For decades, local governments have spent millions of dollars building physical security. Thousands of cameras now watch over our parks, transit hubs, schools, and government buildings. Yet, when something goes wrong, the footage is almost always reviewed after the fact. We have built an expensive network of cameras that no one can watch in real time.
That is the quiet crisis facing public safety, and it sits beneath a more visible one. Communities are demanding a higher standard of security in the face of evolving threats. But the traditional playbook, hiring more personnel and buying more hardware, is no longer viable. Public sector leaders are caught between rising public expectations and shrinking operational realities: a dual crisis of talent and capital.
I talk to public safety leaders daily who feel trapped. The path forward does not require a massive influx of taxpayer dollars or an impossible recruiting surge. It requires making what agencies already own smarter. By applying Agentic Physical Security, an AI layer over existing infrastructure that continuously perceives, reasons, and acts, agencies can transform passive data into proactive protection: bridging the labor gap, maximizing past capital investments, and accelerating justice to build public trust.
Escalating Threats Meet Starved Resources
According to a recent national survey by the Police Executive Research Forum (PERF), agencies across the nation are fighting an uphill battle. Total law enforcement staffing remains below pre-2020 baselines, and a sudden 16.4 percent surge in retirements in 2025 has forced departments to heavily rely on overtime to bridge gaps. Research from Oregon State University details that municipal and state budgets are similarly strained, leaving little room for expenditures as rising costs and inflation force local governments to prioritize immediate operational needs over long-term security infrastructure upgrades.
This creates a dangerous vulnerability. Because there are not enough eyes to monitor these feeds, most video goes unwatched in real time. The National Institute of Justice found that humans lose about 95 percent of their attention on video monitors after 20 minutes. It is a human limitation, at a scale no hiring surge can close.
Making the Most of Existing Infrastructure
The solution is not to buy more cameras or wait for a hiring miracle. This is the shift from surveillance to Agentic Physical Security: an AI layer applied directly onto existing camera networks that does not just detect motion but understands what it is seeing and acts. Detection still leaves a human to interpret every alert. Reasoning is what changes the equation. AI that continuously perceives, reasons, and acts across every feed can tell the difference between a maintenance worker and an intruder and escalate only what is real.
Don’t rip and replace. Retrofit the infrastructure agencies already own, preserving those infrastructure investments while lowering total cost of ownership. It is the rare answer that addresses the talent crisis and the capital crisis at once.
This acts as a force multiplier for existing staff. Instead of asking a single operator to watch 50 screens at once, AI continuously reviews every live feed for genuine risks: a brandished weapon, a forced entry, an unauthorized perimeter breach. When a threat is verified, the system alerts staff so responders can intercept a crisis as it unfolds, rather than discovering it hours late on a recording. The human stays in the loop, not in the bottleneck: people make the decisions that matter, while AI handles the impossible task of watching everything at once. In effect, understaffed departments gain the reach of a fully staffed operations center, using the exact infrastructure they already own.
Accelerating Forensics to Rebuild and Solidify Public Trust
When a major security incident does occur in a public space, the clock starts ticking. Traditionally, a major investigation forces detectives to pull dozens of hard drives and spend days manually watching hundreds of hours of video footage to piece together a timeline. This agonizingly slow process delays justice and stalls the release of accurate information to a demanding public.
Artificial intelligence fundamentally changes the physics of forensic investigations. With AI-powered forensics, investigators can instantly query days of recorded video across an entire city or campus using simple keyword searches, clothing descriptions, license plates, or behavioral patterns. A search that used to take an entire task force a week can now be completed by a single detective in minutes.
This acceleration is vital for building and maintaining public trust. When law enforcement can rapidly solve cases, locate suspects, and present undeniable, visually verified evidence, it brings swift closure to victims and removes threats from the street. Furthermore, it allows public information officers to share precise, factual data with the community almost immediately, cutting through rumors and misinformation. Ultimately, utilizing AI in the public sector isn’t just about operational efficiency; it is about delivering the transparency, accountability, and swift justice that our communities rightfully deserve.
The author, Shikhar Shrestha, is Co-Founder and CEO Ambient.ai.