Don Morron, Podcast Host of AI PHYSEC TODAY, explores the role that AI Agents will play in the physical security industry.
Some people still think generative AI is just good for creating poetry and pictures โ well, someone should tell that to the tech giants who dumped nearly $150 billion into AI in the first half of 2024 alone, according to Forbes.
Meanwhile, predictions are rolling in that AI will have a multi-trillion-dollar economic impact around the globe, as reported by IOFund.
Industry experts say that the real payoff on all this investment will come from a new breakthrough known as AI Agents (Forbes).
In other words, buckle up, because 2025 is about to be the year of AI Agents.
Why AI Agents and why now? Leading technology companies โ Meta, Salesforce, Microsoft, Amazon, Google and plenty more โ believe that AI Agents are on track to exceed the total number of people in the world (Scientific American).
Soon, every white-collar worker will have multiple AI Agents working on their behalf, each taking on tasks that once required human hours or, at best, time-consuming automations.
However, the promise of Agents isnโt just about reducing repetitive workloads; itโs about achieving a new level of reasoning, memory and action in AI systems thatโs poised to redefine how businesses, including physical security end-users, integrators and manufacturers, serve their customers.
Better known in the AI thought leadership space simply as โAgents,โ an Agent is a system or program capable of autonomously performing tasks on behalf of a user (or another system) by intelligently designing workflows and utilizing available tools (IBM).
Thatโs a bit abstract, so letโs break it down in terms of the big three capabilities almost every AI leader agrees on: reasoning, memory and actions.
Yes, Agents really do โthinkโ โ or at least, they employ something very similar to what we humans consider thinking.
Within the AI community, this is referred to as โchain of thought reasoning.โ
The latest frontier models, such as OpenAIโs GPT-o1 & o3 โ often built on hundreds of billions, if not a trillion, parameters โ can now break down problems internally, understand context, interpret user intentions and solve complexities by dividing them into smaller subtasks.
This is a huge shift from previous generations of AI models that essentially just predicted the next word in a sentence.
Those older models could produce coherent-sounding text but were prone to โhallucinations,โ a term the AI community uses for highly confident yet incorrect answers.
Weโre not talking about typical computer storage here but rather an AIโs ability to remember, recall and use past interactions or data in meaningful ways.
Memory in this sense underpins the hyper-personalization that Agents promise to deliver.
Imagine an AI assistant that deeply understands your preferences, your workflow, your communication style and your history of interactions โ so much so that every new task you throw at it becomes quicker, more precise and more relevant.
Weโre not satisfied with just storing data; the real goal is for these Agents to recall and leverage user signals to deliver seamless experiences.
Finally, we come to the secret sauce of Agents: their ability to โactโ on the userโs behalf.
Weโve all interacted with website chatbots and while they sometimes offer decent information, the scope of their help is typically limited.
An Agent, however, can take that conversation and turn it into actual workflow automation.
Hereโs a typical scenario:
Rather than merely chatting, the Agent is executing real actions โ pulling up records, creating new ones and integrating with business systems like ticketing or CRM platforms.
For those of us in physical security โ be it end-users, integrators or manufacturers โ AI Agents hold equally transformative potential within the security products we use.
Think about how an AI Agent might ingest live video streams (possibly thousands, from multiple facilities), apply Large Vision Models (LVMs) to produce a daily AI-generated report that highlights anomalies or important events across these countless video feeds, leaving out the noise of nuisance alarms and unusable video so humans can focus on what matters most.
Imagine dynamic user interfaces that adapt to each operatorโs login, reflecting that individualโs usage history.
A new hire who rarely touches certain features wonโt be bombarded with complicated dashboards, while a veteran user with advanced needs might see precisely the tools they rely on most โ no more, no less.
Right now, the AI industry widely believes that Agents will be the core technology for the next several years, shaping the landscape until yet another breakthrough โ possibly the commercial adoption of quantum computers โ changes the rules again.
For now, though, the hype (and the funding) around AI Agents is centered on the idea that theyโll lead to โhyper-personalization,โ potentially knowing you better than you know yourself, Deloitte says.
Hyper-personalization, in many ways, is the holy grail of next-generation AI.
From Apple to Google, and from small startups to tech giants, everyone wants to be the first to deliver an Agent thatโs part personal butler, part psychic and part wise advisor.
So how does all of this apply to physical security professionals โ particularly end-users, integrators and manufacturers?
For starters, physical security is a relationships and trust-based industry. Indeed, we like to say we protect โpeople, property and profits.โ
Agents hold a promising future in each of those areas because they empower us to enhance our service and solutions in ways that were previously impossible.
Consider, for instance, how an integrator could use an Agent to optimize the deployment of new security technology across multiple client sites.
The AI Agent could automatically track every piece of hardware (cameras, sensors, access control points), note which ones might be due for a firmware update and even schedule on-site visits based on technician availability and location.
It might cross-reference real-time data from support tickets to identify which cameras fail the most or which doors have the highest false alarms.
By taking care of these coordinating tasks, the AI Agent frees up your human team to do what they do best โ build trust with clients and ensure systems are installed and maintained properly.
On the manufacturing side, imagine streamlining your entire supply chain.
An Agent could interface with factories, shipping providers and retailers all at once, adjusting inventory levels in real time based on demand fluctuations.
Meanwhile, your sales and support teams can dedicate their time to nurturing relationships and closing deals, rather than juggling supply chain details.
In short, AI Agents will help us become better at the things only we can do โ building relationships, innovating new products and focusing on creative problem-solving โ while the behind-the-scenes complexities are handled by AI.
This is the vision that so many in the AI community see unfolding in 2025 and beyond.
Nobody has a perfect crystal ball, but the AI industryโs consensus is that Agents are about to kick off a hyper-productivity era in multiple sectors, from customer support and back-office tasks all the way to on-site operations.
This shift might feel sudden, but itโs the natural outcome of nearly $1 trillion in current and future investments (Goldman Sachs), years of research and lightning-fast iteration in AI models.
And the best part? We, in the physical security industry, get to harness these evolving capabilities for one of humanityโs most essential needs: safety.
When we integrate Agents that can reason, remember and take decisive action on our behalf, we enrich our ability to protect people, property and profits.
We also free ourselves to do the work that requires human insight โ nurturing relationships, building trust and solving real-world problems.
The next few years will almost certainly be a time of trial, error and growth.
But as anyone whoโs been around AI in the past decade knows, itโs worth the ride.
By the time 2027 rolls around, we may look back and wonder how we ever managed physical security without the help of a few well-trained Agents working by our side.
Just imagine what tomorrowโs Agents โ armed with advanced reasoning, deep memory and the power to act โ will do for the security industry and beyond.
Don Morron is a recognized authority on the transformative role of AI in physical security.
As host of AI PHYSEC TODAY, a pioneering podcast on AIโs impact in this space, he has published 19 episodes and attracted hundreds of followers eager for insights into evolving trends, cutting-edge applications and the future of AI in security.
Through his new venture, HighlandTech, Don builds Agents that automate business workflows โ serving end-users, integrators and manufacturers alike.
This article was originally published in the special February Influencers Edition of Security Journal Americas. To read your FREE digital edition,ย click here.
