An ambient artificial intelligence scribe cut the time doctors spent documenting each outpatient visit by 15 percent and increased the share of the consultation spent making eye contact with the patient, without shortening the visit itself, in a time-motion study at Singapore General Hospital (SGH) published in JMIR Medical Informatics. Of 39 patients surveyed, 27 agreed that their physician focused on them more when the scribe was in use.
The peer-reviewed study, which the authors describe as the first real-world impact study of an ambient scribe in Asia, observed 169 consultations by nine clinicians across seven specialties between Dec. 27, 2024, and May 30, 2025. Observers timed each visit directly rather than relying on electronic health record timestamps. The study was published on March 31, 2026, received no funding, and the authors declared no conflicts of interest.
What the study found
Documentation time per consultation fell from a mean of 5.3 minutes without the scribe to 4.5 minutes with it, a 15 percent reduction that reached statistical significance in the authors’ mixed-effects model (adjusted difference of 0.8 minutes, P=.04) but not in the crude comparison (P=.07). Eye contact rose from 69.6 percent of the consultation to 77.1 percent, an adjusted gain of 7.4 percentage points (P=.009). Consultation length was statistically unchanged, at 11.5 minutes without the scribe and 10.9 minutes with it (P=.42), as was the total cycle time per patient.
The effects were not uniform. Seven of the nine clinicians reduced their documentation time, by between 7.9 percent and 34.6 percent, and eight increased eye contact. The findings, the authors wrote, suggest that “ambient scribes reallocate clinician effort toward patient interaction rather than enabling faster patient turnover.”
The tool, called Note Buddy, is an in-house system that combines automatic speech recognition with a large language model (GPT-4o) and supports English, Mandarin, Malay, Tamil and Cantonese. Only English, Mandarin and Malay consultations were observed, and 80.8 percent of scribed visits were in English. Notes were transferred into the electronic record manually because the tool was not integrated with it. According to SGH, Note Buddy was developed by SingHealth’s Division of Digital Strategy, launched in August 2024, and by December 2025 had supported more than 5,000 healthcare staff and generated more than 67,000 medical and administrative notes.
Patient surveys were collected in four sessions involving four of the clinicians; 39 of 45 eligible patients (86.7 percent) responded, and 16 of them were aged 65 or older. Besides the 27 (69.2 percent) who agreed the physician focused on them more, 24 (61.5 percent) said they would recommend the technology to other physicians, 20 (51.3 percent) answered neutrally and one said the scribe did not enhance the visit. No patient expressed discomfort.
A second study points the same way
The Singapore results sit alongside a mixed-methods evaluation from Gold Coast Hospital and Health Service in Australia, published in BMC Health Services Research on Jan. 8, 2026. That trial licensed an unnamed ambient scribe for 100 clinicians across 19 outpatient specialties for 16 weeks from July 2024, covering 7,499 scribed consultations. Among 22 patients who returned surveys, 68 percent said the clinician spoke more directly with them and looked less at the screen. Of 43 staff respondents, 84 percent reported a positive effect on efficiency, but 47 percent reported hallucinations in generated notes and 16 percent observed bias. The Australian authors concluded that the evidence of hallucination and potential bias “requires attention prior to any further adoption in routine clinical practice.”
Why it matters
Health systems are buying ambient scribes largely on the promise of time savings, and the Singapore data are a reminder that the saved minutes may show up as attention rather than throughput. A gain of roughly seven percentage points in eye contact, with consultations no shorter, suggests the technology changes how a visit feels more than how many visits fit into a clinic session.
The study also shows that a hospital-built tool, rather than a commercial product, can deliver measurable effects, which the authors argue is relevant where data privacy, cost or localization are concerns. For patients, the survey results are encouraging but modest: most noticed more attention and none reported discomfort, in a small sample at a single institution.
The caveats
This was a non-randomized observational study of nine experienced users selected by convenience from 27 eligible clinicians, in specialist outpatient clinics at a single quaternary hospital. Clinicians chose which sessions used the scribe, observers were not blinded, and interrater reliability was not assessed. The authors note that senior clinicians may already have efficient documentation habits, that manual note transfer likely understated the potential time savings, and that note accuracy was not assessed in this study. The patient survey covered 39 people and may reflect social desirability bias. The Australian study was likewise single-center, with its note-quality analysis based on 18 note pairs scored by one senior clinician.
This article is for general information and does not replace advice from a qualified clinician. Corrections: email the editor using the address on the Contact page.
