Law Enforcement Agencies Turn to AI for Non-Emergency Call Management

Key takeaways

Hampshire and Isle of Wight Constabulary and Thames Valley Police launched "Bobbi," the UK's first AI-powered virtual police assistant, on November 26, 2025.

The system operates 24/7 on both force websites to handle frequently asked non-emergency questions.

Chief Superintendent Simon Dodds, head of contact management for both forces, described the launch as addressing a critical operational need.

"Locally and nationally, contact with the police increases year-on-year, online and over the phone, and it is important that the valuable skillset of our call handlers and digital desk operators is spent on emergencies and complex, sensitive issues," Dodds said.

The AI assistant uses natural language processing technology and follows the same guidelines available to human call handlers.

Bobbi successfully resolved 65% of general enquiries without human intervention during testing, according to operational data.

Thames Valley Police serves over 2.5 million residents, and redirecting even 10% of daily contact to AI could save thousands of staff hours annually.

A spokesperson for the forces emphasized that Bobbi cannot replace emergency services. Users cannot report crimes through the system or use it as an alternative to the 999 emergency line.

The technology was tested by over 200 people during development, including representatives from victim care groups and independent scrutiny panel members.

Canadian police deploy the SARA system

Halton Regional Police Service in Ontario launched an AI system called SARA (Smart Answering Routing Assistant) on November 10, 2025, operating 24/7 after months of testing.

Winnipeg Police Service began its pilot project in the summer of 2025, initially operating several hours daily before expanding to full-time use.

Jeff Hill, Deputy Chief of the Halton Regional Police Service, highlighted the system's potential impact on operations.

"We're not only confident in SARA's performance, but we're proud to be adopting such reliable and cutting-edge technology that will help us reduce the number of non-emergency calls our Communicators spend time on, which at the moment is approximately 160,000 per year," Hill said.

SARA asks callers short questions to determine whether the system can handle their inquiry or requires transfer to a human communicator.

The system immediately transfers calls to human operators if situations evolve into emergencies, the system cannot understand the caller, or translation services are needed.

Ben Sanders, Hyper's founder and chief executive officer, stated that the company developed SARA specifically for public safety communications.

"Halton is proving what's possible when public safety embraces innovation. With their new AI agent, SARA, residents get faster service, Communicators get more time for emergency calls, and together, we can make the community safer," Sanders said.

Chris Lewis, a former Ontario Provincial Police commissioner and CTV public safety analyst, defended the technology's cost-effectiveness.

"It's unbelievable how many calls police get that have nothing to do with active police work," Lewis said. "Of the thousands of legitimate calls they get every day, only certain ones really require an emergency response."

Utah implements statewide AI deployment

The Utah Communications Authority announced in July 2025 that all 27 public safety answering points throughout the state would implement Motorola Solutions' Virtual Response technology by year's end. The system automates receipt and resolution of non-emergency line calls using AI.

Tina Mathieu, executive director of the Utah Communications Authority, explained the operational pressure driving the deployment.

"An estimated 65% of Utah's 911 PSAPs' workloads are non-emergency line calls, like noise or parking complaints, that aren't a true crisis," Mathieu said. "When a 911 call comes in, they have to put the non-emergency caller on hold, taking precious seconds away from responding to the real emergency."

The AI technology allows callers to ask questions and receive answers in their native language through voice or text.

Last year, Utah's 27 emergency centers processed nearly 900,000 emergency calls and over 2 million non-emergency line calls.

Kevin Rose, executive director of the Weber Area Dispatch 911 & Emergency Services District, emphasized the technology's role in workforce management.

"There are some calls where the caller just wants a quick answer—it's not a complex emergency requiring a multi-pronged response," Rose said. "AI can help take stress off our call handlers and free them to focus on the calls that really need their human empathy and expertise."

Addressing operational challenges

The deployments respond to long-standing operational challenges.

In 2024, His Majesty's Inspectorate of Constabulary and Fire & Rescue Services reported that nationally across the UK, roughly 15% to 20% of calls to the non-emergency 101 line are abandoned by callers who waited too long, with rates spiking higher during peak summer months.

Hampshire and Thames Valley Police reported that strategic shifts, including automation, helped reduce average wait times from 13 minutes in May 2023 to 1 minute and 30 seconds in May 2025, according to the Hampshire Police and Crime Commissioner Annual Report 2025.

A call handler for Thames Valley Police with six years of experience, identified only as Sarah, described the mental strain of managing diverse call types.

"You go from a call about a parking dispute to a call about a cardiac arrest, then back to a query about a lost dog. It's exhausting," Sarah said. "If Bobbi can take away the 50 calls a day I get about 'where is the nearest police station,' I can actually focus on the domestic abuse victim who needs me to listen patiently."

Read more: