Researchers at the health tech startup Nabla were exploring the potential applications and constraints of OpenAI’s renowned chatbot ChatGPT two years before its public release. During testing in 2020, they engaged with the app, prompting the significant language model to respond, “I think you should,” to a query about suicide, highlighting the risks associated with utilizing AI for health-related inquiries. Alex Lebrun, one of Nabla’s co-founders, emphasized the lack of trust in using large language models (LLMs) for providing advice, particularly in patient-facing materials.
Despite this skepticism, Nabla found an opportunity to leverage GPT-3 and later GPT-4 in healthcare for data analysis and interpretation. The team at Nabla, based in Paris, developed a tool, the Copilot, to streamline doctors’ workflow by automating the generation of detailed medical notes required for billing and documentation, thus saving time on administrative tasks.
However, the proprietary nature of OpenAI’s software posed challenges as it limited transparency and control due to the unavailability of the source code for review and modification. Consequently, Nabla opted to reduce its reliance on OpenAI and focus on refining existing open source models like Facebook’s LLaMA 2 and Mistral’s offerings. This strategic shift was reinforced by a recent \(24 million Series B funding round led by Cathay Innovation and Zebox Ventures, valuing Nabla at approximately \)180 million post-money.
Nabla’s clinical note-taking tool, launched in March 2023, has gained traction among over 20,000 users, predominantly comprising primary care physicians, psychologists, and various medical specialists. The company secured a significant partnership with the Permanente Medical Group, Germany’s largest healthcare provider, employing 10,000 physicians.
In the competitive landscape of medical note-taking solutions, Nabla faces rivals like Abridge and Microsoft’s Nuance Communications acquisition. Despite the competition, Nabla’s strategy of testing and optimizing different models to enhance flexibility and cost-effectiveness sets them apart. By combining open source and commercial models judiciously, Nabla aims to maintain a competitive edge in the evolving healthcare AI market.
The collaborative efforts of Alex Lebrun and Yann Lecun, both pioneers in AI research, underscore the importance of open source principles in advancing AI technologies. Lebrun’s journey from Facebook AI Research to founding Nabla reflects a commitment to leveraging AI for transformative healthcare solutions. Nabla’s unique approach of training AI models based on patient interactions in digital clinics sets it apart in the industry, emphasizing data privacy and patient confidentiality.
Looking ahead, Nabla plans to expand its services to streamline insurance billing processes and support multiple languages, catering to a diverse global audience. With a customer-centric approach and a focus on data protection, Nabla aims to revolutionize healthcare practices while upholding ethical standards and user privacy.
