Technology Q&A: The Future of Intelligent Search with GenAI

The volume of video data in today’s security landscape is growing at an unprecedented rate. While cameras have become increasingly intelligent, generating rich streams of metadata about people, vehicles, and events, security operators often struggle to harness that intelligence in real-time. Actionable data frequently resides in silos, forcing investigators to toggle between platforms or wade through hours of footage to answer a simple question. Bridging that gap between cutting-edge AI analytics and the everyday workflows of security teams has become one of the industry’s most pressing challenges.

SecurityInfoWatch editorial director Steve Lasky sat down with Adam Lowenstein, Americas Product Director at i-PRO, to explore how the company is addressing this problem head-on with its Active Guard platform. At its core, Active Guard was built to eliminate complexity: it translates metadata from i-PRO’s AI-enabled cameras, as well as from third-party analytics applications, directly into leading video management systems (VMS) such as Genetec, Milestone, and Luxriot Video Insight. The result is seamless search, filtering, and alerting inside the very interface operators already use daily, without the need for custom coding or parallel systems.

With the recent release of Active Guard 3.0, i-PRO is introducing natural language search powered by Generative AI, enabling operators to type intuitive queries such as “person carrying a backpack” or “fire truck” and instantly retrieve relevant results. Lowenstein explains the philosophy behind Active Guard, its open-platform approach, and how AI and GenAI are redefining the way security teams find, interpret, and act on video intelligence.

Steve Lasky (SIW): Adam, let’s start with the basics. What is the i-PRO Active Guard solution?

Adam Lowenstein (i-PRO): At its core, Active Guard is the bridge between our AI-enabled cameras and the video management systems (VMS) that security teams already use every day. Cameras with powerful edge processing generate a wealth of metadata, including attributes like vehicle color, clothing type, or even custom data from third-party apps. Active Guard organizes and delivers that metadata directly into the VMS interface. Operators don’t have to learn a new platform or bounce between systems. They can search, filter, and trigger alerts natively inside Genetec, Milestone, Luxriot Video Insight, and others.

What makes it powerful is that only metadata is passed along, not heavy video streams. That means faster searches, less bandwidth, and far more actionable intelligence without adding racks of servers.

SIW: Why did i-PRO create Active Guard?

Lowenstein: We saw a gap. AI cameras were getting smarter, but the valuable data they captured wasn’t reaching operators in a usable way. Too often, analytics ran in silos, which, while useful in theory, were hard to access when you were working inside a VMS. We wanted to make that intelligence easy and intuitive for every operator to use. Active Guard was designed to streamline the workflow, allowing operators to quickly select from 98 unique identifying attributes directly within the VMS and either search or trigger alerts, receiving results instantly without leaving their normal VMS interface.

And just as important, we built it on open principles. Security shouldn’t be a closed ecosystem where one vendor dictates what you can and can’t do. Active Guard is designed to serve as the glue that connects cameras, third-party analytics, and VMS platforms.

SIW: The latest release, Active Guard 3.0, adds Generative AI. How does that change the equation?

Lowenstein: It’s a big step forward. Up until now, search has been limited to predefined attributes, and our cameras produce 98 of them, which is already an industry-leading number. With version 3.0, we’ve added free-text search powered by Generative AI. Operators can type natural language queries such as “person carrying a backpack,” “fire truck,” or “individual who fell”, and the system interprets what they mean.

That removes the friction of having to guess which attribute checkbox to tick. It feels intuitive, like asking a colleague to find something for you. And because it runs entirely on-premises, sites that don’t allow cloud connectivity, such as airports, government facilities, and utilities, can still benefit without exposing their data.

SIW: How is i-PRO addressing the privacy and cybersecurity concerns that often accompany AI and GenAI?

Lowenstein: First, all of this runs locally. There’s no video being shipped off to the cloud for interpretation. The ‘best shot’ images are generated at the camera and the GenAI engine sits on the Active Guard server, so no outside internet connection is required. That keeps sensitive video under the customer’s control.

