I recently attended two recent industry conferences — Axis Connect & Converge Conference (ACCC) and the joint Eagle Eye and Brivo Cloud Security Summit. I wrote in depth about some of the top takeaways here. But SDM was also given exclusive opportunity to interview some of the biggest thought leaders in the industry at all three companies, where we delved deeper into three important topics: where AI is on the adoption curve, whether cloud can be both open and closed, and the role of the security integrator in all of this.

At the Axis ACCC conference, Fredrik Nilsson, vice president Americas for Axis, showed a slide of the 2024 Gartner AI Hype Cycle, which suggested that generative AI is causing the technology to head into what it calls the “trough of disillusionment.” This is the downward slope of the bell curve, which will eventually even out into a “slope of enlightenment” and finally end up in a “plateau of productivity.” I asked Nilsson; Martin Glen, founder and CEO, Axis; and Dean Drako, founder and CEO, Eagle Eye Networks about where they feel the adoption of AI really is, specifically in the security industry.

“I don’t think we are in the trough of disillusionment,” Drako said. “Just look at the money being dumped into AI. And it is actually delivering the goods. I use ChatGPT, not to write, but as my primary search engine. The adoption curve of ChatGPT is the fastest adoption curve in the history of mankind ­— faster than Google. Everyone has tried it out.”

Drako himself was walking around with an AI-based lapel mic that would give him a summary of every conversation he had at the Summit. Gren also noted that he had started using Chat GPT as his main search engine.

However, Nilsson was quick to point out that, while datacenters are investing a “mind-boggling” amount of money in it, many inside the security industry and out are still in the “playing with it and discovering what it can do” phase, which seemed to be the case with many of the attendees at both conferences.

“A lot of people playing around with AI, and there are some good applications, but who is making money on this now?” Nilsson asked.

“They need to figure out a business model,” Gren agreed.

When they do, Drako believes AI will make it easier to get more out of security systems, particularly with the development of large language models and tools. “In our industry, I believe there are going to be two facets of AI that come to impact the customer,” Drako said. “The first aspect is enabling the user to more easily leverage the tool. I can say, ‘Add John to the access control system, and give him access to these work hours,’ all using natural language. They can use more advanced features of the product without actually having to understand them completely. …  That is the first real impact in the security industry: ease of use and expansion of the features used by the customer.

“The second piece of AI is AI utilized in the analysis of the data. On the access control side, you are looking at patterns — who has been where. On the video side, you can do stuff in real time. You can ask to see any time someone slipped and fell in the last month. Ease of use and analyzing data are the two core functions that AI will serve.”

There is still work to be done on the back end, though. “When it comes to data centers … I found some graphs this weekend from a report talking about the amount of data power that each kind of interaction takes and a regular Google search takes 1X data power in the data center somewhere,” Nilsson said “But when you use it with an LLM, the power it takes from the data center is 10 times that. When you look at analyzing images, it’s 320 times the Google search. And when you use it for video, it’s 10,000 times.”

Another issue Nilsson pointed to was the information used to train the models, particularly for ChatGPT. “Where does most information come from to create those LLMs? I saw a graph where 40% comes from Reddit, which is people like you and I who throw the information in unedited, and [something like] 30% … was from Wikipedia, which I think is pretty good,” Nilsson said. “And the list went down from there, to where 4% came from Target stores — their website — and I thought, ‘Hmm, interesting, so that’s the information we’re reading about.’ Yes, everything they get to train those models comes from the Internet. But where does that information come from?”

Despite this, all agreed AI is coming, and fast. “What is fascinating to me is customers are showing up and asking for it,” Drako said. “Integrators don’t even need to sell it. They just need to show up and take orders. That said, it isn’t easy. One percent of dealers might be able to do that now. But it will get easier.”

One of the paradoxes Steve Van Till, founder and CEO, Brivo, mentioned in his keynote at the Cloud Security Summit was the seeming contradiction of end users wanting both open systems and unified, “one pane of glass” solutions. So I asked him, can systems really be both unified and open? 

“What we are trying to do is have both,” he explained, giving a few examples. “One company doing access control can never match the selection and quality of full-time lock company. .... There are places where open is working very well and not at odds with unified platforms. The thing we are doing differently is, first, we have two companies, Brivo and Eagle Eye, with a common backbone between the two — all your permissions work across both applications. We are inviting certain partners to use that backbone like Brivo Visitor Management powered by Envoy. It gives the user one transaction, one reseller, one login, one support call and one invoice. That arrangement gives most of the business benefits of consolidation or unification because all that workflow is the same, but it also provides the openness of being able to have best-of-breed companies come into that same backbone. We couldn’t have done that five or 10 years ago. We didn’t have market stature yet. We do today. We are at a stage where we can construct an open suite strategy and get companies to be part of that. What we will do is continue to build more first-party applications, but also bring in more third parties.”

