Owkin announced an agreement On September 11, 2026 to grant a license to K Pro, the AI scientist it has employed for its biopharmaceutical research along with patient multimodal data, Servier an international independent pharmaceutical group that is governed by foundations, in order to further Servier’s research into oncology.
The License covers
Owkin’s MOSAIC data will be made available to Servier research teams for analysis within K Pro. In an announcement made in New York and Paris on the same day, the company framed this license as the latest chapter of their collaboration. Their previous joint work involved patient subgroups identification in cancer.
Owkin describes itself as a company that uses agentic AI. K Pro, according to Owkin, combines patient data from multiple sources with biological AI agents. This allows for decision support at every step in the value chain of pharmaceuticals. The platform, according to Owkin, delivers target discovery, biomarker detection, subgroup optimization and clinical trial designs. It is also trained and validated using real-world data from patients and Owkin’s wet lab. Owkin says the system is based on biological tools, a decade’s worth of experience gained from pharmaceutical research and partnerships with AI companies.
Thomas ClozelOwkin CEO and co-founder, Mr. “Our AI scientist is changing the way biopharma works. By pairing K Pro with Owkin’s multimodal patient data, Servier’s scientists can investigate the biology of hard-to-treat cancers and reach answers that manual analysis alone cannot deliver.”
Claude Bertrand (executive vice-president of Research and Development at Servier) said that K Pro’s access to AI-driven analyses and high-quality data from patients in one location, allows his team to generate hypotheses and test them with more confidence. Servier is looking forward to implementing the capabilities in its research on cancer.
K Pro Documented capabilities
Owkin’s product documentation The company describes K Pro’s AI as an agent who answers questions in natural language using multimodal patient data. It returns answers that are “rigorous and publication ready”, according to the company. Documentation states that K Pro ranks gene targets based on statistical evidence from TCGA datasets and MOSAIC databases. Additionally, it evaluates target safety profiles, therapeutic potential, and target tractability using curated database, as well as identifies different patient groups based multi-omics profiles.
According to the documentation, K Pro can test biomarkers, hypotheses, bulk RNA-seq data, individual-cell data, spatial data and clinical data. It also integrates genomics, transcriptomic data and immune data. This software also produces statistical reports that cover gene and targets assessment, prioritization of biomarkers, and analysis of the patent landscape.
Owkin claims that a single intelligent Orchestrator is at the heart of the platform. It dynamically selects and combines specialized AI abilities so users can work with a single agent instead of juggling several tools. According to Owkin, the agent learns continuously from real-world information, user feedback and clinical validation.
K Pro has two tiers according to documentation. A free tier is available for demonstration and exploration of 19 public datasets. The professional tier adds Owkin’s proprietary datasets along with the MOSAIC spatial multiomics dataset. It also includes data uploading, increased usage limits and deployment as SaaS, or at the customer’s location.
MOSAIC Dataset
MOSAIC is an Owkin initiative which provides the patient data that’s covered under this license. describes its dataset As the largest multiomics spatial dataset for oncology in the world. According to the initiative, mapping cell interactions within a tumor’s microenvironment will help uncover a new understanding of disease biology. It can also identify subtypes and biomarkers that may enable more personalized treatment.
MOSAIC reports total 2,725 patients in 11 different therapy areas. It also includes six data modes and 100 variables clinical per patient. This site provides spatial omics information for 2,371 individuals and single-cell data for 2,168. Its per-modality counts include 2,666 digitized H&E samples, 2,266 whole exome sequencing samples, 2,114 bulk RNA-seq samples, 2,371 spatial transcriptomics samples, and 2,168 single-cell RNA-seq samples, in addition to clinical data for every patient.
According to the indications, there are 532 cases of non-small cell cancer (NSCLC), 472 cases of ovarian and breast cancer, 306 cases of glioblastoma in women, 269 cases in men, 194 in diffuse large cell lymphoma in women, 179 in pancreatic, 159 for HNSCC in men, 100 for colorectal carcinoma, 96 with mesothelioma in women, 23 in gastric. MOSAIC’s founding partners are Owkin, Gustave Roussy, Charité, CHUV, Universitätsklinikum Erlangen, and the University of Pittsburgh.
The Partnership for October 2023
A license is a way to extend the relationship between two businesses formalized on October 17, 2023The companies announced at the time that they would initially take on two challenges in translational medicine and digital pathology, with Owkin applying machine learning to Servier’s clinical data to identify patient populations most likely to benefit from Servier’s new therapies. They said that the first two areas they were going tackle would be translational pathology and digital medicine. Owkin was to apply machine-learning to Servier’s data in order to find the most suitable patient groups for Servier’s therapies.
According to the announcement, the project’s first goal was to identify cancer types, tumor environments, and subgroups of de-identified patients that would respond optimally to an asset from Servier. The companies analyzed tumor evolution in order to determine drug combinations which could target additional immune checkpoints, or tumor-cell-intrinsic mechanism. Digital pathology, as an approach to screening that could be used for tissue biomarkers and broader analyses was the focus of the second project. The collaboration was run by a multidisciplinary team consisting of clinicians, scientists and computational medicine specialists from both companies. Bertrand, at that time, held the positions of chief scientific officer and executive vice president for research and development.
