The Rulebook Gets a Timetable

UKRI reported on 8 October that seven health start ups have joined the Science and Technology Facilities Council's DeepTech Catalyst Health programme at Daresbury Laboratory. One of them, Asika, is developing AI agents for hospitals and clinics that automate administrative workflows, with the aim of reducing paperwork and improving operational efficiency. Each company receives a support package worth up to £70,000. The programme launched in 2024 with £1.8 million from the Liverpool City Region Combined Authority Innovation Zone Programme. The announcement describes work in development. It names no NHS site, no deployment and no results.

The MHRA's AI Airlock Phase 3 call, published on 6 October, sets post market surveillance as its priority challenge area. One of its three topics asks how monitoring can detect and respond to safety or performance signals in settings with varying levels of clinical or organisational visibility. The page gives remote, direct to consumer and higher autonomy systems as examples, and refers to contexts with limited human oversight. The call is scoped to AI enabled medical devices. The link to administrative agents is uPull.ai's reading.

The UKRI announcement of 8 October includes HMA Technologies, which is developing a wearable robotic arm support. The aim is to help stroke survivors carry out intensive rehabilitation exercises independently at home, supporting recovery beyond traditional clinical settings. The company joins the DTC Health incubation programme at Sci-Tech Daresbury in the Liverpool City Region. Eligible companies for the programme are based in the Liverpool City Region, are less than five years old and focus on health and life sciences.

The source says the device could help stroke survivors, which describes potential. It reports no trial data, no regulatory status and no NHS pathway. A device used independently at home raises questions for any future pathway: who reviews the exercise data, and who responds when the device or the patient's progress shows a problem. Those questions are uPull.ai's own, and the source does not address them.

Berkshire Healthcare NHS Foundation Trust published a pre tender notice on 10 September for an ambient voice technology solution valued at £3.24 million. The contract runs for two years with optional extensions for two more, and the trust is buying through the Artificial Intelligence Dynamic Purchasing System RM6200. The notice requires every bidder to be certified on NHS England's Ambient Voice Technology Self-Certified Supplier Registry. The trust intends to create a supplier shortlist on 13 October and issue a capability assessment on 14 October, to be returned by 21 October. Tender documents are planned for 10 November, and the trust reserves the right to amend these dates. We read the notice through the Stotles tender aggregator, which reproduces Find a Tender record 085619-2026.

HTN reported on 5 October that Berkshire Healthcare is delivering face to face Copilot training aboard its Health Bus, using practical demonstrations and showing staff how to create emails, presentations, summaries and reports. The final roadshow session runs at Wokingham Hospital on 26 October 2026, and trust colleagues book through the digital learning hub. HTN also records the trust's five year strategy, which cites an initial income and expenditure planning gap of £15 million for 2026/27 and a two percent annual productivity growth requirement. The trust expects AI ambient documentation, electronic record integration and process automation to help. Reading the tender and the training together, one buyer is preparing its supplier requirements and its workforce in the same autumn. That reading is uPull.ai's.

The notice makes registry certification a condition of entry for every bidder. It also asks for a system that supports inpatient wards, outpatient clinics and community settings, with background noise, shared spaces and variable connectivity, and consultations involving carers, family members or interpreters with accurate speaker differentiation. The stated aims include higher quality documentation, reduced errors and shorter turnaround for clinic letters, while maintaining high standards of clinical safety and transparency. The notice sets out requirements and says nothing about how performance will be checked after go live.

On 6 October the government confirmed that it accepts all 44 recommendations of the National Commission into the Regulation of AI in Healthcare, which the Commission published on 10 September. Delivery will run across the UK under a newly established Programme Board, with ten priority areas. The response sets three dated steps. The MHRA will issue draft guidance by December 2026 on managing changes to AI enabled medical devices as they adapt and improve. It will consult in 2027 on how AI enabled devices are qualified and classified, which was the Commission's first recommendation. A full implementation roadmap with timelines and responsibilities follows by Spring 2027. The government is also exploring staged authorisation pathways that would let promising tools enter NHS use earlier under close supervision. We covered the Commission's report in our 29 September to 5 October issue. This story is the government's reply to it.

The MHRA opened applications for the third phase of AI Airlock on 6 October. Phase 3 focuses on post market surveillance and lifecycle regulation, and the programme is funded until April 2029. Applications received before 31 October join the current review wave, the window stays open after that date, and a further shortlisting period follows in Spring 2027. The MHRA expects to select its first wave of innovators in November 2026. There is no application fee. Candidates fund their own studies and testing, including data access, and should expect a testing plan of 6 to 12 months. Testing can use a simulation roundtable, a virtual test space, or a real world setting run outside patient pathways, such as shadow deployment. Organisations of all sizes may apply, and a lead applicant may partner with other organisations.

Professor Henrietta Hughes, Deputy Chair of the Commission and Patient Safety Commissioner for England, welcomed the commitments to monitor AI throughout its use in practice, to strengthen patient involvement in decision making and to provide clear routes to redress when care falls below expectations. The government release also refers to new commitments to keep patients informed about when AI is used in their care. The response describes work still to come, with draft guidance, a consultation and a roadmap all dated ahead.