For decades, contract security has been built around a deceptively simple transaction: a client needs a guard, a provider supplies one, and someone gets paid for the hours worked. But as enterprise security programs have become more distributed and data-dependent, the limitations of that transaction are becoming harder to ignore.

Protos Security has spent two decades challenging that model. In a March 2026 SIW feature, CEO Mark Hjelle described the company’s approach as a deliberate departure from the traditional labor-heavy guarding business, one based instead on orchestrating a national supplier network through software, data and managed services. The premise is straightforward: if security services are increasingly about outcomes, the technology behind those services must coordinate the people, providers, pricing, coverage, and information required to deliver those outcomes.

That strategy has continued to evolve. Protos introduced an AI-powered operational agent in March designed to surface open posts, analyze fulfillment trends, and identify performance or billing discrepancies. The company says its platform now supports more than 25,000 annual service locations through a network of more than 2,800 vendor branch locations and 60,000 off-duty law enforcement professionals.

But the more revealing story may be what's underneath the AI.

That distinction matters. Technology may be dramatically different, but the operating challenge remains the same: get the right security resource to the right place, at the right time, at the right price, and be able to prove that it happened.

From Software Application to Operating Platform

The first major shift was moving the business out of the laptop and onto the web.

The old approach was not designed for a geographically dispersed operation. Installing an application on a laptop and requiring users to connect through firewalls from hotel rooms was hardly a foundation for a national security-services network. The COVID-19 pandemic accelerated the move to cloud and web-based operations.

“When we had to distribute our operations center, with people working from home and not able to get into the same operations center to run the business,” Tabor says, “that piece of the technology of just putting it into the cloud, web enabling all of it, that was a fundamental piece that we had to do.”

Clients themselves were changing. A decade ago, a security customer might have been satisfied with proof that a guard showed up, consolidated billing and the ability to replace a missing officer. Today, enterprise customers increasingly want to understand what happened at individual locations, how incidents are trending, how coverage is performing and whether pricing reflects the realities of individual markets.

“They want to look at the incidents that are occurring; they want to look at trends, not only in security coverage but of what’s happening at individual locations,” Tabor says. “They want more sophistication in other data feeds that are out there, crime statistics, etc., as well as dynamic pricing.”

Security labor is not priced uniformly across geography, and a national account cannot necessarily be managed intelligently with a single rate structure. The technology challenge becomes one of granularity: capturing enough information about local supply, demand and performance to make national programs behave like locally informed operations.

The Network is the Product

This is where Protos’ distributed provider model becomes more than a sourcing strategy. It has become a technology problem.

“Cultivating that network of vendors is a real challenge,” Tabor says, noting that Protos maintains a vendor relationship management function that treats its provider supply chain much like a customer relationship.

A national retailer may ask for identical coverage at hundreds of locations, but the underlying economics can be radically different from one market to another. In one city, qualified personnel may be abundant. In another, coverage may be sparse or demand may exceed available supply.

“It can’t be a one-size-fits-all to be able to pay and to bill correctly for a national customer,” Tabor says. “We have to get granular about that, and we have to be able to track that, trend that, and adapt to changing market needs.”

Technology can consolidate information. The harder part is knowing what information means.

“It comes down to the software helping us consolidate it all into a data mart and query it, slice and dice it however we need to,” Tabor explains. “The hard part is the research that’s required in the local markets to understand what’s going on.”

And that provider network has another economic dimension. This solution can give smaller security businesses access to national accounts that they might otherwise never touch. Rather than requiring a national retailer to maintain relationships with dozens of local vendors, the network aggregates those providers behind a single managed-services relationship.

“We take pride in helping small businesses get access to national accounts,” Tabor says. “These other vendors that perform for us would never have access to a nationwide retailer or distribution centers spread across the nation.”

One Platform, Different Businesses

That model becomes more complicated as Protos expands into off-duty law enforcement, specialized services and remote monitoring. Different service lines have different operating requirements.

Tabor is blunt about the answer.

“A common technology platform cannot bring those businesses together without customization,” he says.

Protos Labs exists in part to solve that problem, working with acquired businesses to understand their operating models and adapt the platform around shared fundamentals such as time and attendance, tracking and incident reporting.

“There is a common thread of time and attendance. You’re punching in, you’re punching out, you’re reporting incidents, but it kind of stops there when you get into these different areas,” Tabor says.

For Tabor, the solution is agility. “How can we modify the platform?” he asks.

The objective is to preserve the specialized capabilities of different business lines while giving the enterprise customer a common experience. That can mean one relationship, one interface and potentially one invoice across different types of security services.

