From Principles to Procedure, Not Just Promises

HTN reported on 1 September that South East Coast Ambulance Service NHS Foundation Trust (SECAMB) has deployed semi-autonomous "auto-dispatch" and "auto-allocation" tools in its emergency operations centre, automatically allocating ambulance crews to category 2 calls and automating call triage under a "hear and treat" model. The emergency operations centre element was planned for rollout from the first quarter of the 2026/27 financial year, meaning it remains a live, recent deployment rather than a one-off pilot from further back.

SECAMB set a target of cutting crew allocation time for category 2 calls by one minute by the end of Q4, and has so far achieved an 11-second reduction. A related intelligent call queue tool has lifted clinician call-handling from a baseline of 1.35 to 1.43 calls per hour, against a target increase of only 0.1 — already ahead of that goal.

HTN's report sets out what the auto-dispatch and auto-allocation tools do and the metrics behind them, but does not state who reviews a call the system allocates or triages incorrectly, or how quickly that review happens. For a tool making live decisions in emergency care, that is the detail worth asking a supplier — or your own emergency operations centre team — for directly, rather than assuming it exists.

From 17 August 2026, new NHS England guidance applies to the procurement of new robotic-assisted surgery systems where the procurement had not already been initiated, HTN reported on 3 September. The guidance states: "Further competition should be the default route for robotic assisted surgery purchasing under national framework agreements."

Direct awards are now limited to four defined exceptions: expanding an already-installed robot's use within existing clinical pathways; genuine urgency or continuity-of-care risk arising from unforeseeable events, capped at 12 months and non-renewable; time-limited rental or bridging arrangements while a trust runs a competitive process; or cases where only one supplier can meet a narrowly defined clinical need that no alternative CE-marked device can meet. A trust taking the direct-award route needs sign-off from its clinical director, chief finance officer and procurement lead, then 15 working days of review by NHS England's Commercial Spend Controls Team before the contract can be awarded.

The guidance states that procurement decisions should "promote healthy markets while ensuring the NHS secures value for money and the best possible outcomes for patients", and that "ultimate accountability for procurement decisions and compliance" sits with the contracting trust. It puts a named process behind the kind of scrutiny the Royal College of Surgeons has been asking for since warning that robotic access was a "postcode lottery" — a story we covered in our 4–10 August issue.

Leeds Community Healthcare NHS Trust and Leeds and York Partnership NHS Foundation Trust published outcomes from a pilot of the Heidi ambient AI scribe at Leeds Gender Identity Service, HTN reported on 24 September.

A pre-pilot staff survey found 72.7% reported a negative impact on their work from administrative workload, with more than half working beyond their contracted hours to complete documentation. After the pilot, 91% of service users said clinicians were "more focused", and 73% felt the same or more comfortable during appointments. Clinicians reported "richer, more person-centred interactions, with less pressure to capture information during appointments and more opportunity to concentrate on the individual in front of them."

HTN's report doesn't state how many clinicians or patients took part, the pilot's start and end dates, or a named clinician's own account of using the tool — worth asking for before treating the results as generalisable beyond this service. It's also a gender identity service, where patients may be particularly sensitive about being recorded, which is a fair question to put to any trust citing these figures as a case for wider rollout.

On 10 September the National Commission into the Regulation of AI in Healthcare published its final report and recommendations for a future regulatory framework — the fuller detail behind the staged "L-plates" approval model we first covered in our 15–21 September issue.

The report draws on engagement with more than 12,000 people, including 78 members of the public in Health Foundation deliberative workshops held between March and April 2026. Alongside staged authorisation, it recommends continuous real-world monitoring of AI tools throughout deployment rather than only at initial approval, a public searchable database of AI device safety data and adverse incidents, a requirement that patients be told when AI is used in their care, and enhanced enforcement powers for the MHRA.

Commission chair Professor Alastair Denniston: "People are open to AI improving their care, but only if it is safe, overseen by humans, and if they know when it is being used." Deputy chair Professor Henrietta Hughes: "Patients want to know when it's used, that it supports rather than replaces clinicians, and clear accountability if things go wrong." MHRA chief executive Lawrence Tallon: "We have heard clearly about the opportunities of AI to speed up and improve care, and to release more of clinicians' time." University of Cambridge DeepMind Professor Neil Lawrence: "We can't rely on a single point of approval and assume the job is done."