Case Study: UNESCO achieves evidence-based education decision-making with Capgemini

Enabling evidence-based education decision-making with agentic AI

UNESCO faced a global foundational learning crisis, with millions of children out of school and many unable to read basic text, while education policymakers often lacked the time, expertise, and infrastructure to use large-scale assessment data effectively. To help address this, UNESCO partnered with Capgemini and AWS on Capgemini’s Global Data Science Challenge, using generative AI to turn underused learning data into actionable insights.

Capgemini implemented an agentic AI solution built with AWS tools such as Amazon Bedrock and SageMaker to analyze the PIRLS 2021 reading literacy dataset and combine it with additional verified sources. The winning system could answer complex policy questions with text and visualizations, helping make insights accessible to non-experts; nearly 1,500 participants took part, and the solution is being released open source to support broader global use and improve education decisions.

UNESCO

Borhene Chakroun

Director for Policy and Lifelong Learning Systems