Second, we designed the system with longevity and security in mind. We’ve been vocal about moving beyond “install it and forget it.” AI capabilities evolve rapidly, which requires regular updates, patches, and governance. That’s why i-PRO was the first company in the security industry to receive ISO/IEC 42001 certification for AI management.  

SIW: One of the things that caught my attention in version 2.0 was the integration of third-party metadata like Vaxtor’s ALPR. Can you explain why that matters?

Lowenstein: Sure. Vaxtor is a great example. They build high-performance optical character recognition tools that can identify license plates, shipping containers, ID numbers, you name it. Traditionally, getting that data into a VMS required custom work by either the VMS vendor or the analytics developer. That’s expensive and slow.

Active Guard eliminates that barrier by allowing Vaxtor metadata to come straight into the same VMS window where you’re already running searches. The same goes for Morpho’s fall-detection analytics and others. We’re creating a neutral bridge that allows third-party insights to coexist seamlessly with i-PRO camera data, without requiring special code. i-PRO’s open platform approach represents a significant win for integrators and end users, as it eliminates development work and enables faster deployments.

SIW: How does this differ from competing solutions in the market?

Lowenstein: Some vendors still take a closed-system approach. They’ll sell you cameras, analytics, and a VMS, but they don’t want you to mix and match. That creates lock-in. Nobody is bringing in third-party plugin metadata, such as Vaxtor, to all the popular VMSs with zero API or SDK work, to my knowledge.

Active Guard is the opposite. It’s not a walled garden; it’s an open marketplace of attributes. If you want to use i-PRO cameras, that's great. If you want to add Vaxtor’s ALPR or another containerized AI app, also great. If your VMS is Genetec or Milestone, you don’t have to wait for them to code an integration; we handle the translation. It’s that bridge function that really sets Active Guard apart. Additionally, suppose you have a non-AI camera, even one from a different manufacturer. In that case, you can utilize an i-PRO X Series camera to extract AI metadata from three additional non-AI camera streams, which also feed directly into Active Guard.

SIW: Beyond search, what are some of the new investigative features in 3.0?

Lowenstein: We’ve introduced Person Image Similarity Search, which uses re-identification technology. Say you have a still image of a person of interest. You can feed it into Active Guard, and the system will identify visually similar appearances across your video archive, even if clothing changes or lighting conditions are poor. It’s not facial recognition, so you’re not dealing with identity or biometrics. It’s about accelerating investigations within a visual context and finding the needle in the haystack.

SIW: How does this all fit into the bigger industry trend of edge AI and containerization?

Lowenstein: The industry is realizing that pushing everything to the cloud is not sustainable for many users. Video is too heavy. Edge AI solves that by creating lightweight metadata at the source. But the next leap is containerization. Using Docker, you can package specialized AI tools and run them directly on the camera. That’s how you get applications like fall detection or environmental monitoring without having to rewrite firmware. Active Guard is the layer that makes those edge insights usable inside a VMS.

Think of it as turning cameras into smart sensors and then giving operators the ability to search and act on that intelligence without ever leaving their normal workflow. It’s that unified, single-pane-of-glass user interface that we aspire to, which simplifies security operations.

SIW: What feedback have you received from integrators and end-users?

Lowenstein: They tell us it saves them time and money. Investigations that used to take hours of scrubbing now take minutes. Integrators appreciate not having to manage parallel systems or explain to campus security why ALPR data lives in a different interface than their video. And importantly, they like that we’re not forcing them into a single ecosystem. Open beats closed every time.

SIW: Finally, where do you see Active Guard going next?

Lowenstein: We’ll continue to expand its role as the metadata bridge, with more attributes, third-party apps, and seamless integration with leading VMS platforms. We will continue to refine natural language search. GenAI is just getting started, and we see enormous potential in making video intelligence as easy to access as typing a thought into a search bar. Our mission with Active Guard is simple: help security professionals find what matters, when it matters, as simply as possible.