Key Highlights
- Vendors who want to stay relevant will need to open their APIs and make them agent-friendly — or risk being routed around entirely.
- Metcalfe's Law made comprehensive multi-vendor integration economically impossible for decades; autonomous agents are about to solve that problem.
- Integrators who embrace agentic tools will accumulate lasting intellectual capital job to job — a competitive differentiator the industry has never had before.
AI is about to radically restructure the multi-billion-dollar systems integration business by changing the way software and hardware products can talk to each other. The reigning paradigm of bespoke integrations crafted by the software vendors themselves is about to end. Instead, we will see agent-coded connectors that can be created by anyone, with minimal effort, and only rudimentary technical skills.
This shift will give systems integrators and end users a massive expansion of choices for how they can use and interconnect the open systems of their choosing — or at least those with agent-friendly APIs.
The math is the problem
What makes this such a big deal is the raw math of connecting a large number of disparate products or endpoints in a multi-vendor ecosystem. Metcalfe’s Law observes that the number of connections between N nodes in a network is proportional to N2 (N squared), which is to say that it gets very big very fast. In a network with 100 endpoints, for example, there are (100)2, or 10,000 possible interconnections. In a network of 1,000 endpoints, N2 equals 1 million. You can see where this is going.
The good news in Metcalfe's Law is that the utility of such networks or ecosystems also increases at the same rate — N2. That mathematical fact has been the basis of value creation in transportation systems, telecommunications networks and the internet as a whole. A railroad system connecting thousands of producers and consumers was the infrastructure of the industrial revolution, and the computer network known as the internet with billions of users is what has given us the modern information economy.
The unfortunate corollary of Metcalfe’s Law for inhomogeneous networks is that the number of possible integrations among N products — and the cost to implement them — also increases like N2. The practical implication is that for any pair of products a customer might wish to connect, it’s very unlikely that the vendors would have built an integration. Sad but true: the universe of possible product combinations is bigger than any vendor can profitably afford to build under the bespoke integration paradigm.
That’s why integration costs going toward zero is such a big deal.
I learned this math fairly early in my journey through the security industry’s maze of integration demands. No matter how many video management systems (VMSs) we integrated with Brivo’s access control system, for example, it seemed that every new customer had a different pet VMS that we absolutely had to support. Typical of a highly fragmented market, most of these products only had single-digit market share. It quickly became obvious that even if we integrated 20 or 30 or more, we would still cover less than 50% of the VMS market.
Systems integrators have been able to solve parts of this problem, but not all of it. Any systems integrator worth their salt has a library of hacks they’ve built up over time to cajole disparate products into working with each other. As my high school shop teacher so elegantly put it, the methodology is often to “get a bigger hammer.” Integrators have techniques to make things fit in ways the manufacturers could never have foreseen, but the resulting system can be a Rube Goldberg contraption that’s both costly and fragile. Some square pegs just won’t go into a round hole.
The impact of agentic AI
Enter AI — specifically, the new breed of autonomous agents that run on a user's own machine and take action across apps and APIs through chat interfaces, simple markdown (text) files, or vibe-coded behaviors. Some of the more prominent entries in this category are projects like OpenClaw, Manus and NemoClaw. Many of them are available open source (i.e., free), and they are cheap to operate because you can run them on your own computer (or, better, on a minuscule fraction of a VM at a public cloud provider). Similar capabilities can be achieved through general-purpose AI tools like Claude Code, or even enterprise platforms like ServiceNow.
Unlike cloud chatbots that mostly answer questions, these platforms actually do things: send emails, manage calendars, browse the web and remember context over time. They're extensible through roll-your-own “skills,” with data and credentials staying on the user's hardware rather than in a vendor's cloud. More to the point, they can also be instructed to carry out tasks like data synchronization or message exchanges between two software or hardware products that don’t natively connect with vendor-supported integrations.
With these new agents, integration comes down to this: if you can log into Product A and Product B, you can automate the exchange of data between them. And if the products have agent-friendly APIs, it only gets better.
For some real-world examples of how this works, take a look at the OpenClaw website. You’ll see myriad user-posted success stories about workflows across a multitude of products. There’s SwitchBot, a home AI agent running OpenClaw locally and controlling an ecosystem of smart curtains, locks, plugs, humidifiers, sensors and camera-based AI detection. For the Raspberry Pi fans, there’s the robotic ROSOrin Pro platform running an OpenClaw AI agent with a 3D depth camera and TOF LiDAR. And then there’s the many examples of integrating personal electronic devices and wearables into automated health tracking and reporting systems.
