The Decoder ↗ · 2026-11-18 00:55 · UTC+8
AI agents are starting to fill out medical forms and file insurance requests in place of human staff
Independent The Decoder Source language: English
OpenAI wants GPT-6 Astra and GPT-6 Luna to automate administrative tasks in US healthcare through computer-use agents that operate existing clinic software and insurance portals the way a human would.
Full text
AI in practice
AI agents are starting to fill out medical forms and file insurance requests in place of human staff
Alan Truly
Nov 17, 2026
Nano Banana Pro prompted by THE DECODER
Key Points
OpenAI deploys GPT-6 Astra and Luna as software agents that operate clinic programs and insurance portals using mouse clicks and keystrokes.
In tests, both models reliably handled medical administrative tasks. GPT-6 Luna cost about 8.5 cents per task.
US companies including CVS Health already use the agents for referrals. The offering currently targets the US market and its data privacy requirements.
OpenAI wants GPT-6 Astra and GPT-6 Luna to click and type their way through clinic software and insurance portals as digital clerks. GPT-6 Luna handles administrative tasks for about 8.5 cents per task, according to the company.
In a recent product announcement, OpenAI targets the lack of APIs in healthcare. The computer-use agents integrated into the Agents API in late September operate existing user interfaces the way a human would. In this case, they transfer data between electronic health records and external web portals.
According to OpenAI, when a patient reports back pain, the agent is meant to search relevant prior findings and draft both an MRI order and a reply to the patient. It should also prepare a prior authorization request on the insurer's portal. Clinicians and the responsible administrative staff review these drafts before they are released. The features are available through the API as well as in ChatGPT for Healthcare and ChatGPT for Clinicians.
Video: OpenAI
GPT-6 Luna completes admin tasks for 8.5 cents per task
In an evaluation described by OpenAI, GPT-6 Astra worked through tasks from electronic health record inboxes, among other things preparing an ibuprofen dosage for a child with a broken arm. Physician reviewers rated 99 percent of the 325 tasks evaluated as clinically safe. In 93 percent of 646 reviews, the reviewers called the prep work very or extremely helpful.
Evaluation of GPT-6 Astra’s inbox tasks by physicians using maximum reasoning effort. | Image: OpenAI
For administrative workflows such as prior authorizations, appeals, or medical equipment orders, OpenAI points to the independent HealthAdminBench benchmark from Stanford University researchers. GPT-6 Astra scored 94.3 percent of possible points. GPT-6 Luna, built for low API pricing, scored 91.8 percent and, according to OpenAI, cost about 8.5 cents per task, 98 percent less than Astra. Both models ran at maximum reasoning effort on the benchmark.
Major US healthcare companies test and deploy OpenAI's computer-use agents
Several partner organizations are helping advance the technology. At CVS Health, the agents already process referrals and specialist consultations across separate systems. According to the announcement, one agent cleared a months-long backlog of outgoing documents in one weekend, while another handles the system steps for tens of thousands of consultations daily. CVS Health published its software architecture as Agent Kit on GitHub.
At insurer Cigna Healthcare, the models merge patient information from more than 10 separate systems. Abridge is developing the agents for planned use in clinical discharge planning. Ambience has demonstrated additional safety mechanisms in which an agent stops when a patient's record is unclear and asks a physician for input. Startups like Assort also use the feature in Codex to complete referrals in practice management software whose APIs don't support that final click.
Data privacy, health records, and Europe
Because the agents work directly in electronic health records, protecting sensitive health data is a core concern. OpenAI says it does not use organizational data to train its base models. ChatGPT for Healthcare offers role-based access controls and audit logs, according to the company. Organizations can also configure the applications and environments so an agent only accesses the records and actions needed for a given task.
For HIPAA-compliant workflows with protected health information, organizations must sign a Business Associate Agreement (BAA) with OpenAI. They also need to use the covered services and apply the required security and retention settings. The healthcare facilities themselves are responsible for proper use, but OpenAI, as a contracting party, also has its own obligations under HIPAA. Verified medical professionals in the US who use ChatGPT for Clinicians, a free individual plan, can also sign a BAA, though it applies only to the features the agreement covers.
In the CVS Health deployment, data stays within infrastructure that CVS controls, according to OpenAI. CVS reviews the agents' work steps through screenshots and action logs. The announcement does not say where those screenshots are processed or how long they are stored. It also does not address the European General Data Protection Regulation (GDPR). So far, the offering targets the US market.
OpenAI frames the technology primarily as a tool against administrative friction in a highly fragmented healthcare system. Whether it sees broad adoption likely will not depend solely on the accuracy rates and costs cited in the company's announcement. Also decisive is how reliably the agents connect to existing systems and whether AI agents can be dependably monitored in practice. Another open question is whether a human is required to decide on clinically or legally relevant steps.