TRAINING AND DEVELOPMENT Digital governance
CATCHING WHAT SLIPS THROUGH: AI-ASSISTED POLICY REVIEW FOR NHS ESTATES AND FACILITIES
Whilst the compliance burden is rising, EFM teams struggle to keep compliance documentation up to date due to budget and workforce cuts in the NHS. Given the number of regulatory documents to be taken into consideration, reviews are very time-consuming. In this article, Niklas Kronewiter and Dr Carl-Magnus von Behr discuss how AI could be a solution for this problem.
This article reports on a research study comparing manual policy review with AI assisted review. The results show that error detection can be doubled without removing human judgement.
A system under strain
The NHS is contending with some of the toughest operational conditions in its history. Waiting lists stand at 7.4 million,1
exceeds £13.8 bn2
the maintenance backlog , and teams are asked
to do more while contributing to a national net-zero commitment. Estates and Facilities Management (EFM) staff sit at the heart of this challenge: their work makes sure that buildings and systems are safe, sustainable and efficient. Nonetheless, more than
50% of the NHS estate is not fit-for- purpose.3 Policies, standard operating procedures, and risk assessments must be written, reviewed, approved, and kept in step with the latest versions of Health Technical Memoranda (HTMs), Health Building Notes (HBNs) and applicable legislation. Outdated and badly written policies have limited usefulness and often end up on shelves rather than guiding EFM staff in following best practice. However, keeping compliant documentation relevant is not only difficult, but also time intensive – one reason is the necessity of finding the right clauses across various regulatory sources.4
Participants in our study reported that a full review of a single
Dr Carl-Magnus von Behr Director, CompliMind (formerly
INNEX.AI)
Carl completed his PhD at the University of Cambridge in 2024, researching information and knowledge access across the NHS. He co-founded CompliMind (formerly
INNEX.AI) with Cambridge AI and machine learning specialists, developing tools to streamline information access and collaboration with NHS capital, estates and facilities teams. His work focuses on compliance, administration and reporting workflows, with an emphasis on evidence, auditability and reducing the administrative burden so staff can focus on safe and efficient operations.
12 Health Estate Journal November 2025
policy can easily consume around ten hours of uninterrupted effort, with as many as ten people involved in the drafting, review, and approval process. From initiation to final approval, seven weeks often pass. Even after this time investment, only about 40% of participants reported they feel confident that their approved documents are fully compliant. In other words, significant time is being spent for a sub-optimal outcome.
When policy implementations fail, the
risks and costs are high What sounds like paperwork is not an abstract problem. In 2021-22, NHS England reported 5,348 clinical incidents caused by estates and infrastructure failure.5
In the aftermath of an incident
on the estate, regulators or coroners typically begin by asking for the relevant policies. Several Trusts have faced serious penalties in recent years: Southern Health was fined £2 m for systemic failings.6 Essex Partnership paid £1.5 m plus £86,000 in costs.7
Maidstone & Tunbridge
Wells was fined £180,000 plus £15,000 in costs.8
Since 2016, fines for large NHS
Trusts can legally reach £20 m.9 As a result, the national PAM returns do not simply ask for the presence of a set of certain policies, but ask for a “current, approved policy and an underpinning set of procedures that comply with relevant legislation and published guidance” for
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76
