Boulder police will put AI on the non-emergency line, with human backup
Axon’s assistant will collect routine reports in English and Spanish. Callers can request a person, and staff will review every AI-generated report.
Loading page…
Axon’s assistant will collect routine reports in English and Spanish. Callers can request a person, and staff will review every AI-generated report.
Listen to this story
Boulder plans to use Axon’s Automated Non-Emergency Triage (ANET) to handle routine police calls, aiming to give dispatchers more capacity for incidents where seconds matter. The department says about 70% of its calls are non-emergencies. The schedule is not fully consistent across reports: CBS Colorado describes a soft launch by the end of October 2026 and full rollout by year-end, while Government Technology reported that all non-emergency calls would move to AI at the end of October. Human escalation and report review remain part of the plan.
ANET is designed for calls about issues such as noise, parking, abandoned vehicles, lost property and extra patrol requests.
The assistant is planned to speak English and Spanish; 911 calls are excluded, and callers can request a human dispatcher.
Dispatchers can review information ANET collects, and staff will review every report it generates.
Boulder police want AI to free dispatchers for emergencies, but the system is still heading toward deployment. The department plans a soft launch of Axon’s non-emergency call assistant by the end of October 2026, CBS Colorado reported October 12. A staged demonstration showed a human handoff when a routine complaint became potentially dangerous.
About 70% of the department’s calls are non-emergencies, according to Brad Riggin, Boulder’s safety communications manager. Police hope automating those conversations will ease staffing shortages and reduce the routine work competing for dispatchers’ attention.
The rollout has two milestones. CBS Colorado’s October 12 account puts the soft launch at month’s end and the expected full rollout at the end of 2026. An October 7 Daily Camera story, republished by Government Technology, described all non-emergency calls moving to AI at the end of October after weeks of phased testing.
What it's really going to do for us is let the dispatchers prioritize those calls where seconds matter, instead of being taken away from those calls by lesser priority calls.
Brad Riggin, Boulder safety communications manager, speaking to CBS Colorado
The system is called Automated Non-Emergency Triage, or ANET. Government Technology’s account describes an assistant that speaks English and Spanish and collects information through conversation. It is intended to replace a menu that asks callers to listen to options and press a number to reach the right destination.
In demonstrations, the voice identifies itself as Ashley and tells callers it is an automated assistant. Its planned remit includes abandoned vehicles, noise and parking complaints, lost or found property, and requests for extra patrols. Some categories have explicit limits: traffic crashes without injuries and animal complaints without a threat to public safety.
CBS Colorado tested a hypothetical wildlife concern. The assistant asked what the animal was doing, then transferred the call to a human on the emergency line as the scenario escalated to possible danger. That demonstrated the handoff in a staged exchange; it was not a result from the planned public rollout.
Police spokesperson Dionne Waugh said non-emergency calls will remain recorded and publicly accessible as they were under the existing system. She also said Axon will not use data from those calls to train the assistant. The no-training commitment does not change the department’s stated recording and public-access arrangements.
There is also a human route for complaints about the system itself. Callers can contact the non-emergency line and tell the assistant they want to lodge a complaint; it will transfer them to a dispatcher. Riggin described staff reactions as mixed, while saying some were excited about a tool that could reduce their workload.
Loading discussion...
Join the conversation
Explain whether human backup makes an AI-first public service acceptable.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.