Agents like OpenClaw are quickly becoming a part of the physical security integration landscape. At ISC West this year, for example, I heard a great use case from Collin Trimble, owner and CEO of Alarm Masters in Houston, Texas. Collin faced the familiar problem of needing to connect two systems that had no native integration support from the respective manufacturers — in this case, Brivo’s access control platform and a third-party intrusion product.
So he’s writing his own connector in OpenClaw.
If you’re reading this and you’re not blown away, you should probably quit the security industry and find something else to do. This is like waking up in a different world. It’s the future of systems integration and it’s going to change the business dramatically over the next 12-18 months, if not sooner. It’s truly a watershed moment.
What changes for integrators?
Integrators acquire the capability to integrate anything with anything — and have a hand in the business logic that connects them — which puts tremendous power back in their hands. This skill set provides a platform to vastly expand the value-add integrators can deliver to end users. Leveraging this ability will separate those who continue growing at the speed of the AI era from those who get stuck in time with the way things were done in the past.
Until now, the core value proposition of systems integration has been three-fold: diagnosing the security needs of the customer, knowing how to install the products that solve their problems, and having the trained labor pool to make it happen. The intellectual capital in this equation is, roughly speaking, the Certified Protection Professional (CPP) body of knowledge plus prior experience in getting disparate product groups to work together. Neither of these is a competitive differentiator to any significant degree.
What agentic integration adds is the ability for integrators to develop truly unique functional combinations of third-party systems that are tailored exactly to their customer needs. This expands the available solution set beyond what manufacturers support with native integrations. Agentic overlays enable custom business logic and behaviors that reside above the individual product layer — without having a cadre of software engineers to make it happen. That’s a bit of Holy Grail, in my book.
As an example, consider the all-too-common problem of having to monitor the health and performance of hundreds or thousands of customer installations, with nearly as many legacy products spread across their respective properties. This is a bigger challenge than first meets the eye because of the N2 problem discussed earlier. Still, many integrators have embarked down this path after jumping over the corpses of various PSIM platforms that tried to do the same. I’ve watched on the sidelines as many of these initiatives launch to much fanfare. Even the largest systems integrators seldom reach the goal line. They hit the N2 wall and retreat.
With the AI tools now available for codeless development, the intellectual property — read, differentiators — that integrators can retain from job to job grows immensely larger. It’s no longer a rinse and repeat operation that leaves little accumulated intellectual capital. To put a finer point on it: these agentic tools can help capture truly differentiating intellectual capital at the point of use. This observation also answers the question everyone is asking about every business and every market everywhere: Where can we apply AI for a true advantage? Right here, folks.
Here's what happens next
What happens across the rest of the industry as AI “puts the integration back in the integrator,” so to speak? I have several predictions.
Technology providers (aka, manufacturers) will invest more in making their APIs more “agent friendly.” For the most part, this means providing API functional documentation in a format that’s consumable by agents rather than humans. Typically, this takes the form of a discoverable “LLM.txt” markdown file somewhere on the API platform. With this guidance, agents can determine how to use one or more APIs to accomplish a given task.
This shift of focus will drive the availability of more “headless” systems; i.e, platforms for which the provider’s GUI is optional, but their API has a much larger surface area. Bonus: the person driving agentic integrations needs not be a programmer per se, and won’t have to learn the arcane API specs of every tech stack they touch. They will be able to do everything they want with natural language descriptions (prompts).
This entire movement — agentic development, headless systems, natural language integration — will drive down the cost of high specialized end user experiences. Many of these will take the form of verticalized interfaces that collect just the functionality that each customer cohort needs. The multifamily user will get everything they need for running an apartment complex, without the distractions of features that were intended for a different market. The CRE user will have a view of their real estate portfolio that allows them to manage the essentials quickly and easily. SMB customers will have compact mobile apps that are a perfect fit for their small business and an on-the-go management style
Open systems architecture and open business policies will become an even higher customer priority than they already are. That’s because all of these agentic integration advantages only accrue in a world where technology vendors develop and promote open systems. What does open mean in this context? APIs that are available to everyone. APIs that are documented and publicly discoverable. APIs that are agent friendly.
The most agent-friendly characteristic of all? No artificial business barriers erected by anti-competitive, ring-fenced platform providers that ultimately just want to keep customers captive. This is how we drive integration costs toward zero.
That’s real intelligence.
About the Author
Steve Van Till
Co-Founder, President and CEO of Brivo, Inc
Steve Van Till is Founder and President of Brivo, the pioneering cloud-native platform, delivering unified access control, video surveillance, mobile credentials and identity solutions for enterprises. A recognized physical security authority, he holds multiple patents and has shaped industry standards as Chairman of the Standards Committee and Board member of the Security Industry Association (SIA). Van Till authored the ASIS International Book of the Year, “The Five Technological Forces Disrupting Security,” and is a recipient of the SIA Lippert Memorial Award.