ChatGPT for Healthcare agents can now click through the EHR for staff
OpenAI adds agents that read screens and click through hospital software, plus upgraded clinical search and more than 20 plugins, with admin controls and audit logs.

OpenAI is adding computer-use agents to ChatGPT for Healthcare, letting AI carry out multi-step work across existing hospital software by reading screens, clicking and typing. The announcement of 16 November also covers upgraded clinical search and more than 20 new plugins.
Agents that work inside the software clinics already run
According to OpenAI, healthcare teams spend much of their time stitching together information from the electronic health record, hospital policies, scheduling systems and payer portals. The computer-use agents are meant to do some of that stitching: they move between applications the way a person would, rather than relying on a custom integration for each one.

OpenAI says the release includes models it describes as more capable, safer and cheaper for clinical and administrative work, plus access controls and audit logs for secure deployments. Organizations pick which applications an agent may touch and which steps need a human to review them first. In the demo OpenAI shows, a doctor talks to the ChatGPT app and asks for help chasing an outstanding MRI order. The agent goes through the patient’s chart and prepares a draft order and a message to the patient, both left for a person to approve.
Abridge, Ambience, Cigna Healthcare and CVS Health are named as organizations deploying agents for tasks such as processing referrals and moving documents between systems. OpenAI points to a separate post, “Advancing computer-use agents for healthcare teams”, for customer stories and research.
More than 20 plugins and a search upgrade rated by physicians
The plugin list covers clinical care, care coordination and business operations. It includes Amion, Availity, Elation Health, IMO Health, Luma Health, Metriport, Oracle HDI, Redox, Stedi, TriZetto, Turquoise and UpToDate Expert AI. All are available now in ChatGPT for Healthcare, including through ChartGPT, which places ChatGPT in the EHR workflow. OpenAI’s examples chain them together. A care navigator comparing two hospitals for a knee MRI might pull a procedure code from IMO Health, check eligibility in Stedi and look up negotiated rates in Turquoise Health. A nurse might fetch a hospital insulin-pump policy from AIVA and the covering endocrinologist from Amion.
Clinical search, which finds, weighs and cites medical evidence, was also improved in both ChatGPT for Healthcare and ChatGPT for Clinicians. OpenAI says physicians in a blind evaluation, across more than 1,300 reviews of real-world questions in several specialties, rated the updated version higher on all eight dimensions measured. Those include factual correctness, safety, actionability and source authoritativeness.
Availability varies by user. Healthcare customers get agents in ChatGPT Work and Codex and the improved search now, subject to workspace settings, and administrators must switch on the new plugins. Verified US clinicians can use ChatGPT for Clinicians for free, with select plugins. Developers can reach computer-use capability through the OpenAI API, ChatGPT Work and Codex, but must contact sales to discuss clinical use. ChatGPT for Healthcare can be bought directly or through an AWS Marketplace private offer. OpenAI offers a Business Associate Agreement for both products, and says content shared with ChatGPT for Healthcare is not used to train models.
What the announcement leaves unanswered
The evidence for the search upgrade is OpenAI’s own. The company does not say in the announcement who the physician reviewers were, how they were chosen, what the previous version scored, or how large the gaps were. “Rated more highly” on eight dimensions is a direction, not a measure, and no independent evaluation has been published.
The agent claims are similarly unquantified. OpenAI calls the models more capable, safe and affordable, but gives no error rates, no per-task cost and no pricing for ChatGPT for Healthcare. For a system that clicks through a chart and drafts orders, the failure rate matters more than the demo. OpenAI’s answer is structural: human review steps, application allow-lists and audit logs. That puts much of the safety burden on each hospital’s configuration. The company also notes that HIPAA-compliant use depends on the customer using eligible services under its agreement, applying required security and retention settings, and extending protections to connected systems.
Regional coverage is another gap. The free clinician tier is limited to verified clinicians in the US, and the announcement says nothing about availability elsewhere. That fits a pattern: when OpenAI launched ChatGPT Health for consumers on January 7, 2026, it excluded the EEA, Switzerland and the UK at launch, as we reported in our coverage of Apple’s health study. OpenAI said then that more than 230 million people ask ChatGPT health or wellness questions every week.
The clinician market is also contested. In the same coverage, press reports say more than 40% of US physicians use OpenEvidence daily and that it is free to doctors. OpenEvidence raised $250 million at a $12 billion valuation in January 2026, according to its own announcement. Those usage figures come from secondary reporting, not an audited source. OpenAI’s differentiator is breadth: its plugins let UpToDate Expert AI, payer-eligibility tools and scheduling systems sit in the same conversation as the evidence search, and its agents act on the result instead of only answering.
OpenAI says it is inviting more healthcare companies to build plugins with it, so the list is likely to grow. Hospitals weighing the agents will need to test them on their own EHR and payer portals, since the announcement offers no figures for how well they cope with the messy, vendor-specific interfaces the demo glosses over.
Sources
Sign Up for the Daily AI Pulse
One email a day. All the stories that matter.
By signing up, you agree to our Terms of Use and acknowledge the data practices in our Privacy Policy. You may unsubscribe at any time.
Alex CarterAI News & Big Tech CorrespondentAlex Carter writes for Aiholics, keeping readers updated on the fast-paced world of AI and Big Tech.
You may also like

Oracle’s business apps head into ChatGPT, with a US-only Marketplace deal

Making Chatgpt better for clinicians: A new era of AI-powered healthcare support

ChatGPT prompts for photos: Editing, pictures of yourself and how image editing works
