French AI startup Bioptimus, which intends to build a large language model (LLM) for biotechnology, has raised a $35m seed round just a few months after it was created, confirming previous reporting by Sifted.
The company was created by top executives from French unicorn Owkin, which uses AI to understand and improve the effect of drugs and treatments on different patient groups.
Bioptimus, which was incubated at Owkin, was founded by the unicorn’s current chief R&D officer and ex-Google researcher Jean-Philippe Vert. Vert will not have an operational role at Bioptimus and will keep his role at Owkin.
The seed round was led by French life sciences VC Sofinnova Partners and public bank Bpifrance Large Ventures. French VCs Frst and Cathay Innovation, US investor Headline, Owkin and billionaire entrepreneur Xavier Niel also participated in the round.
What will Bioptimus do?
LLMs — the same technology that powers ChatGPT — which can generate human language, images, sounds or software, are AI models trained on huge datasets.
Bioptimus wants to create a model that will be trained on data that’s necessary to understand biological processes on various levels. It is called a “multiscale” model because it will account for many different scales of biology — including organs, tissues, cells, molecules and atoms.
Until now, AI models have not been trained on data that represents these different levels of biology, meaning that they cannot represent the full complexity of organisms. Simulations of biological processes, for example an organism’s reaction to a drug, are therefore not as accurate as they could be.
“By harnessing the power of foundation models and advanced algorithms trained on massive amounts of biological and multimodal data across scales, we aim to capture the laws of biology that have hitherto remained too complex to be properly understood,” says Vert.
He adds that he believes Bioptimus has a significant competitive advantage, as it will be able to train its model using the troves of high-quality patient data that Owkin holds, in a field where access to information is heavily regulated.
While a €35m seed round seems substantial at such an early stage, it has become the norm for startups building LLMs — which require large amounts of expensive compute power and top talent.
Last year, French AI startup Mistral raised €105m seed round just a few weeks after launching, while Poolside, which is building an LLM that could write code, raised a first round of $126m.
