie works only if the agent understands the business. Always-on will get the headlines, but agents fail less from bad models than from missing context.
Without a shared definition of “active customer” or the revenue number finance trusts, fluent reasoning can still produce a wrong answer. Ontology learns those definitions, metrics and relationships from data, usage and actual queries.
Is Databricks ahead of its rivals?
Databricks is among the first vendors to deliver an ontology layer, according to Leone. He said it remains in step with the market's broader direction toward autonomous coworkers but pulls ahead on ontology because it learns a company's definitions from actual behavior; teams need not write every definition first.
The field is crowded. Rival Snowflake and hyperscalers AWS, Google Cloud and Microsoft are developing similar tools for autonomous coworkers. Leone said the whole data-platform market is chasing ontologies, including ones AI builds and manages with people approving changes.
Leone ranked Genie Search as perhaps the most valuable new feature. It lets agents avoid separate searches and access grants app by app, with one governed search, one place for permissions and fewer calls to pay for. In practical terms, the agent spends less time hopping among systems and more time assembling an answer.
Can enterprises trust autonomous agents?
Mohan said competitive convergence does not settle the harder control problem:
“They make real progress on governed access, [but] enterprises need to govern what agents do, not just what they read.”
He specified 4 controls: a clear identity for each agent, limits on permitted actions, human approval for consequential steps and a full audit trail covering reasoning and actions.
Leone broadly agreed that Databricks keeps roaming agents mostly in bounds. His missing safeguard is simple: always-on Genie should say when it does not know or is unsure. An unprompted wrong answer can reach many people before anyone questions it, so uncertainty needs to travel with the answer.
Databricks still has not answered the commercial and operational questions. It disclosed no pricing for Ontology, always-on Genie or Search; no general-availability date for always-on Genie or Search; and no accuracy benchmark for autonomous work. Ontology's general availability and Search's grounding in Ontology are clear. The cost, rollout risk and trust at scale are not.
Plot points
- Databricks announced 3 Genie changes on Wednesday: always-on agents, generally available Genie Ontology and Genie Search.
- Since its June 2024 introduction, Databricks has expanded Genie across 5 named areas: One, Agents, Code, Ontology and Search.
- Customer feedback shaped 2 additions: Ontology Manager and business-logic context from external BI platforms.
- Databricks says Genie Search indexes documents, tickets and chats, supplementing federated connectors present at launch.
- Analysts specify 4 production controls: agent identity, action limits, human approval and full audit trails.
via techtarget.com (Source)
More from James Calloway
Show full bio
Correspondent covering people, launches and day-to-day practice in marketing technology at Martech Beacon.
31 articles
Neighbouring fixes
- Databricks Genie Plugins Bring Governed Context to ChatGPT, Claude, Copilot and Grok
- Databricks Enters Marketing With Agentic Customer Data Platform
- Databricks Launches CustomerLake, Its First Agentic CDP for Marketing
- Databricks Launches CustomerLake, Takes Aim at the CDP Category
- Forrester Calls Databricks CustomerLake the Litmus Test for Agentic Marketing