NHS App AI triage cut phone queues 29% in pilot

TL;DR:

• An AI triage tool built into the NHS App reduced phone queue volumes by 29% at Wealden Ridge Medical Partnership in Sussex, according to the NHS Transformation Directorate.
• The tool reaches more than 200,000 patients in London over the next 12 months, before becoming available to all NHS App users nationwide by April 2028.
• NHS England is separately rolling out ambient AI notetaking, with clinicians reporting up to a quarter more time spent with patients.

The distinguishing feature of this announcement is that it carries a number and a date. The tool asks patients about their health issue through the NHS App, adapting follow-up questions and requesting further detail or photographs depending on the answers, then produces a summary to help clinicians with diagnosis and prioritisation. A clinician reviews it before the patient is contacted by phone or SMS with next steps.

Ragu Rajan, GP senior partner at Wealden Ridge, described the benefit as sequencing rather than substitution: “Patients can tell us what they need, in their own time. That gives us a clearer picture before we respond, which helps us get people to the right care more quickly. The AI doesn’t replace clinical judgment; it helps us use our time where it matters most.”

The wider transcription push

NHS England is also extending AI transcription of patient-clinician conversations, starting with hospital appointments that do not require an overnight stay. A Great Ormond Street Hospital study of ambient voice technology found national scaling could create capacity for more than 9,000 additional A&E consultations a day. Chief executive Jim Mackey said clinicians using AI notetaking are finding they can spend up to a quarter more of their time with patients.

Regulation is lagging the roll-out

The deployment timetable is running ahead of the framework meant to govern it. The MHRA’s National Commission into the Regulation of AI in Healthcare found 73% of respondents disagreed that the current regulatory framework is sufficient to ensure safety and performance standards, 61% doubted its handling of data governance and privacy, 61% considered clinical evidence requirements insufficient, and 65% wanted more post-market surveillance. Researchers have also flagged five distinct risks in clinical AI scribe systems, the same category of tool now scaling across hospital appointments.

Looking forward

A 29% reduction at one Sussex practice is a pilot result, not a national one, and the gap between a well-supported trial site and 200,000 London patients is where these programmes usually meet reality. The April 2028 date gives the MHRA roughly 20 months to close a framework gap that three-quarters of its own respondents say exists. For UK health technology suppliers, the direction is now firmly set — triage and documentation first, with clinical decision support following the regulation rather than leading it.