Operator Insights

Reported by others2

1. 

   Mary AlfheimVP Data & AnalyticsHeadspace

   Organization-wide. 30+ active users, up from six analysts fielding requests.

   Databricks, published Oct 5, 2026Read the case study (opens Databricks in a new tab)

   Observation

   At Headspace, metric definitions were agreed before any code, then domain Genie Agents answered behind one routing agent.

   Takeaway

   Agree on metric definitions first. Otherwise an AI agent improvises a different definition for each person who asks.

   What happened

   Headspace brought business and technical leaders together to settle ownership and accuracy thresholds for core metrics, then rebuilt its data model in three months from versioned contracts on Databricks. A routing agent called Aladdin sends each question to one of 11 domain Genie Agents, which show the exact SQL, and agents decline questions outside their governed scope. More than 30 people now query governed data this way.

   • Vendor case study
   • User counts reported by Databricks and Headspace
   • Routing agent powered by Claude

   Deployment details

   Organization-wide. 30+ active users, up from six analysts fielding requests.

   • Vendor case study
   • User counts reported by Databricks and Headspace
   • Routing agent powered by Claude

   Team

   Headspace data and analytics

   Duration

   Three-month data model rebuild

   Workload

   Natural-language questions across 13 domain schemas through 11 Genie Agents

   Product version

   Genie Agents behind a Claude-powered routing agent

   Model

   Claude for the Aladdin routing agent

   Measurement

   Active users, contract change time, and conformed entities

1. BookMyShowSep 23, 2026DatabricksAt BookMyShow, business teams answer their own questions in domain Genie Agents, and the central data team no longer fields every request.

   

   Noel CurtisChief Technology OfficerBookMyShow

   Company-wide. Finance, marketing, live events, cinemas, and engineering. Unity Catalog adopted first.

   Databricks, published Sep 23, 2026Read the case study (opens Databricks in a new tab)

   Observation

   At BookMyShow, business teams answer their own questions in domain Genie Agents, and the central data team no longer fields every request.

   Takeaway

   Invest in the catalog first and build a narrow agent for each domain. Point users to those agents, which answer better than the general chat.

   What happened

   BookMyShow had run every analytics request through one central team. After moving to Databricks and Unity Catalog, it built more than 80 domain-specific Genie Agents for teams such as marketing, live events, and cinemas. It steers users to those agents because outcomes are better than in the general Genie One chat. The central team now maintains Unity Catalog and pipelines while business teams answer day-to-day questions in Genie. Databricks credits the 90% drop in dashboard and report tickets to teams using Genie Code for independent analysis.

   • Vendor case study
   • Ticket reduction reported by Databricks and BookMyShow

   Deployment details

   Company-wide. Finance, marketing, live events, cinemas, and engineering. Unity Catalog adopted first.

   • Vendor case study
   • Ticket reduction reported by Databricks and BookMyShow

   Team

   BookMyShow central data team and business teams

   Workload

   Self-serve questions in 80+ domain Genie Agents, Genie One, and Genie Code

   Product version

   Genie One, Genie Agents, and Genie Code with Unity Catalog

   Measurement

   Dashboard and report tickets to data engineering, down 90%, credited to Genie Code for independent analysis