Last month, after 24 years in policing, I moved from service leader to service enabler. As a newly appointed member of the Salesforce Policing, Justice and Public Safety team, I attended the 2026 Modernising Criminal Justice Conference, this time listening, rather than speaking. Police, CPS, HMCTS, HMPPS - the people who run the Criminal Justice System (CJS) were all in the room, and one thing was unmistakable: frontline teams are working harder than ever, and they're still being held back by the gaps between our legacy systems.
As a former Chief Constable, the focus is always on officers and staff - workload, wellbeing, resource levels, leadership and training - with a relentless pursuit as to how you help them to deliver the best service possible within your available budget. But sitting on this side of the room, the clearer it became that everyone working inside the CJS is experiencing the same problems. Victims chasing updates. Witness time wasted with multiple court date changes. Defendants unprepared owing to incomplete case files. Families who can't get a straight answer. They are all caught by the same challenges: a chain of organisations that don't speak to one another, knitted together by humans doing the work that technology should have started doing years ago.
However, the mood at the conference wasn't defeatist. There was a real, shared appetite for genuine transformation. The harder question is how we move from conversation to action. Three distinct challenges stood out for me, and I want to call out a fourth that few want to say out loud..
When a police officer makes an arrest, it sets off a chain of data right across the CJS. Most of the time, that data hits a wall at the agency boundary.
Evidence captured by policing doesn't flow cleanly to the CPS and vice versa back into victim and witness care. Disclosure is rebuilt by hand. HMPPS staff routinely recreate offender profiles from scratch. Intelligence sits in one system while the investigator who needs it is logged into another. We're asking brilliant, highly trained professionals to act as human routers between platforms that were never designed to talk to each other. And the cost of that doesn't just land on them; it lands on the victim who has to retell their story to four different people, on the witness who never gets the call about the new hearing date, on the defendant whose case takes another three months it didn't need to take.
Anyone who's run a police force knows the dread of a multi-year, multi-million-pound rip-and-replace IT programme. The good news is we don't need one. The technology to sit over legacy systems and join them up already exists, and it's been running at scale in other sectors for years.
Think about how a global bank tracks a single customer across cards, mortgages, savings, fraud alerts and call-centre conversations, in real time, with the right person seeing the right thing at the right moment. Think about how an airline can rebook two hundred passengers across three continents within minutes of a cancellation, sending each of them a personalised message on the channel they prefer. Think about how the largest retailers in the world know, to the second, where every order is in the supply chain, and who needs to know about it. None of that is magic. It's an intelligent data layer, sitting over many older systems, connecting them with role-based, compliant access. The same fabric, with the same discipline around open standards and interoperability, is what justice needs. And critically, it's what makes the harder work - investigations, intelligence, disclosure, pattern detection across force boundaries - actually possible. Forces shouldn't have to pick between joined-up data for safeguarding and joined-up data for investigations. It is the same problem, and it has the same solution.
The CJS is drowning in admin. Investigators, prosecutors, court clerks, and probation officers are losing huge chunks of their shifts chasing files, retyping the same information, and waiting for someone in another building to confirm something they already know.
Scheduling is the example I keep coming back to, because it's where the silo tax becomes most visible to the public. There was talk about prison transport as the bottleneck, and it is one. But the harder version of the problem is matching a court date to the availability of a victim who has childcare to arrange, a vulnerable witness who needs special measures, an interpreter, an officer in the case, a defence solicitor, a custody slot, and the transport van. Today, that coordination happens in inboxes and spreadsheets and phone calls. When it falls over, the people who pay are the victim and witness who took a day off work for nothing, and whose faith in the system takes another knock.
We have moved a long way past digital forms and static databases. Agentic AI can now take on real operational workflows, provided it sits on a platform you can actually trust to handle police and justice data.
