ICON has announced a multi-year collaboration with Anthropic to bring Claude’s AI capabilities across the clinical trial lifecycle, expanding the company’s digital innovation and AI strategy through its secure agentic platform, Orbis.
The partnership will focus on four production capabilities inside Orbis. These include site intelligence and study planning, where ICON’s OneSearch and OnePlan tools will use frontier model reasoning to improve site selection and feasibility assessments. A second area, predictive intelligence, will use AI to detect enrollment risks, operational trends and study signals in real time across active trials.
The collaboration will also target protocol optimization, applying AI to protocol design and scenario modeling in an effort to reduce amendments and speed study startup. The fourth capability, ecosystem integration, will embed ICON’s clinical expertise into client systems so users can securely access ICON insights directly within Claude.
ICON said it will also roll out Claude across the organization in a role-based approach. Claude Code will be used for developers, Claude for knowledge teams, and Claude Science for scientific and clinical teams.
The company said the work with Anthropic supports three core elements of its AI strategy: strengthening the intelligence layer behind Orbis, boosting productivity through automation, and building domain-specific agents within clinical development workflows.
ICON Chief Executive Officer Barry Balfe said the company is determined to lead AI acceleration of clinical trials and believes the combination of ICON’s clinical development expertise and Anthropic’s frontier AI will help sponsors make better decisions earlier and accelerate delivery of medicines to patients.
Anthropic’s Pip White said AI can help address long-standing operational challenges in clinical development, noting that study enrollment is a major bottleneck that causes delays in many trials. The companies said the collaboration is aimed at improving execution and reducing friction across the trial process.