Van Till added, however, that hardware openness is much less involved than software openness, and any sort of suite solution is going to be inherently somewhat “hybrid” proprietary. He doesn’t necessarily view it as a problem for the industry.

“There is this whole history of software,” Van Till explained. “If you became a customer when SQL databases came out, it was a standard. Every one of them put in all these proprietary extensions. If you really leveraged those, you became locked in. The same thing happened in cloud. Public cloud infrastructure in beginning was a raw server. AWS now has over 1,000 SKUs. Like many customers, we have taken advantage of all those things. But now we are very locked in to either Amazon or Azure or Google — every bit as much as proprietary locks. Early in these technologies, they tend to be fairy open, but, as they try to differentiate themselves in the market, they do so by adding more features that [are] proprietary. Even if you look at APIs, the endpoints that are available are unique to the way they look at a problem, the framework of it being on an IP network. Those are things everyone understands. It is ‘open’ at a level that you can integrate to it.”

Steve Burdet, manager, solutions management, Axis Communications — another company committed to being open platform — also spoke with SDM about this topic at ACCC, regarding his company’s cloud path. “We have a handful of cloud products. We want it to be open and flexible,” he said. “One thing we hear is customers don’t want to be locked in only to find out they have to go back to the proprietary world. So, we want to build good tools on top of our Cloud Connect platform, but we also want to make those same products available to all of our partners. … I like to think of it as an open way to provide an exit strategy.

“It is frightening people that they are locked into their cloud. That is the business model for some competitors. That is not the world I want to live in,  or that we want to create. … It is your data and your system,” Burdet added. “I like the opportunity of thinking of a way to exit if they need to. … Making it easier will hopefully invite people to come to our cloud. … It  is a fundamental problem nobody really wants to solve. Without having to solve it, at least we are not locking them in.”

3. How Can Integrators Best Navigate Their Changing Role?

At both partner conferences, as well as Securing New Ground, also held in October, I heard discussions around whether security integrators were up to the challenge of adopting and utilizing these fast-paced technologies. Many are, to be sure. But I asked these top leaders what advice they would give their integrator resellers, and whether those that aren’t on the AI/cloud adoption path are in danger of becoming “device hangers” and losing business to the more managed services providers.

“I have been talking about this for a year or two now,” Drako said when asked how integrators can best adapt to the AI world. “Get in and become more intimate with your customer. If you think about it, our relationship with the customer today is modest. The customer calls up and says, ‘I need some security.’ You install it and move on. Now, it is going to be not just about security, but helping them run their business. You need to understand in more detail what problems you are trying to solve. We literally had a sandwich shop wanting to make sure all sandwiches were being made per the recipe for consistency and quality control. Today, it is about understanding the goals of the business and crafting and configuring a solution that can help them meet those goals.”

John Szczygiel, executive vice president and COO, Brivo, gave the same advice he had when asked about adapting to RMR selling: Don’t sell the buzzword, sell the solution. “They are not really selling AI. They are selling a feature or capability that might be powered by AI,” he said. “A few years ago, everything was ‘big data.’ Today, AI is a little like that: ‘Here are six different features that in some way leverage AI.’ AI is a tool that can help your system do a lot more. My advice to integrators is: make sure that capability maps to something the customer appreciates. You are not selling ‘AI’ just like you are not selling ‘RMR’.

“Nobody really goes out to buy AI,” Szczygiel added. “They buy features assisted by a tool. So you need to be able to explain how things work. … Someone has to have the answers.”

He also suggested bringing in your manufacturer partner on these technical discussions to make sure all the customer questions get answered. This was also something Burdet discussed with SDM.

“Some integrators are making so much money pulling cables and hanging things that they are swamped with work just doing that. They are doing so much, it is hard to focus on all the new things,” Burdet said. “But I do believe the term ‘integrator’ implies a lot more than just installing. There is so much need and opportunity for connecting. … Integrators know their customer needs better than anyone. Whether they are investing in that now, I really hope they are. I hope they are getting deeper and wider with their existing customers and also attracting new customers.

“We have professional services departments that can help them with that. Integrators are not necessarily having all these conversations alone. … I would hope they are thinking about that. Everyone here [at ACCC] is smart and doing great business, but some who are simply following flows of money could just become subcontractors, eventually.”

Gren also had thoughts on this topic, adding that the company’s commitment to always going through the channel helps integrator partners feel comfortable having them join those customer conversations. “That is the beauty of our business model,” he said. “We don’t sell direct, no price negotiating. We have the model that builds in the trust so we can talk freely to the end customers.”

Nilsson closed with this thought: “What the integrators appreciate is the knowledge we can transfer. It is interesting, sometimes end customers like to complain about the integrators. But I think, while we don’t have an easy job developing products or our own chipsets, it is the integrators that make all these solutions from different vendors work together. They have the really tough job.”