AI is already being used across the NHS to reduce admin burden, improve productivity and support the workforce.
As the NHS builds on this progress, attention is turning to clinical-facing tools that tackle one of the most persistent pressures on staff time: documentation. Ambient voice technologies (AVT) are emerging as a practical option to reducing the burden of more traditional approaches to documentation and other administrative tasks. AVT uses speech recognition and generative AI to securely capture conversations between clinicians and patients and turn them into structured notes and letters. The aim is to improve consistency of documentation and allow clinical time to be used as effectively as possible – reducing time spent in administrative tasks and, in turn, releasing capacity to spend with patients requiring greater levels of care. Globally, around 60 ambient scribe products are now in use and adoption is accelerating. Indeed, 16 January 2026 saw NHSE publish a national registry of 19 suppliers of AVT. The self-certified registry requires suppliers to comply with standards on clinical safety, technology and data protection. The registry's launch builds on NHS guidance published in 2025 that emphasised the importance of adopting platforms from a cohort of eight suppliers with regulatory authorisation. AVT is treated as a medical device, typically class IIa or IIb, depending on functionality and risk. Products must meet the standards required covering clinical safety and information governance, including secure data handling, encryption and audit trails. Where systems generate summaries using large language models, outputs still require clinical review. Human oversight is essential to manage transcription errors and correct misleading content.
Deployment also requires strong data governance. NHSE's guidance expects Data Protection Impact Assessments and compliance with Caldicott principles - including consent, data minimisation and transparency on how recordings are processed and stored. The NHS TEST framework is now supporting evaluation of emerging digital health solutions, including AVT. It helps NHS teams assess whether a technology is safe, effective and ready to scale. A Great Ormond Street Hospital led evaluation of the TORTUS solution across nine London NHS sites, bridging 2024 and 2025, covered over 17,000 patient encounters. Reported outcomes included a 23.5% increase in patient interaction time, shorter appointments and a 35% reduction in clinician overwhelm. Economic modelling suggested potential savings of up to £176m a year (if scaled nationally) through reduced documentation time and increased throughput.
Several similar pilots are underway across the country, facilitated by Health Innovation East, Health Innovation Kent, Surrey and Sussex, and Health Innovation Yorkshire and Humber. Early feedback points to improved administrative efficiency and patient satisfaction. The pilots have also identified practical challenges around consent workflows and data management, particularly in settings where appointments consult on sensitive matters, eg mental health. Findings such as these reinforce a consistent lesson gleaned from the adoption of digital technologies across the NHS: outcomes depend on how tools are introduced, governed and embedded into pathways and not only on the capability of a product.
The value of AVT depends on effective integration with EPR systems, reliable performance in real clinical conditions and clear operating procedures. Implementation of AVT will require testing of interoperability, regulatory status, training requirements and how clinicians will review and sign off AVT outputs. All of which indicate that NHS leaders should assess AVT as a service improvement tool, not a standalone digital product
In mental health and other areas, teams need to agree where AVT supports practice and where it risks undermining trust. Clinician discretion and patient choice should be built into local operating models, supported by clear guidance and staff training. Funding may be available through productivity programmes, capital allocations and AI-specific funding routes. However, investment should be paired with local evaluations that test whether time saved in administration translates into improved access, reduced waiting pressures or the capacity for delivering better quality.
Health innovation networks can help NHS organisations move from expressing interest to informed adoption. Their assistance includes convening early adopters, co-ordinating evaluations, sharing learning nationally and helping teams apply national guidance consistently. Support also includes working with suppliers so that AVT solutions meet NHS requirements for safety, interoperability and real-world usability. AVT has the potential to release staff time and reduce administrative burden. Realising that potential depends on robust evidence, considered governance and implementation efforts designed equally around clinical pathways and patient trust.
