Clear.bio, a health-tech startup founded in 2019 that offers a 12-week digital nutritional intervention to help people reverse Type 2 diabetes, used Google AI to automate how it identifies and scores healthcare practices as potential referral partners, according to a Google Cloud customer case study.
As Clear.bio grew across the Netherlands and expanded into Germany and France, it needed its sales and marketing effort to focus on the healthcare practitioners most likely to become valuable partners. Identifying those practices by hand was time-consuming and inefficient, and the company also wanted to automate its whole data prediction workflow so that expansion would not be held back by manual work.
In an AI Sprint with Google for Startups, Clear.bio used Gemini to identify practices that fit its growth strategy, analysing factors such as technology adoption, patient demographics and treatment patterns. It trained and deployed an AutoML machine learning model on the Gemini platform that makes the final prediction and assigns each practice a 'Best_Class' score. BigQuery serves as the data warehouse, loading, extracting and preprocessing the training data, and Cloud Run functions automate the pipeline, triggering the process whenever new CSV files are uploaded and managing the end-to-end data flow. Engineering, product, scientific, marketing and partner management teams worked on the project together.
Within the first few months, the conversion rate of high-value healthcare groups interested in partnerships rose by 30%. The sales team now concentrates on the top 20% of practices most likely to generate referrals, which cut the time spent on outreach by almost 20%, and the model's predictions are 92-94% accurate. "It saves our team time, reduces guesswork, and helps us reach out to the most promising practices more efficiently," said CTO Parul Vats. Gijs van der Spek, Account Manager for Regional Healthcare Organisations, said the tool "consistently boosts engagement and significantly increases my chances of closing new contracts."
Separately, Clear.bio uses AI to translate continuous glucose data into personalised nutritional insights, combined with support from human dietitians, and 69% of its patients saw a clinically meaningful drop in HbA1c. The company plans to scale the predictive model to more teams, win more clients and expand to new countries. "We are excited that the model has already shown great results over the last six months," said co-founder and CSO Madelon Bracke.