ICON has entered into a multi-year collaboration with Anthropic to integrate Claude’s artificial intelligence capabilities across the clinical trial lifecycle.
The partnership will build on Orbis, ICON’s secure and governed agentic AI platform, and support the contract research organization’s broader digital innovation and AI strategy.
ICON will work with Anthropic’s Life Sciences research team to explore and deploy AI technologies intended to improve clinical development planning, study execution and operational decision-making.
The companies will initially focus on developing four production capabilities within Orbis: site intelligence and study planning, predictive trial intelligence, protocol optimization and integration with client technology systems.
ICON plans to incorporate Claude’s reasoning capabilities into its OneSearch and OnePlan tools to improve the selection of clinical trial sites and the assessment of study feasibility.
Site selection is an important part of clinical development because sponsors must identify hospitals, medical practices and research centers capable of recruiting eligible patients and following the study protocol.
A poorly selected site may enroll too few participants, take longer than expected to begin operations or lack the necessary staff and patient population. These issues can delay a trial and increase costs.
By applying AI to ICON’s trial data and operational expertise, the companies aim to help study teams evaluate potential sites more efficiently and make better planning decisions before enrollment begins.
The second planned capability, Orbis predictive intelligence, will be designed to identify enrollment risks, operational trends and other signals across active studies in real time.
Clinical trials generate information from numerous sources, including sites, patient enrollment systems, laboratory results, monitoring activities and operational reports. Detecting emerging problems early can allow study teams to adjust resources, contact underperforming sites or revise recruitment strategies before a delay becomes more significant.
ICON and Anthropic will also apply AI to clinical trial protocol development and scenario modeling.
A protocol defines how a study will be conducted, including which patients can participate, what treatments they receive, what information is collected and how outcomes are evaluated.
Protocols can become difficult to execute when they include overly complex eligibility criteria, excessive site requirements or procedures that place significant burdens on patients and investigators.
Changes made after a trial begins are known as protocol amendments. These changes may require additional regulatory submissions, site training, patient documentation and system updates, potentially increasing costs and delaying the study.
ICON plans to use Claude to model potential protocol scenarios, identify operational issues and reduce the likelihood of avoidable amendments. The companies believe this could accelerate study startup and improve the precision with which trials are designed.
The fourth planned capability will allow ICON clients to access the company’s clinical insights directly within Claude through secure integrations.
This approach is intended to bring ICON’s expertise and trial intelligence into the systems that pharmaceutical, biotechnology and medical device companies already use.
ICON did not disclose which client data or workflows will be available through these integrations. The company emphasized that Orbis is designed as a secure and governed platform, which is particularly important because clinical development can involve confidential scientific information, commercially sensitive study plans and protected patient data.
Beyond the four production capabilities, ICON will deploy Claude tools across its internal organization through a role-based rollout.
Developers will receive access to Claude Code, while knowledge-oriented teams will use Claude for research, analysis and other professional workflows. Scientific and clinical personnel will use Claude Science for activities requiring greater specialization in life sciences information.
The deployment is intended to automate high-volume, repeatable work and allow employees to devote more time to activities requiring clinical judgment, scientific expertise and direct collaboration with customers.
ICON said the collaboration supports three central areas of its AI strategy.
First, Anthropic’s models will strengthen the intelligence layer supporting Orbis. This layer connects ICON’s proprietary data, clinical experience and AI capabilities across different phases of a trial.
Second, Claude’s deployment across developer and knowledge teams is expected to improve productivity by reducing manual work.
Third, ICON and Anthropic plan to develop domain-specific AI agents embedded directly within clinical development processes.
An AI agent can perform a sequence of tasks toward a defined objective rather than responding only to an individual prompt. In clinical trials, an agent could potentially monitor operational information, identify an enrollment risk, gather supporting evidence and recommend a response for review by a study team.
The companies did not say whether any agents will make independent clinical or operational decisions. Human oversight, data governance and regulatory compliance are expected to remain important as the capabilities move into production.
ICON provides outsourced drug and medical device development services to pharmaceutical, biotechnology and healthcare companies. Its work spans study planning, patient recruitment, clinical operations, data management, regulatory support and other functions required to evaluate experimental treatments.
Anthropic develops the Claude family of AI models and has been expanding its focus on specialized applications for scientific research, healthcare and life sciences.
The collaboration reflects growing interest among clinical research organizations in using generative and agentic AI to address operational bottlenecks.
Clinical trials can be delayed even when the underlying scientific program is promising. Challenges involving site activation, patient enrollment, protocol complexity, data review and coordination across global research centers can extend development timelines.
Anthropic cited study enrollment as one of the largest operational problems in clinical development, noting that enrollment difficulties contribute to delays in as many as 80% of trials.
ICON believes combining Claude’s reasoning capabilities with its clinical trial data and global delivery infrastructure could help sponsors identify risks sooner, make faster decisions and operate studies more efficiently.
The companies did not disclose the financial terms of the collaboration or provide a timeline for launching the four planned production capabilities.
KEY QUOTES:
“We are determined to lead AI acceleration of clinical trials. Working with Anthropic gives us direct access to frontier AI capability and the people who build it.”
“By combining ICON’s clinical development expertise, global delivery model and proprietary trial intelligence with Anthropic’s frontier AI capabilities, we believe we can help sponsors make better decisions earlier, execute studies with greater precision and ultimately accelerate the delivery of medicines to patients.”
Barry Balfe, Chief Executive Officer of ICON
“AI holds real potential to address long-standing challenges in clinical development and make a measurable difference for patients.”
“Far too often, the barrier to getting medicines to patients faster is operational, not scientific. Take study enrollment as an example. It’s one of the biggest bottlenecks in clinical development and causes delays in up to 80% of trials.”
“This is exactly the kind of operational challenge that ICON’s trial-delivery expertise, paired with Claude, is well suited to address.”
Pip White, Head of Ireland, UK and Northern Europe at Anthropic
