Etihad Airways puts AI in business users’ hands and turns hours of catering cost analysis into minutes with Databricks Genie

Breaking down data silos for AI readiness

Etihad Airways set out to put the power of AI in every business user's hands, technical or not. Leaders no longer wanted static dashboards showing what happened. They wanted to know why, what action to take and what was coming next.

But Etihad's data was spread across disparate tools and duplicated in silos. In finance, the challenge was even bigger. When a new finance analytics function launched, its workspace held one schema and no data. Reporting relied on legacy BI tools and RPA bots moving data around.

Empowering business users with Genie Agents

Etihad built its lakehouse on the Azure Databricks Data + AI Platform, bringing more than 80% of company data into one place as reusable data products that are built once, governed once and used everywhere. The finance team first reset its foundation, focusing on the data model, semantics and governance and then delivered AI/BI Dashboards and Genie Agents directly to business users.

Adoption of Genie Agents for catering analysis

Finance's first release was a Genie Agent and AI/BI dashboard that itemized roughly 10,000 food items from hundreds of caterers across Etihad's network. Teams can now ask what's driving costs and compare prices across cabin classes and routes, cutting analysis that took hours down to 2 minutes. Since then, the team has launched more Genie Agents, including one for exploring metadata across the company, plus a RAG application over about 10,000 fleet contracts built with Databricks AI Search. The small team uses Genie Code with markdown playbooks and guardrails to rapidly prototype new dashboards and Genie Agents.

Scaling Genie Agents with governance and user training

Genie Agents sit on Unity Catalog, so the same attribute- and role-based access controls govern every query, dashboard and agent, keeping sensitive data from reaching the wrong user. Unity Gateway tracks agent cost and access. The finance team also invested in training for business users, and within about six months, 66 users were active in a single 30-day period, supported by the smallest analytics team at Etihad. Across the platform, demand for data products is now outpacing supply.