AI & agents
OpenAI puts computer-use agents to work in healthcare software
- Published
- Source
- OpenAI Blog
Summary
OpenAI announced computer-use agents for healthcare that read screens, click, and type their way through multi-step work in the software hospital teams already use. In its own evaluations, 99% of 325 reviewed electronic health record (EHR) inbox tasks were judged clinically safe, and physicians rated 93% of 646 assessments as very or extremely helpful. On HealthAdminBench, an independent benchmark from Stanford researchers, GPT-6 Luna scored 91.8% at 8.5 cents per task, 98% cheaper than Astra. CVS Health said a production agent cleared a months-long document-sending backlog over a weekend, and open-sourced its Agent Kit.
Why it matters for our work
Agents are starting to do real work even in older medical systems that lack APIs. The safety and cost figures come from OpenAI's own evaluations and benchmarks, but they show that delegating repetitive work to agents while keeping human review steps is becoming practical in healthcare.
Translated from the Korean original. Summaries may be translated and edited. Commentary reflects our perspective; forecasts remain the source’s views.