How UFC Uses IBM watsonx to Triple Fight Insight Generation

UFC, the world's premier mixed martial arts organization reaching 700+ million fans and broadcasting to 950+ million households across 170+ countries, needed to transform how it delivered fight insights across 40+ annual live events. The existing process relied heavily on manual expertise and institutional knowledge to analyze large volumes of fast-changing, domain-specific data — a time-consuming approach that limited scalability and speed during fight weeks.

UFC partnered with IBM to build the UFC Insights Engine, an AI-powered platform that automates fight and fighter insight generation. The core infrastructure uses IBM watsonx Orchestrate to enable a pre-fight insights agent, allowing UFC editors to mine insights through natural language queries. The unified conversational interface routes complex questions to appropriate AI components, integrating text-to-SQL pipelines for structured data queries and retrieval-augmented generation pipelines powered by IBM Granite and Llama models.

IBM watsonx.ai extracts insights from UFC's historical data library — including fight statistics, athlete profiles, and event outcomes — by identifying key performance indicators and generating contextual analysis. IBM watsonx.governance ensures explainability and transparency in AI outputs, while IBM watsonx.data manages vectorized data retrieval for efficient information access.

The platform is estimated to deliver a 3x increase in insight volume and a 40% reduction in query generation time, generating hundreds of contextual, data-driven insights across fighters, matchups, and events. The system has significantly improved response times to stakeholder content requests during fight week, enabling faster, smarter, and more engaging content delivery across broadcast, digital, and social platforms.