From Dashboards to Decisions

That same philosophy is reshaping the client dashboard. Security companies have offered dashboards for years. But displaying information is not necessarily the same as producing intelligence.

Tabor points to a familiar frustration: a client clicks from national to regional to local views, drills down through multiple screens and eventually discovers that the dashboard still does not answer the question being asked.

“We’ve all probably experienced being given a set of clickable dashboards where you can drill down, drill down, drill down,” he says. “But they’re not dynamic in the way that you might want them to be.”

The user may want to pivot the data around service lines rather than geography, or around incident types rather than locations. In many cases, the answer has been to export the data and manipulate it manually.

“Can my people get actionable insights?” he asks. “It’s kind of like a binary yes or no, and that’s an easy one to knock down.”

Consider an executive preparing an operational review. Previously, that person might have needed to download an incident report, a time-and-attendance report and a financial invoice, then combine the information to create a useful picture.

“If you can do that with Agentic Help or AI Help to be able to cultivate that report the way you need to,” Tabor says, “and you can just get it with the press of a button or subscribe to a feed where that’s sent to you, I mean, there’s real measurable results there.”

For Tabor, the immediate AI payoff is not science fiction. It is time.

“At this point, I think that’s kind of where we are,” he says. “You can save time in your profession, your field, with your field people who are monitoring what’s happening and trying to take action based on that.”

That philosophy is reflected in the Protos operational assistant introduced this year. Unlike a conventional chatbot that routes users toward pre-approved answers, the assistant has access to the same fundamental data available through the client portal and uses large language models to infer what the user is asking.

A conventional chatbot can leave users feeling as though they have simply reached an existing report through a different interface. The operational assistant intends to interpret the question and navigate the underlying data.

But that flexibility creates a new challenge: trust.

“Our challenge is to make sure those inferences are shaped by the right values and the right things we know our client wants to see,” Tabor says, “without adding so many guardrails that it feels like all you did was get the same report by typing a question instead.”

AI Informs. Humans Decide.

For physical security, that line matters because operational decisions can carry real-world consequences.

Tabor’s formula is simple.

“AI informs, and humans decide.”

AI can identify patterns, assemble information, recommend actions and surface anomalies. That does not mean it should independently decide how a security operation responds.

“There will be a day when we have such high-quality data and such high-quality metrics that actions can be taken autonomously,” Tabor says. “But I don’t think we’re quite there.”

For now, the better application is to compress the distance between information and judgment. AI can say: here is what the data shows; here is the recommendation; here is why the recommendation makes sense. The security professional remains accountable for the decision.

The ROI Question is Getting Simpler

For security executives evaluating AI and operational technology, Tabor argues that the first ROI calculation does not have to be complicated.

Start with the baseline.

How long does it take today to get an actionable answer? How many reports have to be assembled? How many dashboards have to be navigated? How much time does an operations team spend reconciling information that already exists inside the organization?

“Security executives should be looking at what that baseline time to actionable insights is, and whether I can reduce it significantly,” Tabor says.

There is also an unavoidable reality in a rapidly changing technology market.

“A certain amount of a leap of faith” is necessary, Tabor acknowledges, because organizations do not want to be left behind. But the counterweight is discipline: do not spend so aggressively on emerging technology that the economics of the underlying security service suffer.

By that measure, the best technology investment is the one that measurably reduces friction.

What Comes Next

Asked what technology shift could most change the delivery of physical security services over the next three to five years, Tabor offers an answer that begins well before humanoid robots.

The near-term opportunity is making the existing security ecosystem more seamless,  reducing the friction involved in matching a client need with a qualified provider at an appropriate rate.

More information can make those matches smarter: certifications, qualifications, local market conditions, provider performance and the identity of the person actually assigned to the post.

“We’re really doubling down in Protos on cultivating that full ecosystem and depth of knowledge to make those connections quicker,” Tabor says.

Beyond that lies the more futuristic end of the spectrum: drones, AI-monitored video and autonomous security robots.

“I think we’re still probably a few years from that being a reality,” he says, “but it has some interesting possibilities.”

That may be the most useful way to view the next phase of security-services technology. The industry does not necessarily need to leap from human guards to autonomous machines. It needs to eliminate friction between every component of the existing system—clients, providers, dispatchers, field personnel, data, billing, and decision-makers—and then use AI to make the entire chain more intelligent.

Protos’ evolution suggests that the real competitive advantage may not come from any single piece of technology. It may come from the operating platform that makes all those pieces work together.

The original problem was matching a security need with a provider. Twenty years later, the question is no longer simply whether that match can be made. It is whether technology can make the match faster, smarter, more transparent, more economical, and ultimately more accountable.

That is a much bigger software problem. And, increasingly, it is becoming a security problem, too.