The convergence of AI, IoT and building automation: Why interoperability will define smart infrastructure

By Varun Joshi, Co-founder & Director, Unwired Connect 

Most conversations I hear about AI in buildings start with the software: the models, the analytics, the dashboards. I get why. That’s the part you can demo. But I keep coming back to a simpler question that rarely gets asked in those rooms. What is all that intelligence plugged into?

AI needs something underneath it, a dense and dependable layer of sensing and actuation. If the building has no connected endpoints, then “AI-driven building management” is a dashboard with no hands. It can watch and it can report, but it can’t do anything. The smart drivers and controllers sitting in the ceiling are where AI actually touches the physical world, and how well that handoff works will decide how much value AI ever delivers in a building.

The real bottleneck is silos

Think about a typical commercial building. There’s HVAC, lighting, access control, a BMS, and sensors dotted around. Any of these may be perfectly good on its own. The trouble is that they usually come from different vendors, speak different protocols, and were never meant to have a conversation.

So the building ends up with several smart systems and not much intelligence. Occupancy data from one system can’t inform another. Energy numbers live in one portal and lighting schedules in another, and someone has to stitch it all together by hand. An AI model can’t optimise what it can’t see, and it can’t coordinate systems that don’t share a language.

That’s why I see fragmentation, rather than a lack of technology, as the thing holding smart infrastructure back. The way forward is to build products that are protocol and platform agnostic from the start, so a customer is never forced into one vendor’s world.

Open standards are what make it real

Anyone can promise interoperability. Delivering it is harder, and I think it only counts when it rests on open standards and independent certification.

Lighting control is a good example. DALI-2 certification, run by the DALI Alliance, gives you confidence that a driver from one manufacturer will work with a controller from another. If you’re a system integrator on a large project, that matters a great deal. It lowers 

commissioning risk, it frees you from relying on a single supplier, and it keeps the design flexible as the building’s needs change over the years.

Choice of ecosystem matters just as much. Projects differ. One might suit a Bluetooth mesh platform like Casambi. Another might do better with the wide device range around Tuya, which covers Bluetooth mesh, Zigbee and Wi-Fi. A third might simply call for wired DALI-2. A healthy market lets customers pick what fits each job instead of squeezing every job into one architecture. Manufacturers have a part to play here too. Standards are only as strong as the number of companies willing to take part in them and get their products certified.

It isn’t wired versus wireless

People have argued about wired against wireless for years, and I think the argument misses the point. In practice, most buildings end up with both.

A large share of the buildings that will be standing in 2040 already exist. Many of them can’t be rewired, and often shouldn’t be, given the cost and the disruption to the people using them. Hybrid setups solve that. Wireless DALI and DALI gateways let a building gain intelligence without anyone opening up ceilings and walls.

So interoperability isn’t only about how new devices get along with each other. It also means getting along with what’s already there. A solution that needs a clean slate will only ever reach a small fraction of buildings. One that can be layered onto existing infrastructure reaches far more of them.

Why lighting is the place to start

If you had to design a sensing and actuation network for a building from scratch, you’d struggle to improve on lighting. It’s already in every room. It’s powered, spread evenly through the space, and installed by professionals. That makes it a very natural backbone.

Once lighting is connected, it does more than light a room. Occupancy sensing shows how a space is really used, which is often quite different from how it was planned. Daylight harvesting dims the artificial light as natural light comes in, saving energy without anyone noticing a drop in comfort. Energy metering gives you consumption figures down to the fixture or zone. And the drivers and luminaires themselves report on their health and runtime, which is useful asset data you’d otherwise have to go and collect.

The nice part is that you get this sensory network without a separate infrastructure project running alongside. For many buildings, lighting is simply the easiest and most affordable place to begin.

Moving from reactive to predictive

The big change comes once devices share data across systems and AI has real material to learn from.

At that point a building can stop reacting and start anticipating. It can learn occupancy patterns and bring a space to the right condition before people walk in. It can spot a driver or fixture that’s starting to degrade before it fails and disrupts the day. It can adjust energy use continuously as conditions change, instead of following a fixed schedule that was set months ago.

For facility managers, this changes the nature of the job. Today many of them find out how the building performed when the monthly bill arrives, and then try to work out what went wrong. With shared data they see what’s happening live and where to step in. Maintenance gets planned instead of urgent, and wasted energy is caught as it happens.

Lighting that looks after people

Efficiency and cost savings matter, but buildings are there for the people inside them, and this is where connected lighting gets especially interesting to me.

Human-centric lighting adjusts intensity and colour temperature through the day to follow our natural rhythms. Add personalised scenes, and AI and IoT begin to improve how people feel and how well they work. Light that follows the body clock, suits the task, and respects individual preference can make an office, a classroom or a care home a noticeably better place to spend time. Doing this at scale takes the connected, interoperable foundation I’ve been describing. Without it, these ideas stay stuck as one-off pilots.

Working with everything else

It’s tempting to see smart infrastructure as a race to add features, with every vendor piling on more capabilities, more dashboards and more proprietary integrations. I don’t think that’s how this will be decided.

The companies that come out ahead won’t be the ones with the longest feature list. They’ll be the ones whose products work with everything else. Integrators and customers have grown wary of lock-in, and I think they’re right to be. They want systems that can grow and change, and that can connect to whatever arrives next, including AI capabilities nobody has built yet.

Interoperability is the groundwork for AI-ready buildings and smart spaces in general. Intelligence will keep getting better, but it can only do useful work in the physical world if the devices beneath it are connected, open and able to cooperate. Getting that foundation right should be our first priority as an industry.