How to Navigate the AI Maze at ISC West

That excitement was quickly followed by a second wave centered on agentic AI and partially autonomous systems, with broad promises about how dramatically security operations might improve. This blending of terminology created confusion. As AI became a marketing label rather than a technical distinction, it became harder for integrators and consultants to assess what was genuinely new vs. what had simply been rebranded. In many cases, advanced analytics and rule-based automation were presented alongside true machine-learning capabilities, blurring the line between common analytics and actual artificial intelligence.

We saw this firsthand: Several years ago, when video surveillance companies were heavily marketing AI, a client asked us to review a proposed “AI-powered” video analytics platform that had been presented as capable of dramatically reducing false alarms when layered over the client’s existing video management system. When we looked more closely, the core functionality was largely a collection of rule-based analytics that already existed in many VMS platforms but typically required intentional configuration.

The software itself was solid and did simplify the setup of these analytics, but the way it was described created unrealistic expectations about how adaptive they would be once deployed. In practice, the system still required significant programming and tuning rather than independently learning scenes and conditions as the client had assumed.

Once our project team aligned on what the technology could realistically deliver, the conversation shifted away from excitement about AI and toward more practical factors, such as camera placement, lighting conditions, and how the system would actually be managed day to day. In the process, we even discovered that the client’s existing cameras had stronger onboard analytics that were not being leveraged and could address many of the challenges they were experiencing.

How to Evaluate What You See

This dynamic will be especially visible on the ISC West show floor. Vendors who differentiate themselves will focus on how solutions are deployed, supported, and sustained once they move into real-world environments, and how clearly they can explain where the technology fits, how it integrates, and what it requires to perform consistently over time.

At this stage of AI adoption, the primary risk is misalignment. When use-cases are loosely defined, integration paths are unclear, or operational ownership is left unanswered, even strong AI solutions struggle to move beyond pilots and demos. As integrators and consultants engage vendors on the show floor, the most telling signal will not be how advanced a solution sounds, but how clearly it is explained and how it translates into successful deployments.

Listen for clear use-cases, clear integration paths, and clear ownership once the system goes live. Those signals are often more valuable than feature lists or demo performance. When you visit a booth, start by leveraging your real-world experience and knowledge of the challenges your clients are facing. This is an invaluable perspective…use it! With that in mind, here are five specific areas to address:

1. Clarity of use case. Strong solutions can clearly explain the specific problem they are designed to solve and the conditions under which they perform best. Vague answers or overly broad claims often indicate that the technology is still searching for a true operational fit.

2. Ease of integration. AI rarely operates in isolation, and solutions that succeed are those designed to work cleanly within existing environments. Ask how a solution integrates with current platforms today – not future roadmaps – and what dependencies or constraints come with that integration.

3. Ownership. AI systems require tuning, monitoring, and ongoing adjustment to remain effective. Conversations that quickly clarify who is responsible for training models, managing false positives, and maintaining performance over time tend to reflect solutions that are ready for real deployment. When those answers are unclear, long-term success becomes harder to predict.

4. Deployment realities. Solutions that acknowledge real-world conditions – occupied facilities, mixed infrastructure, limited staffing – signal maturity. If a solution only works under ideal circumstances, it is unlikely to hold up once it leaves the booth.

5. Value in operational terms. The most compelling AI conversations connect technology back to measurable outcomes: reduced manual effort, improved response times, better decision-making, and clearer insights. When value is expressed this way, it becomes easier to assess whether a solution will deliver benefits beyond the demo.

Where Integrators Win or Lose

As AI capabilities continue to mature, the role of integrators and consultants is becoming more defined and more consequential. Simply specifying or selling the latest AI-enabled solutions doesn’t provide value; implementing them responsibly does. That includes coordinating across design and IT stakeholders, setting realistic expectations with clients, and planning for how systems will be operated, maintained, and adjusted over time.

Execution starts with integration. Solutions that work cleanly within existing environments consistently outperform those with deeper feature sets but heavier dependencies. Clean integration reduces friction, simplifies deployment, and lowers the operational burden once systems go live. Reliability matters more than novelty, especially in environments where security systems are expected to perform continuously, not impress occasionally.

This dynamic often plays out in the field. On one deployment, a client initially gravitated towards a sophisticated analytics platform with an extensive list of AI-driven capabilities. The demos were impressive and the feature set appeared to solve a wide range of potential security challenges, including the promise of bringing the data from disparate systems into one place; however, once we began evaluating how the system would integrate with video management and access control platforms as well as their HR platform, scheduling, and meal planning systems, it became clear that the deployment would introduce significant complexity vs. its promise of simplification.

The system required custom integrations and ongoing tuning that the client’s security team was not staffed to support. After stepping back and re-evaluating the operational goals, the project team ultimately implemented a more streamlined approach by focusing on integrating the video and access control systems under a single platform that could then natively integrate cleanly with the rest of their environment. While this did not leverage all of the AI features, it consolidated their systems (and data), allowed them to leverage analytics now available within the integrated environment, and created a foundation where future AI add-ons could be integrated as the video and access control platform developed them.

Just as important, this alternative solution deployed faster, required far less ongoing maintenance, and produced more consistent results for the operators responsible for monitoring the system. In the end, the success of the project was not determined by the depth of AI capability, but by how well the technology fit into the environment and the people responsible for running it.

This illustrates a critical point: The trusted advisor role is vital. Clients measure success on dashboards, but they also measure it by confidence – confidence that systems will perform as expected, that alerts are meaningful, and that someone understands how the technology behaves in real conditions. Integrators and consultants who prioritize clarity, reliability, and accountability earn that trust and differentiate themselves in the process. Focus on execution over novelty and you will turn AI promises into measurable outcomes for your clients long after the ISC West show floor clears.