How Bundesliga built Captain, an AI agent for fans, using Amazon Bedrock

This article describes how Bundesliga built Captain, an AI-powered fan companion app using Amazon Bedrock that provides real-time insights, statistics, and personalized storytelling during live matches.

• Captain uses agentic AI to proactively surface match insights and personalized storylines to fans in real time
• Multi-agent architecture routes questions: Nova Pro for simple queries, Claude Sonnet 4 for complex reasoning
• Text-to-SQL workflow converts natural language questions into database queries via Amazon Athena and S3 Tables
• Research agents autonomously investigate milestone moments to uncover compelling data-driven stories within seconds
• Data foundation uses S3 Tables (Apache Iceberg), Amazon MSK ingests 200M data points per match
• Chat assistant includes Router, Stats, and Video agents with content safety guardrails
• Dynamic routing reduced chat service costs by 35% while maintaining answer accuracy
• Vector database stores question-to-SQL templates for faster, cheaper repeated queries

Captain demonstrates how agentic AI transforms fan engagement through autonomous data analysis, personalized storytelling, and real-time insights at scale during live football matches.