Insights, Success

About Hexis

Hexis Performance was founded in 2018 to simplify and unlock elite-level sport nutrition for everyone. Hexis leverages AI to personalise best-practice nutrition and deliver it at scale. It is built by elite sport practitioners with international academics in data science, behavioural science, nutrition science, performance science and biostatistics.

We interviewed the CEO of Hexis Performance, David Dunne, about how EvoML solves their challenges in building and deploying AI into their product to scale personalised nutrition.

1.Business Challenge

Nutrition is a complex science. To provide personalised nutrition recommendations, a deep understanding of multiple areas, such as metabolism, biochemistry, physiology, and performance science are required. In fact, there are very few expert nutritionists who are qualified to develop elite-level nutrition plans.

In addition to the complex science, there are many additional factors to consider when developing an individually tailored nutrition solution for optimal performance. As a result, expert nutritionists can only develop several nutrition plans, at a push, in a day. Even then they are limited by time as the resources that can be developed to support these plans, e.g. recipe and shopping solutions.

Hexis needed to leverage cutting-edge technologies like AI to convert this complex nutrition science and individual requirements into simple and powerful nutrition practices, enabling more people to access elite-level personalised nutrition in their day-to-day lives.

2.Build and Scale AI

At the moment, Hexis has a relatively small team of in-house data scientists. Hexis needed an AI solution that can boost their small team’s productivity to build efficient models that can be deployed instantly and scale to thousands of users. Nutrition is a hard science, Hexis wanted a robust platform that can quickly recommend personalised nutrition plans for hundreds of thousands of users simultaneously. They also needed a white-box process which can explain the causality between nutrition and performance. These requirements led Hexis to partner with TurinTech.

EvoML is cloud-agnostic and easy to set up. Hexis got EvoML up and running on their existing infrastructure on Azure in less than an hour. Using EvoML has been a very easy and smooth process. With a few clicks, Hexis analysed a vast amount of data and generated machine learning models that can provide better nutrition recommendations. Additionally, EvoML continuously monitors user interactions with the Hexis App, and optimises the models with latest user behaviour changes, so users can get dynamic nutrition for maximum performance.

David: “The EvoML platform is very robust. And we find the intuitive interface and explainability feature very helpful. It has empowered us to quickly and easily create ML models, while allowing us to understand the reasoning behind why the model made the decision it did. “


Thanks to EvoML, Hexis’s two-person data science team can work like a ten-person team. The intuitive interface enables them to use EvoML without extra training. In addition, efficient models embedded in the Hexis app run fast on mobiles, providing best nutrition recommendation immediately for optimal user experience.

In the future, Hexis plans to leverage EvoML to generate hundreds of models that will further empower its capabilities to generate better nutrition plans, more personalised training sessions and recommend better recipes to users.

David: “It would’ve taken months for our small team of data scientist to build the model and ensure it is ok to deploy and be used in production. EvoML has made this process so much simpler and faster. “