Skip to content

Open Systems

AI in Security: Part 1 of 2, the Short-Term View

One year ago, ChatGPT went mainstream and started the AI hype train. Security technology manufacturers jumped aboard, promoting AI-driven improvements in video analytics and security software user interfaces.

So is your job about to get meaningfully easier because of AI?

Not really.

As with any major new technology, people generally overestimate the short-term impact.

Based on what we saw at ISC West 2024 and in speaking with vendors and our clients, here is our take on the short-term:

You will soon be able to use more natural language in searches like “Show me the White Tesla S3 in parking lot B last week”. This is somewhat better than using dropdown lists, and we don’t begrudge UI improvements. Other queries like “What time did Amy Cooper leave on Monday?” just don’t move the needle enough. We will get interested when we can ask “Show me the most recent car collisions” without seeing false positives and missing events. But noone is advertising that yet.

As reported by IPVM (paywalled post) and seen by our own ISC West attendees, the new “AI” product enhancements are mostly smoke. The Best New Product winner couldn’t find a “garbage can” in a video search, and a search for “smoke” returned BBQ grills. Another product mistook hands and phones for guns. So far this seems more like rebranded analytics in beta.

Where we see some short/medium-term AI promise is in smarter tagging and analysis of security datasets. By dataset that we mean the people & objects in the video, access control entries/exits, and other sensor data like temperature. With clean tagging and good analysis (two different things!), then you could ask, “Show me the anomalies.” Better yet, the system will proactively tell you about the potentially worrisome stuff.

Alas that’s a few years off from being reliable, and these engines are going to need your own site-specific training effort to handle “permitted exceptions” like shift changes.

While we are skeptical about the short-term impact of AI on your security solutions, it’s going to eventually help. After all… with any major new technology, people generally underestimate the long-term impact.

How We Choose Technology Suppliers

Picking physical security technology suppliers has never been more complex than it is today, and it promises to get only more difficult given the number of new and exciting technologies being developed – and the number of manufacturers and distributors hawking them.

Last year, Secuni used more than sixty suppliers. Of course, some are central to a solution while others provide just a few peripherals. To be an effective design-build outfit, management of supply chain is an important part of our success. Not all the new suppliers will be successful, and we do not want to strand you with obsolescence. The biometrics and analytics markets are littered with skeletons of defunct companies.

What are the key factors for us?

  1. We are hesitant to adopt new suppliers without extensive field testing. We also feel it is fair to ask new suppliers for financial bona fides to establish their potential longevity as a going concern.
  2. Product support is a massive difference-maker: we expect thorough sales support, and training for both ourselves and our clients.
  3. Open architecture is key. We do not advise our clients to invest in full solutions because there are so many great open-architecture software developers. That flashy feature from the closed system will soon be copied by others and be available to most of the open-architecture market.
  4. Commitment to current and emerging standards, like ONVIF and OSDP.

Our own team knows that we can’t represent every good product. To maintain expertise, we limit ourselves to a set of solutions that we feel satisfy different market needs, while recognizing that we cannot address all needs – and that many of the suppliers we don’t represent have great products that solve client problems very successfully.

This helps us stay grounded and tuned into the reality that although product selection is important, it’s ultimately the implementation and operational teams that create successful outcomes more than specific product features.

We also encourage our clients to spend time in the selection process. Consider how many hours you will spend operating the system after selection! The upfront investment of time, while hard to find for most of us, is critical to ensure the best fit. Often, during the selection process, we uncover important additional requirements that help the vendor selection process.

Centralizing Your Security Systems

It’s common for large companies with multiple facilities in the US (and overseas) to have a fragmented set of security systems at all those sites. You might have several kinds of video surveillance solutions running at the same time, and access control systems that don’t talk to each other or your HR database.

Our Perspective on Open Architecture

We firmly believe in the long-term value of using open-architecture security hardware and software. In other words, fundamentally extensible components that can easily integrate together.

In contrast, some security product lines lock users into a proprietary platform. This may meet your requirements and save capital expenditure in the first few years, but at the cost of higher operating expense and narrowed future choices.

Systems based on open architecture can accommodate new innovations faster, easier, and at lower cost.