La Mesa PD AI Assistant: Non-Emergency Call Automation & CX

What happened

The La Mesa Police Department in California has deployed an AI-powered virtual assistant to handle non-emergency calls from the public. The system is designed to manage routine enquiries — such as reporting minor incidents, requesting information, or filing non-urgent reports — without routing callers to a live dispatcher.

The deployment represents a deliberate effort to reduce the volume of low-priority calls occupying human operators, freeing sworn staff and civilian dispatchers to focus on time-critical emergencies. The AI assistant is intended to be available around the clock, providing residents with a consistent point of contact regardless of staffing levels or peak-demand periods.

Why it matters

Public-safety communications sit at one of the most emotionally charged intersections of government service and citizen experience. When a resident reaches out to a police department — even for a non-emergency — they are often anxious, frustrated or confused. Deploying an AI at that moment is not merely an operational efficiency decision; it is a deliberate choice about what kind of experience the institution wants to deliver at a vulnerable touchpoint. The behavioural stakes are high: perceived indifference or robotic deflection can erode institutional trust far more durably than a slow response time.

For service designers and CX practitioners, this case illustrates a tension that is becoming increasingly common across both public and private sectors — the trade-off between scalability and emotional resonance. Automating non-emergency intake can genuinely improve outcomes if the system is well-calibrated, but it demands rigorous journey mapping to ensure callers never feel abandoned, misrouted or dismissed.

The Renascence take

Most commentary on police AI deployments fixates on surveillance or bias. What gets missed is the quieter, more consequential design question: who decides what counts as "non-emergency" — the institution or the caller? That asymmetry is where trust is won or lost.

From a behavioural standpoint, the moment a resident is redirected to an automated system, their internal narrative shifts from "I am being helped" to "I am being managed." La Mesa's real design challenge is not the AI's accuracy — it is preserving the caller's sense of agency and acknowledgement throughout the interaction. A customer-obsessed operator in any sector deploying similar triage automation should invest as heavily in the emotional architecture of the conversation flow as in its technical capability. Specifically: always give the caller a clear, frictionless path to a human, and never let the system be the one to decide the matter is unimportant.