OpenAI Brings Computer-Use Agents to Healthcare, Automating Clinical Workflows With Human Oversight
November 16, 2026
OpenAI announced computer-use agents for healthcare, available through ChatGPT for Healthcare, ChatGPT for Clinicians and the OpenAI API, that can take multistep actions in existing software for tasks such as authorizations, referrals and discharge planning.
The agents are designed to do more than draft and summarize: they can move information and complete workflows across electronic health records, payer portals and other disconnected applications, including systems with limited or no API access, while leaving people in control.
In a demonstration using synthetic patient data, an agent reviewed records and drafted an MRI order, visit note and patient reply for clinician review, then prepared an insurer authorization request for an authorization specialist.
In OpenAI’s evaluation of GPT-6 Astra on electronic health record inbox tasks, reviewers rated 99% of 325 tasks clinically safe, and physicians found the work very or extremely helpful in 93% of 646 assessments.
OpenAI also improved clinical search in ChatGPT for Healthcare and ChatGPT for Clinicians so it can find, assess and cite medical evidence; physicians rated the update higher across all eight measured dimensions in more than 1,300 blind-review evaluations.
On Stanford researchers’ HealthAdminBench, GPT-6 Astra scored 94.3% of available rubric points and GPT-6 Luna scored 91.8% across 135 tasks and 1,698 subtasks; Luna cost 8.5 cents per task, 98% less than Astra.
More than 20 healthcare plugins, including offerings from Aiva, Availity Extend, Luma Health, Oracle HDI, Redox and UpToDate Expert AI, support work such as record review, prescription preparation for clinician review, care coordination, insurance checks and operational analysis.
At CVS Health, one production agent reportedly cleared a months-long document-sending backlog over a weekend, while another handles system steps for tens of thousands of specialist consults daily; deployments use validation checks and human oversight, with screenshots and action records.
Cigna Healthcare is using OpenAI models to gather member context from more than 10 clinical and administrative systems for case managers preparing outreach, reducing time spent searching; the deployment moved from concept to launch in under eight weeks.
Ambience is testing agents that prepare prior-authorization requests from chart evidence, pause for clinician input when records are ambiguous and require approval before submission, while Abridge is testing cross-system discharge coordination with select partners.
Assort, Bunkerhill, Forus, Gravity and Tennr are also testing agents for referral management, authorization checks and medication approvals; OpenAI said it is working to improve agent speed and cost while partners handle specialized implementation.
OpenAI says organizations can restrict agent access and actions, require approvals and review audit logs; HIPAA workflows require eligible services covered by an OpenAI Business Associate Agreement and appropriate security and retention settings, and organization data is not used to train its foundation models.
Summary based on 4 sources
Get a daily email with more AI stories
Sources

OpenAI • Nov 16, 2026
More ways to move care forward with ChatGPT for Healthcare
Fierce Healthcare • Nov 16, 2026
HLTH 2026: CVS Health, Cigna and Abridge collaborating with OpenAI to test computer-use agents
Unite.AI • Nov 16, 2026
OpenAI Brings Computer-Use Agents to Healthcare Workflows
OpenAI • Nov 16, 2026
Advancing computer-use agents for healthcare teams