Customer Success Story : Retorik ai
We were looking for a cloud provider aligned with our objective of hosting our AI platform in a sovereign environment. Thanks to Scaleway, we have found a trustworthy partner that not only helped us fulfill this goal but also helped us limit our CO2 emissions, while keeping our cloud budget under control.As such, Scaleway enabled us to reduce our costs by 15%.
About the company
Retorik AI is a generative artificial intelligence platform developed by Davi that addresses productivity challenges in high-demand professions through verbal and non-verbal capabilities. The platform generates intelligent robots capable of understanding human language and emotions, enabling them to interpret and respond appropriately to users, taking into account the content and emotional tone of exchanges. These robots provide precise, contextual answers in a given field of expertise, guaranteeing the user reliable, tailored information. What's more, they are endowed with an appearance, social attitude and personality that make them more human and interactive, offering a natural and engaging user experience.
Challenges
Retorik ai goes beyond traditional language models to optimize workflows and improve efficiency, equipping digital employees with professional, behavioral, social and emotional skills.
Until now, they hosted their data in France, but with a foreign provider. As their need for AI computing resources grew, they realized that they couldn't develop sovereign and ethical AI models while being exposed to U.S. cloud law. What's more, they couldn't guarantee total security for their end-customer data.
Retorik AI therefore needed to guarantee the sovereignty of its AI models and the security of its customers' data, while reducing its cloud costsThey also wanted to avoid being subject to out of Europe specific’s jurisdiction, and address their environmental concerns, including the reduction of CO2 emissions.
Results
After developing an infrastructure on Kubernetes, we logically built their installation on Kapsule; Scaleway's managed Kubernetes product. This enabled them to design their R&D and production stacks for their internally hosted LLMs, thanks to the availability of H100 nodes in Kapsule. The Scaleway teams also provided them with invaluable advice on correctly sizing their environments and ensuring the stability of their K8s clusters. Thanks to Scaleway, they were able to save 15% on cloud costs.
Future Plans
Retorik ai aims to significantly improve the performance of the models they serve. In addition, they want to reduce the maintenance efforts of their teams, while guaranteeing the security of their customers' data. To this end, they are currently part of the private beta version of Generative APIs.