OpenAI Advances Computer-Use Agents for Healthcare Teams

OpenAI is positioning computer-use agents as a way to reduce manual information movement in healthcare workflows. The agents perform multi-step tasks by clicking, typing, and moving between applications, allowing AI to handle repetitive work in software teams already use. The announcement reports new evaluation results for model performance, safety, and cost in clinical and administrative settings, and describes collaborations with Abridge, Ambience, Cigna Healthcare, and CVS Health. Access is available today through the OpenAI API, ChatGPT for Healthcare, and ChatGPT for Clinicians.

To show the workflow, OpenAI built a demo using synthetic patient data: a patient messages a clinician about lower back pain, and a computer-use agent reviews visit notes, physical therapy records, and prior imaging in the EHR. The agent drafts an MRI order, an encounter note, and a patient reply for clinician review, then uses the same records to prepare a prior authorization request in the insurer’s portal for an authorization specialist.

OpenAI evaluated GPT-6 Astra on clinician EHR inbox tasks shaped by physician input. Reviewers assessed whether the model’s actions, notes, orders, and patient messages would be helpful and clinically safe. Overall, 99% of 325 reviewed tasks were rated safe, and physicians selected “very helpful” or “extremely helpful” in 93% of 646 assessments.

For administrative work, OpenAI evaluated computer-use agents on HealthAdminBench, an independent Stanford benchmark for common administrative workflows such as prior authorizations, appeals, and medical equipment orders. GPT-6 Astra earned 94.3% of available rubric points, while GPT-6 Luna earned 91.8% despite costing 98% less at 8.5 cents per task. Both models were evaluated at maximum reasoning effort.

The post also describes partner efforts and production results. CVS Health uses computer-use agents to automate repetitive work across clinical systems; one agent cleared a months-long document-sending backlog over a weekend, and another handles system steps for tens of thousands of consults each day. CVS Health is open sourcing its Agent Kit, built with OpenAI’s Agents SDK, and keeps production data within its enterprise-controlled infrastructure, with no production data used to train foundation models.

Abridge is developing computer-use agents to help clinicians carry out care plans, including a planned inpatient use case that adds EHR information and nurse/clinician context. A clinician can ask the Abridge Assistant to help resolve discharge barriers, with agents working concurrently on medication access and home-health coordination while the clinician stays focused on the patient. Cigna Healthcare is using OpenAI frontier models alongside its clinical expertise to bring together information from systems that case managers previously navigated manually across more than 10 separate clinical and administrative systems. Ambience built additional controls so agents stop and ask clinicians when records are ambiguous, wait for approval before submitting, and provide step-by-step audit logs.

On safety and data controls, ChatGPT for Healthcare provides role-based access controls and audit logs. Organizations can add review and approval steps, configure which applications and environments agents can use, and limit access to records and actions needed for each task. Organization data is not used to train OpenAI foundation models. For HIPAA-compliant workflows involving protected health information, organizations must use eligible services covered by their Business Associate Agreement with OpenAI and apply the required security and retention settings.

Healthcare organizations, clinicians, and developers can get started now: organizations can contact OpenAI’s sales team or enable computer use in ChatGPT Work and Codex, U.S. clinicians can sign up for a free ChatGPT for Clinicians account, and developers can build agents with the OpenAI API and the computer use guide.

Advancing computer-use agents for healthcare teams

View Original