NHS Puts AI to Work on Operating Theatre Time
Eight NHS hospitals began a year-long evaluation of ambient surgical AI on September 9, 2026.
Proximie, working alongside Amazon Web Services and Deloitte, is deploying its Intelligence Suite across 104 operating theatres, catheterisation laboratories, and endoscopy suites.
The programme is described as the largest surgical AI evaluation the NHS has run.
The goal is not to put new technology in surgeons' hands.
It is squeezing more use out of theatres that already exist.
What the Intelligence Suite Actually Does
Proximie's platform runs ambient sensors in the background of the operating theatre.
It captures surgical and operational data automatically, without requiring input from clinical staff.
That data is de-identified and anonymised at the point of capture.
AI models then convert the anonymised feed into real-time operational insight across the full patient pathway.
The output is designed to show hospital teams exactly where time is lost between procedures.
The Scale of the NHS Evaluation
The 12-month programme covers eight hospital sites and 104 theatres, cath labs, and endoscopy suites.
Six of the sites are in London, with two elsewhere in England.
The programme is targeting more than 20,000 procedures a year across the participating sites.
AWS is supplying the cloud and AI infrastructure, so no hospital needs to add on-site servers as the programme scales.
Deloitte is handling delivery and transformation work across the NHS Trusts and the NHS London region involved.
What Independent Data Already Shows
The 104-theatre pilot itself has not produced results yet, since it only began in September 2026.
But Proximie's platform is already running at other NHS trusts, and one of those deployments has an independently sourced data point worth noting.
At University Hospitals Coventry and Warwickshire, operations director Juliet Starkey described a specific discrepancy in her own trust's numbers.
Her hospital's own reporting system logged one theatre list at 66% utilisation.
Proximie's ambient sensors recorded the same list at 88% utilisation, clearing the NHS benchmark requiring at least 85% of a planned session to go toward active patient care.
Starkey made these comments publicly at the AWS Public Sector AI Symposium in London.
Why This Isn't an Autonomy Story
Nothing in this programme touches surgical decision-making.
The Intelligence Suite doesn't guide a surgeon's hands or suggest a clinical next step.
It answers a scheduling and utilisation question: how much of the theatre time that's already funded is actually being used.
That distinction matters because "surgical AI" gets used to describe two very different problems, one operational and one clinical.
This pilot sits entirely on the operational side of that line.
