Okin launches the "Bio AI Center" in full swing... Collaborates with NVIDIA and Anthropic to kick off pharmaceutical innovation

A biotech pharmaceutical AI startup founded ten years ago, Owkin, has launched a series of autonomous AI agent products focused on clinical research and new drug development, officially initiating commercialization. The company announced that, based on collaborations with NVIDIA and Anthropic, it will build the next-generation AI ecosystem aimed at leveraging real patient data for precise analysis.

Owkin plans to promote its agent based on multimodal patient data collected from over 800 hospitals worldwide through the newly launched AI infrastructure “Agent Platform.” Unlike most previous biotech AI companies that relied on laboratory data, Owkin emphasizes that its AI learns primarily from clinical data of real patients, thereby enhancing data realism and reliability.

The company stated that these agents will be capable of performing functions such as ▲biomarker detection ▲interpretation of complex medical datasets ▲assisting clinical trial decision-making, which is expected to significantly accelerate new drug development and clinical success rates. The ability to use patient-centered data while protecting individual privacy is also highlighted as a technological differentiator.

Owkin CEO Thomas Klozel said, “So far, the costs and time required for new drug development have been prohibitively high. The release of this agent architecture will be a revolutionary turning point for the pharmaceutical industry. Utilizing specialized AI agents can efficiently perform various analyses and clinical trial designs, and we have designed it so that any researcher can easily access it on existing platforms.”

This platform is not just a tool but aims to become a “smart biological AI hub” that facilitates real-time data flow between hospitals, research institutes, and pharmaceutical companies. Klozel described it as “biological artificial superintelligence,” emphasizing, “Over time, AI will be able to autonomously learn and reason across the entire field of biology, enabling the derivation of entirely new treatment and diagnostic methods.”

The first agent most focused on technically is the “Pathology Explorer.” It is provided through Anthropic’s healthcare professional platform Claude for Healthcare Life Sciences, developed based on Owkin’s pathology data interpretation models. This agent can identify cell types, locations, and biomarkers solely from digital pathology images, with an accuracy 23.7% higher than existing models. At the same time, it uses only one-fifth of the parameters of the previous model, significantly reducing computational costs and processing time.

Owkin will provide these agents via API, compatible with common model context protocols in the healthcare industry, allowing easy integration into existing hospital or pharmaceutical systems. These agents will continuously improve through ongoing patient data refinement and laboratory-based testing.

Meanwhile, the collaboration with NVIDIA will focus on enhancing Owkin’s foundational model “OwkinZero.” The plan is to improve model performance in biological reasoning and scalability based on NVIDIA’s NeMo framework and high-speed image processing tool cuCIM. Klozel stated, “The combination of NVIDIA’s AI ecosystem and Owkin’s capabilities will give rise to an AI system capable of understanding and interpreting biological backgrounds at levels that no previous model has achieved.”

Owkin’s efforts are seen as a signal of fundamentally transforming the existing paradigm of the pharmaceutical industry. The shift from a lab-centered approach to a patient-centered one, and from closed analysis to open AI agents, is attracting significant attention for the potential to bring about innovative clinical outcomes.