Oxford University Hospitals NHS Foundation Trust
Oxford University Hospitals pilots four AI note-taking tools, with nearly 90% of clinicians reporting time saved
Oxford University Hospitals tested four ambient voice tools that listen to consultations and draft clinical notes. The Trust reported that nearly 90% of clinicians saw less documentation time and 99.7% of patients consented.
Key figures
Copied exactly from the sources. Copy any figure with its citation.
Nearly 90%
Clinicians reporting reduced documentation time (Trust statement)1
87%
Users reporting reduced documentation time (Digital Health report)2
over half
Clinicians saving 30 minutes or more each day1
73%
Participants reporting improved wellbeing1
What it took
Cost, team size and timeline were not disclosed in the sources.
The problem
Doctors and nurses spend a large part of their day writing notes and letters. Oxford University Hospitals wanted to see whether AI could ease that load without hurting care or patient trust.
What they built
The Trust's digital innovation team, TheHill, ran a pilot of ambient voice technology. A microphone and AI software capture the spoken consultation and turn it into a draft clinical note or letter. The clinician reviews it before it goes into the health record. Four products were tested: Accurx Scribe, Heidi, Lyrebird Health and TORTUS.
How they did it
The pilot ran from July to November 2025 across inpatient and outpatient settings with different workflows. Digital Health reports 160 uses by 136 people, among them clinicians, nurses, other health professionals and operational colleagues. An evaluation report summarises their experience.
Results
The Trust reports that nearly 90% of clinicians saw less documentation time and over half saved 30 minutes or more a day. It also reports 73% with improved wellbeing and 99.7% of patients consenting to the tool. Digital Health reports 87% of users saved time, most often one to 10 minutes per interaction, and that users agreed the record was accurate in 81.5% of logged encounters.
All reported figures
| Measure | Figure | Citation |
|---|---|---|
| Clinicians reporting reduced documentation time (Trust statement) | Nearly 90%1 | |
| Users reporting reduced documentation time (Digital Health report) | 87%2 | |
| Clinicians saving 30 minutes or more each day | over half1 | |
| Participants reporting improved wellbeing | 73%1 | |
| Patients consenting to use of the tool during the pilot | 99.7%1 | |
| Different people who used the tools | 1362 | |
| Logged encounters where users agreed the record was accurate | 81.5%2 |
Lessons for your project
Test several vendors side by side in different settings (inpatient and outpatient) before buying, as Oxford did with four tools.
Keep a clinician review step. Notes were checked by the clinician before entering the record.
Measure both time saved and staff wellbeing, and track patient consent, since trust affects whether the tool can be used.
Use the pilot to learn how to structure procurement and rollout, not only whether the tool works.
Caveats
- The results are self-reported by staff in surveys and in-pilot data capture, and are published by the Trust that ran the pilot.
- The Trust says nearly 90% and Digital Health says 87%; both are shown, with their sources.
- The sources name four products but not the language models behind them, so the model is shown as undisclosed.
- The Digital Health page returned an access block to automated requests, so an archived copy was read.
- Trust figures cover 'clinicians'; Digital Health's 87% covers all users including nurses, allied health professionals and operational staff. The pilot was small (160 recorded uses).
Sources
- Successful AVT pilot at OUH shows promising results for staff and patients
Oxford University Hospitals · 2026-02-23 · Company blog · ouh.nhs.uk
- NHS pilot of AI scripting tech reduced admin time for 87% of users
Digital Health · 2026-02-25 · News · digitalhealth.net
Summarized in our own words from the sources above. Spotted an error? See corrections.