Case Study: Capgemini achieves 100% developer adoption with Mistral AI’s private software engineering assistant

Capgemini boosts developer adoption from 30% to 100% with Mistral AI

Capgemini, a global system integrator, faced the challenge of leveraging generative AI for its regulated clients in sectors like aerospace and defense, who required self-hosted solutions due to strict data sovereignty requirements. Their initial open-source coding assistant had low developer adoption and insufficient accuracy, preventing widespread use. To address this, Capgemini turned to vendor Mistral AI and implemented its Codestral model within their private software engineering assistant.

By deploying Mistral's Codestral in a self-hosted environment, Capgemini achieved a dramatic improvement. Code completion accuracy jumped from 50% to 90%, leading to a developer adoption rate increase from 30% to 100% across all client projects. This solution delivered systematic productivity gains and quality enhancements, with over 50 client projects reporting improvements in less than six months, all while meeting strict data privacy and compliance mandates.

Capgemini