The commercial parallels are already mature. Banks run AI agents that triage millions of fraud cases a day, escalating only the ones that need a human eye. Insurers use them to draft first-pass settlements within hours of a claim. Telecommunication companies use them to resolve customer issues across SMS, app and phone without the customer ever having to repeat themselves. Translate that into our world and you get an agent that reviews an incoming case file and flags missing evidence before it reaches the CPS; one that cross-checks court listings against the availability of victims, witnesses, transport and counsel, and flags a clash before it costs anyone a day; one that drafts the first version of a routine probation compliance report so the officer can spend their time on the parts that need a human judgement.
The point everyone in that room kept coming back to is that this only works if it is built responsibly. Does the model do what it claims? How does a human stay in the loop? Is it legal, proportionate and necessary? Those questions need to be answered in the open, with independent evaluation, and with the CJS agencies in the driving seat; not behind a vendor's curtain. That posture, of radical transparency and continuous testing, is one I'd argue policing should welcome rather than fear.
The real measure of a justice system is how it treats the people it exists to serve. And here I want to be direct: we cannot keep navel-gazing into our own efficiency problems when the thing that should be driving every conversation is trust and confidence.
Right now, too many victims and witnesses are bystanders in their own cases. They ring three agencies to find out a trial date. They get a letter about a hearing that's already been moved. They withdraw, because the system has worn them down. That's where attrition comes from, and it pulls vulnerable people out of the process altogether. The cost of that is not measured in pounds. It is measured in cases that don't get prosecuted, in offenders who don't get held to account, and in a public that gradually stops believing the system works.
Apply the principles modern organisations already use to keep their customers informed, and the picture changes quickly. Secure, automated updates over the channels people actually use - SMS, WhatsApp, a portal - tailored to what each victim and witness has asked for. A single, clear view of where their case is, what's next, who to contact. The same kind of clarity people now expect from their bank or their parcel tracker. The same kind of personalisation a healthcare provider gives a patient navigating a treatment plan.
This isn't a leap of faith. It's the kind of thing that's been done at scale across financial services, healthcare, retail and government for years, and bringing that cross-sector experience into policing and justice is one of the genuinely useful things a partner from outside this sector can do. Transparency builds trust, and trust is the bedrock of policing by consent and public confidence in the courts.
I'll address the elephant in the room, because it came up inside conversations all day. Technology is seen as expensive. Sometimes it is expensive. But the honest comparison is not technology cost versus zero. It is technology cost versus the price we are already paying for not having it.
Lost officer hours retyping data that already exists somewhere else. Court days collapsed because four diaries couldn't be reconciled. Cases dropped because a victim disengaged after the third unanswered phone call. Disclosure errors that end up in front of a judge. Civil claims and reputational damage when something falls through the cracks between agencies. Every one of those has a cost, and stacked together they dwarf the price of giving professionals the tools they need and citizens the service they deserve. If the conversation is only about the line item, we are having the wrong conversation.
The line that stuck with me from the conference was Justice can't wait. The pressure on police, courts, prisons and frontline staff is immediate. The pressure on victims, witnesses and the wider public is just as real, and a good deal less visible. They all deserve tools and a service that match what we are asking of them.
Stepping out of operational policing and into the technology sector, the thing that has struck me most in these short few weeks is that the answers to our hardest justice problems are not waiting on some future invention. The capability is here. It's running, today, in industries that long ago decided fragmented data and disconnected services were not acceptable. We don't need to reinvent the wheel - we just need to connect the spokes, in the open, alongside the people doing the work, and with the citizen kept firmly at the centre of why we are doing it at all.
The themes discussed at Modernising Criminal Justice reinforced a clear message: the challenges facing the justice system are interconnected, and solving them requires a more connected approach to data, processes and citizen engagement.
Join our upcoming webinar, Beyond Crisis: A Roadmap for UK Criminal Justice Transformation (25th June, 12:00pm), where we'll explore how connected data, intelligent automation and AI-powered engagement can help justice organisations improve victim outcomes, reduce operational burden and deliver measurable results - without replacing existing systems.