OpenAdaptPatient note entry in the EMR
OpenAdapt can add a note to a patient's chart as a program learned from a demonstration. On the public OpenEMR demo, an 18-step note task passed 19 of 20 runs with no AI model calls. That test used a screen check of the saved note, not a database read, and an older engine version.
Updated 2026-10-18
The task
- Sign in and find the patient.
- Open the chart and scroll the dashboard.
- Go to Patient Messages and start a note.
- Type the note, a new value each run.
- Save.
What's been tested
On 8 Jul 2026, on the public OpenEMR demo (synthetic patients, shared, resets daily), the program passed 19 of 20 runs at 39.2 seconds median, with 0 AI model calls. In the 20th, it stopped and the note stayed in the unsaved form. On the same task, a claude-sonnet-5 computer-use agent passed 10 of 10 at 70.4 seconds median, about $0.55 a run. The agent scored higher.
Both were judged by the same screen check of the saved note, not a read of the EMR's data. It's a small sample on engine version 0.1.0. The comparison in full.
How it was judged
The 19 of 20 result was judged by a screen check of the saved note, not by a read of the EMR's own data.
Questions
Who writes the note?
A person or your own tool writes it. OpenAdapt only types it into the chart, after a person approves.
Related: referral intake · worklists · how it works
Sources
- OpenAdapt engine benchmarks, OpenEMR public demo, 8 Jul 2026 (docs/BENCHMARKS.md): an 18-step add-a-note task on the public demo (synthetic patients, resets daily). The compiled program passed 19 of 20 runs at 39.2 s median with 0 model calls; in the 20th the note stayed in the unsaved form. A computer-use agent, Claude Sonnet 5 (claude-sonnet-5), passed 10 of 10 at 70.4 s median, about $0.55 a run at list price. Both judged by the same screen check of the saved note, not a database read. Engine version 0.1.0.
See it on your process
Bring one task your team types every day. We'll tell you whether it fits, and what a test on your own system would cover.