Thought Leadership by Leandro da Cunha, Surveillance BU Executive, Duxbury Networking

For a long time, surveillance was measured by how much footage a business could capture. The weakness in that model becomes clear after an incident, when the question is no longer whether the video was recorded, but whether anyone can find the right moment quickly enough for it to matter. A business may have hundreds of hours of footage available, but that does not mean the system is helping people make better decisions.

In many environments, the problem is not too little video, but too little intelligence around the video already being captured. Operators cannot watch every stream with the same level of attention throughout the day. Investigators cannot afford to spend hours searching manually through recordings after an incident. Security teams cannot treat every movement, object, vehicle, or person as equally important. As surveillance estates become larger and more connected, the value increasingly sits in what happens after video is captured.

Turning video into useful alerts

This is where AI video analytics becomes important, provided it is used properly. However, AI should not be treated as a magic layer that suddenly makes a site intelligent. Its value lies in reducing noise, identifying patterns, and helping operators focus on events that deserve attention. A perimeter breach, loitering after hours, an object left in a restricted area, a vehicle moving in the wrong direction, or a person entering a controlled zone all become more useful when the system can flag them quickly and consistently.

For local companies, the appeal is not only better detection. It is also the ability to improve existing surveillance environments without replacing every camera immediately. Solutions such as the Ironlink AI NVR with CVEDIA Analytics for Milestone XProtect point to where the market is moving, giving partners a way to add edge AI and analytics to new or existing video estates through the head-end. That matters because many customers need a practical upgrade path rather than a complete rebuild.

Storage is part of the intelligence layer

Storage is often treated purely in terms of how much footage can be saved and at what cost. While this is important, it is only part of the picture. Once analytics, metadata, searchability, retention requirements, and evidence quality are part of the discussion, storage becomes part of the intelligence layer.

If footage may be needed for an investigation, compliance process, insurance claim, disciplinary matter, or criminal case, it must be stored in a way that preserves its usefulness. Poor retention planning, weak indexing, inadequate performance, or unreliable infrastructure can turn recorded video into something that technically exists but is difficult to use.

The newer generation of AI-enabled recording infrastructure brings together recording, storage, switching, AI acceleration, and health monitoring. That can help reduce complexity while giving customers a more manageable foundation for video recording and analytics.

Identity needs governance

Facial recognition adds another layer. There are legitimate environments where identity matters. Access-controlled buildings, high-risk facilities, campuses, estates, and specific investigative use cases may need to know whether a known person has entered a space or whether someone appearing in footage can be matched against an authorised list. Platforms such as SAFR are built for these kinds of enterprise facial recognition use cases across live video, access control, and mobile environments.

But facial recognition needs a clear purpose, proper authorisation, controlled access to watchlists, defined retention rules, and careful consideration of privacy obligations. The technology can be valuable, but only when the governance around it is as deliberate as the deployment itself.

Connected surveillance carries connected risk

Surveillance is becoming a data system, and data systems carry risk. They rely on networks, servers, storage, software, access controls, remote management, and sometimes cloud services. That makes them part of the organisation’s operational technology environment.

The recent attacks against South African internet infrastructure are a reminder of how exposed connected systems can become. While these incidents were not surveillance-specific, they underline an important point for any connected environment: availability, segmentation, remote access control, monitoring, and resilience cannot be afterthoughts.

A surveillance system that cannot be accessed during an incident, cannot retrieve footage quickly, or depends on weakly protected infrastructure has a different kind of failure. It may still have cameras and recordings, but it may be unable to support the response the business needs.

This is why the next phase of surveillance is not about collecting more video for its own sake. It is about turning video into usable intelligence, supporting responsible identity decisions, storing footage in a way that preserves value, and protecting the infrastructure that keeps the system available.

At Duxbury Surveillance, this is the focus. Partners and customers are asking what can be detected, what can be searched, what can be proven, and what will still work when pressure arrives. More footage does not automatically mean better surveillance. Better surveillance comes from the ability to recognise what matters, act sooner, find evidence faster, and trust that the system will hold up when it is needed.