Databricks Bets Big on AI Coworkers with Genie One, Ontology, and ZeroOps at Summit 2026

The assistant is available on the web and through native iOS and Android apps, and it integrates across the collaboration tools teams already use, including Slack, Microsoft Teams, and mobile apps . For enterprise workflows, users can tag Genie in a Microsoft Teams thread to receive context-aware answers governed by Unity Catalog. Genie for Microsoft Teams and M365 Copilot entered Beta at the summit . Genie One can also connect to external knowledge stores such as Google Drive and SharePoint, expanding its ability to reason over broader enterprise context .

Genie Ontology: A Live Map of Your Business

A key architectural shift is Genie Ontology, which entered Public Preview on June 15, 2026 . Rather than having Genie search for context from scratch with each query, the Ontology acts as a continuously updated map of the business. It automatically extracts and ranks knowledge from dashboards, notebooks, pipelines, and connected workplace applications . Databricks calls it a "web of all knowledge" that lets Genie One identify trustworthy sources and answer with full business context .

On internal benchmarks published in May 2026, Databricks claimed that Genie with Ontology improved query accuracy "from 32% to over 90%" compared to a leading coding agent . This is a striking number, but it is important to note the caveat: the 32% to 90% gain is on internal Databricks evaluations, not external, independently verified benchmarks. External analysts have flagged this as a useful directional signal, but less useful as a head-to-head comparison against other agentic stacks .

Genie Agents Evolve from Genie Spaces

Databricks used the summit to evolve its Genie Spaces concept into Genie Agents. Users can now save any Genie conversation as a reusable, shareable agent that inherits the conversation's memory, sources, instructions, and behavior. Coworkers can call on the agent by name, and it repeats the same reasoning approach . Data teams can still create Genie Spaces as domain-specific environments that configure trusted data, metrics, and business rules to power Genie One answers .

Genie App Builder and Genie ZeroOps

The Genie ecosystem now extends beyond analytics. Genie App Builder lets teams "vibe code" simple business applications governed by Unity Catalog, expanding the platform into lightweight app creation .

Genie ZeroOps, announced in private preview, aims to put data and AI operations on autopilot. Initial support includes jobs, pipelines, tables, and ML workloads, with apps and Lakebase databases on the near-term roadmap .

A Pay-As-You-Go Pricing Shift on July 6, 2026

Perhaps the most scrutinized business change is the new pay-as-you-go pricing model for all Genie products—Genie One, Genie Spaces, and Genie Code. Starting July 6, 2026, each user receives 150 DBUs of free LLM usage per month (roughly $10.50 in US East, though the dollar equivalent varies by region) . Usage beyond the free allowance is billed by DBU consumption, while SQL Serverless compute is billed separately .

Account administrators can begin setting budgets and cost controls immediately through the Unity AI Gateway Budget capabilities, with tracking available across accounts, workspaces, groups, and individual users .

Broader Summit Context: LTAP, Lakehouse//RT, and Notable Keynotes

The Genie announcements were part of a larger platform story at a summit Databricks calls the world's largest data and AI conference. Co-founders Ali Ghodsi, Matei Zaharia, Arsalan Tavakoli-Shiraji, and Reynold Xin were joined on the keynote stage by guest speakers including Satya Nadella (in a pre-recorded fireside chat), OpenAI's Greg Brockman, and PepsiCo's Magesh Bagavathi .

Two major infrastructure announcements rounded out the event:

• LTAP (Lake Transactional/Analytical Processing): A new architecture that unifies OLAP and OLTP on a single copy of data in the lake, designed to eliminate ETL and replicas. The foundation of LTAP is Lakebase, which now handles 12 million database launches per day across the platform .
• Lakehouse//RT: A real-time Lakehouse powered by a new compute engine called Reyden, delivering millisecond query latency at tens of thousands of concurrent users and agents directly on governed Delta Lake and Apache Iceberg tables. Databricks reports up to 16x better performance than existing real-time serving stacks, with response times as low as 10ms on smaller datasets .