The best AI scribe for GPs
Hanah was the best AI scribe across six GP consultations in our benchmark, with a note ready to finalise 1 minute 50 seconds after the consultation and the fewest note errors in every one.
THE READING ROOM
A clearer view of AI scribes.
Ideas, explanations and a closer look at the evidence. Understand the technology—and how we put it to the test.
START HERE / OUR METHODOLOGY
How we test AI medical scribes.
The consultations, metrics and review process behind the benchmark. What we measure, and what the results can tell you.
Articles & analysis
17 articles · newest firstRESULTS
REVIEWS
Heidi review
We tested Heidi on 12 recorded consultations, five times each. It was second of six on speed and hallucinations, captured 81% of clinical facts and averaged 9.0 note errors.
METHODOLOGY
The ScribeStandard benchmark
ScribeStandard records the same 12 consultations through six AI scribes, at least five times each, and grades every note blind against the clinical facts said in the consultation. Every recording, transcript, note and grading reason is published.
RESULTS
The best AI scribe for allergy and immunology
Hanah was the best AI scribe in our allergic reaction test, with a note ready to finalise 2 minutes 21 seconds after the consultation and the fewest note errors.
RESULTS
The best AI scribe for cardiology
Hanah was the best AI scribe in both of our cardiac consultation tests, suspected heart attack and worsening heart failure, with the fewest note errors in each.
RESULTS
The best AI scribe for dermatology
Hanah was the best AI scribe in our eczema consultation test, with a note ready to finalise 2 minutes 1 second after the consultation and the highest coverage.
RESULTS
The best AI scribe for French
Hanah was the best AI scribe in our French and English consultation test. Three of five Hanah notes were ready to finalise as written, and no other scribe had one.
RESULTS
The best AI scribe for gastroenterology
Hanah was the best AI scribe in our gastroenteritis test, with a note ready to finalise 1 minute 44 seconds after the consultation and the fewest note errors.
RESULTS
The best AI scribe for Hindi
Hanah was the best AI scribe in our Hindi consultation test, with the lowest word error rate, no hallucination flags and a note ready to finalise in 52 seconds.
RESULTS
The best AI scribe for mental health
Hanah was the best AI scribe in our low mood consultation test, with a note ready to finalise 1 minute 37 seconds after the consultation and the fewest note errors.
RESULTS
The best AI scribe for myotherapy
Hanah was the best AI scribe in our 50-minute myotherapy consultation test, with a note ready to finalise 1 minute 40 seconds after the consultation and the fewest note errors.
RESULTS
The best AI scribe for neurology
Hanah was the best AI scribe in our suspected TIA consultation test, with a note ready to finalise 3 minutes 41 seconds after the consultation and the highest coverage.
RESULTS
The best AI scribe for pain management
Hanah was the best AI scribe in our pain management dictation test, with a note ready to finalise 40 seconds after the dictation. The top scribes were close on accuracy.
RESULTS
The best AI scribe for urology
Hanah was the best AI scribe in our urinary symptoms test, with a note ready to finalise 1 minute 46 seconds after the consultation and about a ninth of the note errors of the next scribe.
RESULTS
Hanah is the best AI scribe in our benchmark
Across 12 recorded consultations, Hanah's notes had the fewest errors and took the least time to review, and Hanah ranked first on 19 of the 20 measures ScribeStandard publishes.
DOCUMENTATION QUALITY
Completeness and unsupported claims
A note can contain every relevant fact and still include claims the consultation does not support. ScribeStandard reports coverage and hallucination flags as separate measures.
WORKFLOW PERFORMANCE
Measuring session bandwidth
ScribeStandard measures all browser traffic in a recording session, including app loading, WebSockets and note finalisation, as well as the audio upload.