Sevilla FC x IBM watsonx (Scout Advisor)
Three hundred thousand scouting reports, one question in plain language, a summary in two minutes.
Sevilla FC's data department built Scout Advisor with IBM, a generative AI application on watsonx that uses Llama 3.1 70B Instruct in a RAG pipeline to query a database of more than 300,000 internal scouting reports in natural language. In production since January 2024, the system cuts the time to summarise reports on a single player from hours to seconds. Chief Data Officer Elías Zamora describes the goal as bridging the gap between scouts' qualitative judgement and quantitative data analysis.
Por qué importa
This is the most explicitly documented case in the entire scouting segment: named model, named platform, report count, stated processing time. The structural risk sits apart from the technical one: a system that condenses hundreds of scouts' human judgement into a single answer concentrates the editorial interpretation of talent inside one model, trained on one proprietary dataset. Sevilla has already started selling consultancy to other clubs to replicate the architecture, which means the same judgement model risks spreading across other European scouting departments.
Companies