Databricks announced the general availability of the Genie One Model Context Protocol (MCP) server for all users. It lets any agent query Genie One as a peer, with answers grounded in Genie Ontology across structured and unstructured data. Governance runs through Unity Gateway, with a noted gap around per-user permissions for service-principal auth.
Databricks has renamed several AI agent products: Genie Spaces are now Genie Agents, Multi-Agent Supervisor is now Supervisor Agent (GA in February), and Knowledge Assistant went GA in late January. A new front door called Genie One sits above these for business users. The changes, effective June 2026, have caused confusion as many blog posts still use old names.
Databricks has renamed several AI agent products: Genie Spaces are now Genie Agents, Multi-Agent Supervisor is now Supervisor Agent (GA in February), and Knowledge Assistant went GA in late January. A new front door called Genie One sits above these for business users. The changes, effective June 2026, have caused confusion as many blog posts still use old names.
A four-year Databricks admin found Anthropic's official Claude Academy tutorial on using Databricks for data analysis underexplained and outdated, particularly around connecting Claude to Genie One. They shared a YouTube video guide as an alternative to help others integrate Claude with Databricks.
A four-year Databricks admin found Anthropic's official Claude Academy tutorial on using Databricks for data analysis underexplained and outdated, particularly around connecting Claude to Genie One. They shared a YouTube video guide as an alternative to help others integrate Claude with Databricks.
MathKernel: An evidence-aware multi-engine mathematics kernel and MCP server
MathKernel: An evidence-aware multi-engine mathematics kernel and MCP server
A developer reports testing Databricks' Genie Code for multi-step data tasks (profiling, training, logging, dashboards) because it respects Unity Catalog permissions and asks before mutating tables, unlike generic MCP-based agents that lack lineage and governance.
A developer reports testing Databricks' Genie Code for multi-step data tasks (profiling, training, logging, dashboards) because it respects Unity Catalog permissions and asks before mutating tables, unlike generic MCP-based agents that lack lineage and governance.
A Databricks user reports that a newly set up Genie agent fails to deliver production-level responses despite following instructions and sample SQLs, and asks the community how to improve accuracy for business rollout. The post highlights the practical challenge of semantic modeling and tuning needed before Genie Agents can be trusted by business users.
A Databricks user reports that a newly set up Genie agent fails to deliver production-level responses despite following instructions and sample SQLs, and asks the community how to improve accuracy for business rollout. The post highlights the practical challenge of semantic modeling and tuning needed before Genie Agents can be trusted by business users.