AI News

Trending story

Databricks and Turing Build Enterprise Data Agent Benchmark

2 reports2 story updatesUpdated Nov 11 23:29

Understand this story

AI overview

Databricks partnered with Turing to build an enterprise data agent benchmark, comparing Genie with two frontier agents in a real Databricks workspace. According to information published on the Databricks blog on November 11, 2026, Genie achieved a 91% pass rate, while the two frontier agents achieved 45% and 58% respectively; Genie's average token usage was less than half that of its competitors, and it was 3 to 4 times faster. Databricks attributed the gap to the business context that Genie Ontology extracts from notebooks, dashboards, and saved SQL queries, and announced that Genie Ontology is now generally available.

AI-generated from public reports

Latest development · Report

Databricks Upgrades Genie to an Autonomous Enterprise AI Collaboration Assistant

Developments

1 confirmed developments

  1. Databricks and Turing build an enterprise data Agent benchmark

    Source:Databricks Blog

Reports and discussion timeline

Browse reports and discussion by time. Their evidence levels remain distinct.

Nov 11

  1. Unite.AI

    Databricks Upgrades Genie to an Autonomous Enterprise AI Collaboration Assistant

    On November 11, 2026, Databricks launched a new version of Genie, positioned as an autonomous AI collaboration assistant capable of continuously carrying out long-running enterprise work. It supports scheduled or event-triggered tasks, asks follow-up questions to fill in missing information, and manages multiple tasks simultaneously, with users deciding which operations require approval. Genie Ontology and Genie Search for Google Drive, Jira, Confluence, and Workday are now generally available, and native MCP connectors are also GA, enabling retrieval of context and execution of operations such as creating tickets and updating records. Benchmarks from Databricks' collaboration with Turing show that Genie achieved a 91% pass rate during task runs, compared with 58% for the stronger of two anonymous frontier agents, consuming an average of 431,829 tokens and taking 100.3 seconds per run.

  2. Databricks BlogFirst-party

    Databricks Releases Genie Benchmark: Genie Pass Rate 91%, Token Usage Less Than Half of Competitors

    Databricks partnered with Turing to build an enterprise data agent benchmark, comparing Genie with two frontier agents in a real Databricks workspace. Genie achieved a 91% pass rate, while the two frontier agents achieved 45% and 58% respectively; at the same time, Genie's average token usage was less than half that of its competitors and it was 3 to 4 times faster. Databricks attributes the gap to the business context that Genie Ontology extracts from notebooks, dashboards, and saved SQL queries, and announced that Genie Ontology is now generally available.

Related stories

  1. Related storiesBaidu Wenxin releases ERNIE 5.5 Flash Preview
  2. Related storiesGoogle launches Gemini 4 Hydrogen on Antigravity
  3. Related storiesArena.ai teased the upcoming release of Tencent Hunyuan Hy4, and Tencent Hunyuan subsequently launched the official version
  4. Related storiesDeepSeek open-sources inference framework DeepInfer