Great Ormond Street Hospital DRIVE unit (NHS England-sponsored)
GOSH DRIVE tests TORTUS AI scribe at nine London NHS sites: direct care share 70.0% to 86.5%, appointments 8.2% shorter
An NHS England-sponsored study led by Great Ormond Street Hospital's DRIVE unit tested the TORTUS AI scribe across nine London sites, from hospitals to ambulance teams. At core sites the median share of the consultation spent on direct care rose from 70.0% to 86.5%, and appointments were 8.2% shorter.
Key figures
Copied exactly from the sources. Copy any figure with its citation.
70.0%
to 86.5%
Median share of the consultation spent on direct care at core sites, baseline versus with the AI scribe2
23.5%
Relative rise in that direct care share (the report also gives it as 16.5 percentage points), as stated in the key findings and press release1
8.2%
Reduction in overall appointment length with the AI scribe1
13.4%
Increase in patients seen per shift in A&E (St George's University Hospital)1
What it took
The problem
Doctors and other clinicians spend much of each consultation typing notes and writing letters. NHS England wanted to know whether ambient voice technology (software that listens to a consultation and drafts the note) could free clinician time and improve patient care across very different settings.
What they built
Great Ormond Street Hospital's DRIVE innovation unit led an evaluation of an AI scribe supplied by TORTUS, a UK start-up. The tool turns the conversation into a draft note and letter, using templates that suit each clinician. It makes no clinical decisions, and the clinician checks and edits everything before it enters the record.
How they did it
The study ran for 12 months, from May 2024 to April 2025, at nine London sites across hospitals, GP surgeries, mental health care and ambulance teams. Core sites used a before-and-after design, observing clinicians with usual documentation and then with the scribe. York Health Economics Consortium modelled the financial effect. The work also led to the NHS T.E.S.T. framework for assessing health technologies.
Results
Over 17,000 patient encounters were evaluated. The median share of each consultation spent on direct care rose from 70.0% to 86.5% at core sites (a 23.5% relative rise, per the key findings), appointments were 8.2% shorter and A&E at St George's saw 13.4% more patients per shift. Clinicians reported a 35% fall in feeling overwhelmed by notetaking, and 92% of patients and families consented. In A&E, the report puts the cut in documentation time at 51.7% per clinician. NHS England then published a registry of 19 AI notetaking suppliers in January 2026.
What went wrong
Patient experience did not differ significantly with or without the scribe, and accuracy satisfaction was no higher than baseline. Some clinicians worried about lost nuance, omissions, and pressure to see more patients. Early sites rated the product lower while it was still being improved.
All reported figures
| Measure | Figure | Citation |
|---|---|---|
| Median share of the consultation spent on direct care at core sites, baseline versus with the AI scribe | 70.0% to 86.5%2 | |
| Relative rise in that direct care share (the report also gives it as 16.5 percentage points), as stated in the key findings and press release | 23.5%1 | |
| Reduction in overall appointment length with the AI scribe | 8.2%1 | |
| Increase in patients seen per shift in A&E (St George's University Hospital) | 13.4%1 | |
| Reduction in clinicians feeling overwhelmed by notetaking | 35%1 | |
| Patients and families consenting to AI scribe use | 92%1 | |
| Reduction in documentation time per clinician in A&E | 51.7%2 | |
| Patient encounters evaluated across the 9 sites | Over 17,0003 |
Lessons for your project
Measure before and after on the same clinicians. The core sites observed each clinician first with usual note-taking, then with the scribe in use.
Test across different settings. Hospitals, GP practices, mental health and ambulance teams gave different results, with A&E showing the clearest capacity gain.
Budget for templates, training and integration. The report lists at-the-elbow training, custom templates and workflow fit as key adoption enablers, and notes the tool was not integrated with every patient record system.
Keep clinicians in the loop. The AI makes no clinical decisions and clinicians check and edit every note before it is saved.
Agree how freed time is used. Some clinicians feared pressure to see more patients, especially in services for complex needs.
Caveats
- GOSH DRIVE, which led the evaluation, and NHS England, which sponsored it, publish all the figures. No independent party confirmed them.
- The 23.5% is a relative change in the median share of the consultation spent on direct care (from 70.0% to 86.5% at core sites), not a gain in minutes. The report's results section also gives 16.5 percentage points absolute and 23.6% relative, and its discussion loosely says 16.5%, so it is inconsistent. The case leads with the 70.0% and 86.5% medians.
- Clinicians chose to take part, not all professions were represented, surveys were not validated and patient surveys were not paired before and after. Patient and carer experience scores showed no significant difference with and without the scribe.
- The A&E result is based on a single hospital and few clinicians, per the report. Its national savings figures of 176 million and 658 million pounds are modelled projections, not results, and are not counted as outcomes here.
- Clinicians raised concerns about missing nuance, omissions in developmental and psychosocial detail, and satisfaction with accuracy of notes did not differ from baseline. Technology, templates and hardware were changing during the 12 months.
- The sources name the product, TORTUS, but not the AI model behind it. The deployed date is the start of data collection. The NHS England page is dated 15 January 2026, though its text says 16 January.
Sources
- GOSH-led trial of AI-scribe technology shows 'transformative' benefits for patients and clinicians across London
Great Ormond Street Hospital · 2025-09-04 · Press release · gosh.nhs.uk
- The use of Ambient Voice Technology with Generative Artificial Intelligence in Multiple Clinical Settings Across the NHS
GOSH DRIVE Unit · 2026-02-11 · Research paper · gosh.nhs.uk
- NHS backs AI notetaking to free up more face-to-face care
NHS England · 2026-01-15 · Press release · england.nhs.uk
Summarized in our own words from the sources above. Spotted an error? See corrections.