AI in public services: enhancement, not replacement

Guest blog by Kamila Hankiewicz, CEO and co-founder at Untrite

We're automating bureaucracy instead of eliminating it. And calling it progress. I've spent the last five years watching AI transform emergency call handling in UK police forces. The most telling metric? Operators disagreed with the AI less than 13% of the time. That's not a failure rate, but rather the entire point.

The automation delusion

Walk into any force control room and you'll see call handlers drowning—transcribing, typing, searching multiple systems, assessing risk, all while someone's reporting a crime. The average handler captures maybe 40-70% of what callers actually say. The rest vanishes into the void of cognitive overload.

The obvious fix seems simple: automate everything. Let AI take the calls. Remove the human "bottleneck." Except the bottleneck isn't the human. It's the mountain of paperwork we've buried them under.

Working with Humberside Police, we learned something the automation-first crowd keeps missing. Give operators AI that handles transcription, data entry, and system searches in real-time, and something shifts. They stop frantically typing and start actually listening. Call handling times dropped 29%. But what matters more: operators reported feeling less stressed and more confident—one called it "a safety net." They could focus on empathy, judgment, the messy human nuances no algorithm can parse.

The NHS Crisis nobody wants to talk about

The NHS 10 Year Health Plan commits to becoming "the most AI-enabled health system globally by 2035." Ambitious. Also possibly delusional if we keep thinking AI means "replace expensive humans with cheap algorithms."

Consider the numbers: 935,950 emergency calls monthly. 8.9% vacancy rates. 42% of staff report work-related stress. 29% frequently consider leaving. The NHS wants to recruit its way out of this. Even with 23.7% staffing increases since 2019, the system barely keeps up when demand is growing 37% and your workforce is burning out.

The Plan requires 2% annual productivity improvements. Traditional efficiency drives mean "work faster, see more patients, take fewer breaks." That's not productivity—that's exploitation with a dashboard.

What if productivity meant something different? Call handlers spending zero minutes on transcription because AI handles it. Instant access to patient records without toggling between six systems. Applied to NHS emergency services, that 29% efficiency gain translates to capacity for 271,000 additional calls monthly. No recruitment. No retention crisis. Just removing the administrative sludge that stops healthcare professionals from doing healthcare.

What actually works

Three principles from deploying AI where mistakes have consequences:

Design with removal in mind. Every feature should eliminate a process, not accelerate it. Our operators spend zero time on manual transcription now. That task didn't get faster—it vanished.
Preserve human override. In emergency services, where context is everything, this balance isn't negotiable. The day an algorithm starts making life-or-death decisions without human oversight is the day we've lost the plot.
Measure what disappears. Celebrate approvals operators no longer need, forms they don't fill, searches they never run manually. Productivity isn't doing bureaucracy faster—it's doing less of it.

Technology isn't the hard part. We've proven AI can handle 90-95% transcription accuracy even with regional accents and emotional stress. The challenge is earning trust from frontline staff who've watched too many "transformative" IT projects crash spectacularly.

The choice ahead

AI in emergency services isn't about replacing dispatchers or automating triage. It's about giving frontline workers superpowers—perfect recall, instant access to relevant information, real-time decision support—while keeping human judgment at the center.
The question for public services isn't whether to adopt AI. It's whether we'll use it to reduce bureaucracy or just make bureaucracy easier to tolerate.
I know which future we're building. And it's the one where the 999 operator actually has time to care about the person on the other end of the line.

Kamila Hankiewicz is CEO of Untrite, building decision intelligence platforms for emergency services and high-risk, time-sensitive complex enterprise operational workflows.

Health and Social Care Programme activities

techUK is helping its members navigate the complex space of digital health in the UK to ensure our NHS and social care sector is prepared for the challenges of the future. We help validate new ideas and build impactful strategies, ultimately ensuring that members are market-ready. Visit the programme page here.