Most notably it taught me that most conversations about technology in policing happen at the wrong altitude. I spent years managing offenders, leading investigations that were complex, high-stakes and under-resourced. When you are in that environment, what you need from technology is very specific: you need it to cut your workload and surface the things that matter most.
However, too often tools are designed or procured by people who understand the technology but have never had to manage a caseload of sixty high-risk offenders and answer for every decision they make about which ones to visit. The gap between the boardroom and the officer trying to cross-reference three separate systems is often enormous – and it’s getting wider. Particularly in the context of three of policing’s most pressing technology challenges.
Community offender management
The core problem is scale meeting scarcity. The registered sexual offender population in England and Wales now exceeds 209,079, representing roughly a 50% increase over the past ten years. The proactive capacity to visit, monitor and disrupt is finite. If every case is treated with equal intensity, the highest-risk individuals get less attention than the risk they pose warrants. Risk is dynamic. A static assessment made at the point of release can quickly become outdated. An offender's housing situation, relationships, digital behaviour and contact with children can all change, with each development potentially increasing or reducing risk.
A monitoring model built on periodic, calendar-driven reviews will often miss the escalation signals that matter most. What is needed is automated risk assessment that ingests data continuously and alerts the officer at the exact moment new information is received that signals the need for intervention. A senior officer who has never run a caseload tends not to feel the urgency of that shift the way a frontline offender manager does. That is the gap I am referring to.
Online child sexual abuse material
While this problem is different in character from offender management, it shares the same structural flaw: the volume of offending material has completely outgrown the reactive model. Reports of AI-generated child sexual abuse material to the Internet Watch Foundation more than doubled between 2024 and 2025.
An “investigate every referral” approach can no longer absorb that volume. Officers working in this space know that viscerally, because they are living with the backlog and the psychological cost of the exposure to offending materials. The technology exists to use hash-matching and automated classification tools, linked to the Child Abuse Image Database, to triage known material quickly and reduce needless repeated human exposure to the worst content. Our work with SeaCAT is an example of that.
The Crime and Policing Act 2026 has also strengthened the legal framework, extending the definition of abuse material to cover AI-generated images and criminalising the creation of AI models optimised to produce such content. Converting new powers into effective operational practice requires senior leaders who understand the caseload. Victim identification should be a first-order priority. That prioritisation decision must be made at the top, informed by what is happening at the bottom.
Cybercrime
In this context, ranging from malicious communication to dark web activity and digital stalking, we’re talking about a skills and awareness gap as much as a technology gap. Investigating cybercrime demands digital forensics, cryptocurrency tracing, dark web investigation and network analysis skills that are scarce, expensive and highly sought after in the private sector. Digital stalking is a particular blind spot, sitting at the intersection of the high-volume offending and genuine danger.
Too many frontline officers encounter such offences without the tools or training to handle it correctly. And too many senior officers do not register it as a distinct capability gap requiring a dedicated response pathway.
The tiered model which sees local forces handling volume, Regional Organised Crime Units providing regional specialist support, and the National Crime Agency leading on the most serious threats is the right approach. But it only works if there is baseline competence at every level, and if senior leaders actively invest in building and sustaining it. Public-private partnerships are part of that answer, but only if they involve genuine knowledge exchange at all operational levels.
What’s the takeaway?
Chief Officers must start by getting closer to the ground on all three of these problems: community offender management, online child sexual abuse material and cybercrime. Getting direct feedback from the detective constable in the cybercrime unit; the offender manager with a MAPPA (Multi-agency public protection arrangements) caseload and the digital forensic examiner working through a device backlog is the most valuable intelligence you have on whether your technology investment is working.
In each of these areas, the scale of demand has already outstripped what human effort alone can address. That means the decisions made about automation, triage tools and workforce design have real consequences: for victims, for officers and for the general public. Those decisions should be made with a clear understanding of the challenges the frontline is actually facing. That level of understanding does not come from a vendor demonstration.
ILAS: Intelligence Lead Assessment Service
ILAS is an intelligence automation application that continuously ingests sub-threshold data points to monitor individuals for change and alerting officers and staff at the inflection point where an intervention may be needed. It replaces traditional periodic, calendar-driven reviews with a constant, automated monitoring function that tracks change across a population and surfaces the individuals whose circumstances have shifted to a point requiring action. This allows officers to focus their face-to-face capacity on cases that genuinely require it, rather than applying equal intensity across a caseload irrespective of current risk.