Bioptimus is a newly created Paris-based generative AI startup. The most engaging fact about Bioptimus is its plan to implement all the known artificial intelligence prototypes we learned over time. The primary focus of these AI models will be on bio.
It makes reason to launch a startup with a sole focus on biology because training data access is more difficult in this area. Bioptimus must cope with private and sensitive clinical data, which presents various data issues, while web crawling is gradually being replaced by licensing agreements with content creators by OpenAI,
Bioptimus will also require a significant investment because, like other AI startups, it will employ skilled researchers and train its models on expensive GPUs. For this reason, Sofinnova Partners is leading a $35 million seed round that the business is raising. The funding parties include Bpifrance’s Large Venture fund, First, Cathay Innovation, Headline, Hummingbird, NJF Capital, Owkin, Top Harvest Capital, and Xavier Niel.
Bioptimus is not an overnight sensation. Jean-Philippe Vert will lead the business as executive chairman and co-founder, serving in a non-operational capacity. He is the chief research and development officer at Owkin. Owkin is a French biotech unicorn that uses AI to find new medications and enhance diagnostics.
Since Rodolphe Jenatton was a top researcher at Google, he has more extraordinary expertise in artificial intelligence than most CTOs. A number of the co-founders were formerly Google DeepMind researchers.
Owkin has partnered with world-class university hospitals to gather multimodal patient data as part of his work for top biopharma. Bioptimus will use this unique dataset to train its core model.
A moonshot scheme from Owkin
Bioptimus may be viewed as a moonshot venture or Owkin’s spin-off business. However, why didn’t Owkin choose to develop an internal foundational model? Since developing innovative AI models is challenging, establishing a distinct organization made more sense.
Vert told TechCrunch that Although Owkin still needs to build biology [foundational models] on its roadmap, it is eager to work with and support organizations such as Bioptimus. It takes significant resources in data volume, processing capacity, and various data modalities—which are more straightforward to unlock as a distinct entity—to train massive foundational models. Bioptimus is more suited to accomplish this since it is a “pure player” in core models.
Additionally, the company and Amazon Web Services have partnered. Amazon’s data centers will likely be used to train the company’s model. Bioptimus can now focus on refining its AI model with sufficient funding and exploring the potential applications it may provide to the biotech research community.
According to Vert, the AI we develop will eventually help with precision medicine, disease detection, and the synthesis of unknown biomolecules for application in the environment or medical field.
