Owkin Licenses Its K Pro 'AI Scientist' and Multimodal Patient Data to Boehringer Ingelheim

Drug discovery AI just signed one of its most interesting deals of the year. On September 2, 2026, Owkin — the Paris- and New York-based “agentic AI” company pursuing what it calls Biological Artificial Superintelligence — announced a license agreement with Boehringer Ingelheim covering K Pro, Owkin’s AI Scientist, together with multimodal patient data. The goal: accelerate the discovery process across multiple indications in oncology and immunology, two of Boehringer’s core therapeutic areas.

The announcement, made jointly from New York and Paris, is the third major pharmaceutical license Owkin has landed for K Pro in 2026, following a multi-year co-development deal with Sanofi announced June 5 and a separate multi-year K Pro license with AstraZeneca in May. Taken together, they mark a shift in how large pharma is buying AI: not one-off consulting projects, but standing software licenses wrapped around proprietary data networks.

What Boehringer is actually getting

Under the new agreement, Owkin will license multimodal oncology data to Boehringer Ingelheim and will generate new multimodal data in immunology. That second clause matters — this is not just a data dump, but a commitment to produce new immunology datasets in collaboration with Owkin’s global patient data network, which the company builds by licensing, sourcing, and generating multimodal patient data with academic and clinical partners.

Boehringer’s research teams will access and analyze the data through K Pro, which Owkin describes as a single environment to interrogate data and run reproducible analyses. K Pro’s reasoning capabilities support both human-led analysis and self-driven campaigns for hypothesis generation, testing, and prioritization.

On the product side, K Pro is an AI agent for biology built around a single intelligent orchestrator that dynamically selects and combines specialized AI skills across drug discovery and development. Documented functions include ranking therapeutic targets and gene candidates using statistical evidence from TCGA and Owkin’s MOSAIC dataset, evaluating target tractability and safety profiles, characterizing patient subgroups from multi-omics profiles and clinical outcomes, validating biomarkers across bulk RNA-seq, single-cell, spatial, and clinical data, and generating publication-ready statistical reports covering target assessment, biomarker prioritization, and patent landscape analysis.

The MOSAIC connection

The deal builds on a pilot the two companies ran in 2025, in which Owkin delivered deep spatial insights into the tumor microenvironment of a gene target through its MOSAIC dataset. That pilot evidently convinced Boehringer that the dataset-plus-agent bundle was worth a full license.

MOSAIC — the Multi-Omics Spatial Atlas in Cancer — is described by the project as the world’s largest spatial multiomics dataset in oncology. According to the project’s own figures, it currently spans 11 therapy areas, six data modalities, and roughly 100 clinical variables per patient, totaling 2,725 patients: spatial omics data for 2,371 of them and single-cell omics for 2,168. The six modalities are spatial transcriptomics, single-cell RNA sequencing, bulk RNA sequencing, whole exome sequencing, digitized H&E histology, and clinical data for every patient.

The project’s founding partners read like a who’s-who of European and American oncology centers: Owkin, Gustave Roussy, Charité, CHUV, Universitätsklinikum Erlangen, and the University of Pittsburgh. Owkin’s target scale for MOSAIC is 7,000 patients across seven cancer indications — non-small cell lung cancer, ovarian, bladder, mesothelioma, glioblastoma, breast cancer, and diffuse large B-cell lymphoma — which would make it more than 100 times larger than any existing spatial omics dataset at launch. By mapping interactions between cells within the tumor microenvironment, MOSAIC aims to uncover new disease biology, identify more precise patient subtypes, and discover biomarkers and drug targets.

Why Boehringer wants this

Boehringer Ingelheim is not a company that needs saving — its clinical pipeline, as of May 2026, lists more than 80 active preclinical and clinical projects, and half of that pipeline is anchored in external collaborations. Its oncology roster includes zongertinib, a HER2 tyrosine kinase inhibitor, and obrixtamig, a DLL3/CD3 T-cell engager; immunology programs target inflammatory bowel disease, systemic lupus erythematosus, and systemic sclerosis.

But spatial biology is where much of the frontier target discovery is happening, and building a MOSAIC-class dataset in-house would take years and hundreds of millions of dollars. Licensing one — plus the agentic layer that reasons over it — is the faster path. The 2025 pilot focused on spatial insights into a single gene target’s tumor microenvironment; the new deal scales that across multiple indications.

The business model shift

What makes this deal notable beyond pharma-AI cooperation is the structure. Owkin spent much of its early life as a services-and-collaboration company — its 2021 unicorn round was a $180 million investment from Sanofi tied to four collaborative projects. The 2026 wave of deals is different: K Pro is offered in free and professional tiers, with the professional tier adding proprietary data access, data upload, and deployment on client infrastructure. Owkin is becoming a product company, and big pharma is buying seats.

The company has raised roughly $334 million across eight funding rounds and was valued above $1 billion at its Sanofi round. It has also been reorganizing around its core: in March 2026 its diagnostics division spun out as Waiv with $33 million to commercialize AI precision testing in oncology, and its Bioptimus affiliate continues building foundation models for biology separately. Owkin itself is doubling down on K Pro as the product.

Owkin’s co-founder and CEO Thomas Clozel, MD, framed the deal in sweeping terms: “This agreement between Owkin and Boehringer Ingelheim shows how cutting-edge AI has the potential to enable data-driven pharmaceutical research. We believe the next decade of drug development will be defined by access to deep multimodal patient data and to AI Scientists — like K Pro — capable of reasoning over that data.”

What to watch

The obvious question is whether AI agents reasoning over spatial multiomics data can actually move discovery timelines — a claim every AI-drug-discovery company makes and few have proven at scale. The Sanofi and AstraZeneca deals will be the comparable cases: all three licenses were announced within four months of each other, and if K Pro campaigns surface validated targets in that window, the “AI Scientist” pitch stops being marketing and becomes a procurement category.

Second, watch the immunology data generation. Owkin’s strength to date has been oncology, where MOSAIC gives it a structural moat. Immunology is newer territory for its data network, and the Boehringer deal effectively funds an expansion of that moat into a second therapeutic area — with the data presumably staying in Owkin’s network for future licenses.

Third, patient data governance. Multimodal patient data licensed across borders to a German drugmaker is exactly the kind of arrangement that European regulators watch closely. Owkin’s federated, partner-collaborative model was designed for this, but as these deals multiply, so does scrutiny.

For now, the scorecard is simple: three top-20 pharma companies have now licensed the same AI Scientist. That is the strongest signal yet that agentic AI has found its first durable enterprise market — not in chatbots, but in biology.