AstraZeneca Turns to AI Agent to Cut Drug Discovery Time

Artificial Intelligence & Machine Learning
,
Healthcare
,
Industry Specific

Owkin AI Will Analyze Trials, Forecasts, Other Rival Activity for Pharma Giant

Marianne Kolbasuk McGee (HealthInfoSec) •
May 13, 2026    

See Also: Reduce Cloud Risk in Healthcare with Security by Default

The use of Owkin’s technology – an AI agent pitched to the biopharma sector as expert in sifting through patient data – is a component of AstraZeneca’s ambition to improve research and development, clinical trial work and cut down the typical eight or nine years it takes for new drug discoveries.

AstraZeneca has told investors it aims to hit $80 billion in revenue by 2030, driven by the development of new drugs – and that AI will help the company achieve that goal.

Other parts of the company’s AI strategy involve AstraZeneca working with other partners, including a pilot program with the U.S. Food and Drug Administration to improve early-clinical trial efficiencies (see: US FDA Piloting Use of AI for Real-Time Clinical Trials.)

“We are continuously investing in frontier AI solutions with capabilities not only to accelerate better decisions across the drug development lifecycle, but also to derive novel and innovative insights,” said Jorge Reis-Filho, chief of AI for science innovation at AstraZeneca in a statement provided to ISMG.

Other major pharmaceutical makers betting on AI in a big way include Amgen, Bristol Myer Squibb, Novo Nordisk and Merck.

Paris-based Owkin’s three-year licensing agreement with AstraZeneca, disclosed Wednesday, calls for the end-to-end development of AI agents to run on Owkin’s K Pro agent, integrated within AstraZeneca’s IT infrastructure and decision workflows. The arrangement is the latest work between the two companies.

Owkin biopharma AI agents will help AstraZeneca monitor and analyze a variety of competitive data points from a number of sources, said Atanas Kamburov, vice president of solutions and platform delivery at Owkin. The firm on its website says K Pro exposes its reasoning to users and that it has guardrails to stymie hallucinations.

Use cases include analyzing ongoing and upcoming clinical trials, likely trial completion timelines and competitor strategies in specific disease areas, information gleaned from conference disclosures, news flows and other publicly available signals about pharmaceutical development activity, Kamburov said. Patient data will come from 800 hospitals.

Advanced functions include bio-statistical forecasting involving recruitment rates, trial centers, disease incidence and trial trajectories, he said. This type of analysis not only applies to what competitors are doing, but can also help AstraZeneca monitor internal efforts, he said.

“How should the parameters of a certain trial be tweaked so that the readout happens sooner,” is an example of the kinds of questions that can be answered faster with AI agents helping, Kamburov said.

AstraZeneca remains in control of access, authentication and data governance involving the use of the AI agents for competitive intelligence, he said. Owkin’s tools integrate with AstraZeneca’s single sign-on system, maintain audit trails and operate in a separate dedicated instance for AstraZeneca, he said.

Owkin previously worked with AstraZeneca to develop an AI solution to help pre-screen potential clinical trial patients with certain genetic mutations for breast cancer.

AstraZeneca invested about $14.2 billion to $14.6 billion on research and development in 2025, or roughly a quarter of its revenue, The Wall Street Journal reported.

A key part of the strategy to integrate AI agents into IT environments like AstraZeneca’s involves managing data across a vast, global enterprise, Kamburov said.

Security is baked into every layer of this architecture, he added. “We follow a zero trust approach, moving beyond simple perimeter defense to a model where every interaction is continuously verified,” he said.

Regardless of the data source, “we maintain strict ‘need-to-know’ access, ensuring that only authorized personnel can engage with specific information,” he said.

“In pharma, scientific integrity is paramount,” he said. “We have established ironclad guardrails to ensure that AI-generated outputs are not just innovative, but factually accurate and scientifically sound.”