AI Agents will transform operations and the human role in security, says Rahul Yadav, Chief Technical Officer, Milestone Systems.

The transformation

As a technology leader who has spent years working at the intersection of AI and video security, I’ve witnessed numerous transformative shifts in our industry.

But none quite compared to what we’re seeing emerge for 2025.

We are standing at the threshold of fundamental changes that will reshape not just how we think about security technology, but how we interact with AI across every industry sector.

The convergence of advanced AI capabilities with practical applications is creating unprecedented opportunities for innovation and efficiency.

The Era of Agentics

The most significant shift ahead is what’s becoming known as the Era of Agentics.

Unlike traditional AI systems that follow prescribed steps, AI Agents are autonomous systems capable of understanding contexts, making decisions and taking actions independently.

These Agents – similar to but far more sophisticated than today’s chatbots – use generative, training-based approaches rather than deterministic programming.

By 2025, we’ll see these Agents emerging across different products and services, from video analytics to automated security responses.

Think of AI Agents as digital colleagues that can handle complex tasks without constant human direction.

They can wait for specific conditions, respond to prompts or act on their own initiative when they detect relevant situations.

Most importantly, they learn from their actions and adapt to new scenarios, much like human operators do.

In security applications, this means systems that can automatically identify potential threats, coordinate responses and even predict incidents before they occur.

The real power of these Agents lies in their ability to reason and adapt.

Unlike traditional software that needs explicit programming for every scenario, these systems can understand context and make nuanced decisions.

This capability will transform everything from access control to emergency response, creating more intelligent and responsive security environments.

Beyond thinking: The age of AI that acts

In the evolution of AI, we’re witnessing a pivotal shift from systems that merely analyze to those that take decisive action.

While traditional metrics like IQ measure cognitive ability and EQ gauges emotional awareness, a new capability is emerging: the power to act intelligently and autonomously – AQ (Action Quotient).

Think of Tesla’s self-driving cars, which don’t just process road conditions, they smoothly navigate complex traffic scenarios in real time.

This shift toward action intelligence is particularly relevant in security operations.

Traditional monitoring systems alert operators to potential issues, requiring human intervention for every response.

In contrast, high-AQ systems can assess situations, initiate appropriate responses and adjust their actions based on changing conditions.

This capability will transform how we approach security management, making systems more proactive and less dependent on constant human oversight.

The implications extend far beyond simple automation.

These systems will be able to coordinate complex responses across multiple subsystems, from access control to emergency communications, creating more comprehensive and effective security solutions.

The key is that these actions aren’t just pre-programmed responses – they’re intelligent decisions based on real-time analysis and learned patterns.

The human element

Despite these technological advances, human roles aren’t disappearing, they’re evolving.

As Microsoft’s CEO aptly noted: “It’s not AI that will replace you, but someone using AI who will.”

Success in 2025 and beyond will depend on how effectively we learn to work alongside these AI systems, using them to augment our capabilities rather than replace them entirely.

Consider how coding has evolved: today, even young students can create sophisticated programs using AI-assisted tools.

This democratization of technology doesn’t eliminate the need for human expertise; instead, it elevates our role from routine tasks to higher-level decision-making and oversight.

Security professionals will need to develop new skills focused on managing and directing AI systems rather than performing routine monitoring tasks.

The key to success will be in learning to work with AI as partners, not tools, because the better we collaborate, the smarter and faster we all become.

Humans excel at understanding context, making nuanced judgments and handling unexpected situations – skills that will become even more valuable as routine tasks become automated.

The evolution of AI models

The landscape of AI is becoming more sophisticated and specialized.

We’re seeing the emergence of three crucial model types: Small Language Models (SLMs) for specific applications, Vision Language Models (VLMs) specifically designed for video processing and Large Multimodal Models (LMMs) that can handle multiple types of data simultaneously.

This evolution represents a shift from traditional analytics to more comprehensive, learning-based systems.

These models don’t just follow pre-programmed rules, they learn from each incident and improve their responses over time.

This is particularly crucial for smart city applications, where systems need to process and understand multiple data types simultaneously.

This evolution in AI models is driving a parallel shift in computing infrastructure.

We’re moving from traditional CPU-based processing to GPU-focused architectures, fundamentally changing how we approach system design and programming.

While major tech companies invest hundreds of millions in training large base models, security companies can leverage these foundations to create specialized applications with more modest hardware investments.

A mid-sized security operation can now establish effective AI capabilities with an investment of $200,000 to $300,000 in GPU infrastructure, which is a fraction of what was required just a few years ago.

This democratization of AI capabilities means that even smaller security organizations can begin implementing sophisticated AI-driven solutions, though they’ll need to carefully consider their specific needs and use cases to make the most effective use of these resources.

What makes this development particularly significant is the increasing accessibility of these technologies.

While training large-scale models remains resource-intensive, organizations can now leverage pre-trained models for specific applications, making advanced AI capabilities more accessible to a wider range of users.

This democratization of AI technology will accelerate innovation across the security industry and spur on the development of AI Agents.

Responsible innovation

Looking ahead to 2025, responsible technology development will become a crucial competitive advantage.

However, this doesn’t mean stifling innovation with excessive caution.

The key is finding the right balance, taking calculated risks while maintaining ethical standards and user trust.

For US companies, this means staying ahead of innovation curves while building trust with users and stakeholders.

Just as consumers choose trusted brands for their smartphones and personal devices, organizations will increasingly select security technology partners based on their track record of responsible innovation and ethical AI deployment.

The challenge lies in maintaining this balance while keeping pace with rapid technological advancement.

Think about it: would you trust a self-driving car made by a company with a sketchy reputation?

The same principle applies to security tech – ethics and trust aren’t just nice-to-haves; they’re deal breakers.

This requires developing clear frameworks for AI governance while maintaining the flexibility to adapt to new technologies and use cases.

Great data makes great AI

In this landscape of emerging technologies, it is important to emphasize one fundamental truth: “Great AI requires great data.”

Organizations that have invested in data quality are already seeing accelerated benefits from their AI initiatives, while those lacking robust data infrastructure risk falling behind.

In 2025, the focus on data quality will become even more critical as synthetic data and accelerated computing push the boundaries of what’s possible with AI.

The convergence of these trends in 2025 promises to usher in a new era of AI capabilities, where success will depend not just on adopting the latest technologies, but on building solid foundations in data quality and governance.

Improving data will always give you a better competitive advantage in the market.

The future of video management

The video management landscape is undergoing its own transformation.

Traditional video management systems (VMS) are evolving from passive recording and playback tools into intelligent platforms that can automate complex workflows and security responses.

This shift will fundamentally change how organizations approach security operations.

Security centers that once required large teams of operators will become streamlined, AI-augmented environments where human expertise is focused on high-level decision-making and complex situations.

Routine tasks like event management and incident reporting will be largely automated, with AI Agents handling initial assessments and responses.

This evolution doesn’t mean complete automation, rather, it represents a more efficient partnership between human operators and AI systems.

The key is finding the right balance where technology handles routine tasks while human operators focus on situations requiring judgment, empathy and complex decision-making.

This transformation will require new approaches to training and workforce development, as security professionals adapt to roles that emphasize system management and strategic oversight rather than routine monitoring.

The security industry stands at a pivotal moment.

The technologies we’re developing today will shape not just how we approach security challenges, but how we think about the relationship between human operators and AI systems.

By embracing these changes while maintaining our commitment to responsible innovation, we can create security solutions that are more effective, more intelligent and more responsive to the complex challenges of tomorrow.