TRAINING AND DEVELOPMENT Compliance clarity
An example showing the CompliMind review assistant assessing a draft policy regarding the reponsibilities of waste managers, as utilised by the research experiment.
a multitude of hard and soft FM areas. Especially in areas such as water safety, fire safety, health and safety and anti- ligature, regulators demand precisely where estates teams carry the statutory duty to act.
Why current processes fall short
Most policy reviews today remain largely manual. Staff compare PDF versions of local text with national guidance,
Niklas Kronewiter Visiting Researcher, University of Cambridge
Niklas graduated from the London School of Economics (LSE) in 2024 and conducted this study as a Visiting Researcher at the University of Cambridge’s Cyber-Human Lab. He is passionate about the application of machine learning in an entrepreneurial context and in the healthcare sector.
sometimes on paper printouts marked up with highlighters. Tracking which sections were reviewed against which version of an HTM is difficult. There is limited searchability, and no consistent way to ensure that every clause has been addressed. This leads to a patchwork of local processes and a lack of standardisation or digital-first tools. Staff are left with the choice to copy from another Trust or reinvent-the-wheel every time a policy is up for review. The result is duplication,
variation, and occasional gaps that only come to light after an audit or incident.
Can AI help?
Large language models (LLMs) have made it possible to read, compare and extract information from long, complex documents. This opens a door for a new kind of computer-assisted workflow: one that automatically matches local policies to national guidelines and highlights what is non-compliant and what might be missing. However, in a time where AI is often glorified to improve every domain of knowledge work, real-world evidence helps to distinguish vague claims from facts. Yet, despite the challenges of quantifying a qualitative domain, this difficulty became the motivation for the experiment described here.
Building the experiment
To design a real-life experiment, we drafted an SOP for the Waste Manager
November 2025 Health Estate Journal 13
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
