Turning Building Data Into Better Decisions With Willow

[00:00:00] Do you struggle to keep your AI agents from acting outside of compliance? Well, Denodo provides an AI data layer connecting your data systems, keeping guardrails consistent across all of your data platforms. So start scaling your business carefree with Denodo. And you can do that by visiting denodo.com to learn more. But now, on with today's show.

[00:00:26] What if the smartest person in your building turned out to be the building itself? Well, if you work in a facilities team, you might begin your morning opening 17 browser tabs, checking alarms, chasing work orders, and deciding which problem deserves attention first. Yep, welcome to the world of alert fatigue.

[00:00:51] Well, my guest today is Bert Van Hoof, CEO of a company called Willow. And he returns to the podcast to talk about how operational AI can turn the overload into practical decisions. We will discuss why a dashboard that reports 500 events is less useful than a system that identifies the three that actually matter.

[00:01:17] And also explain why and recommends an action and checks whether it worked afterwards. And you'll also hear around predictive maintenance, digital twins, occupancy data, human judgment, and essentially how all these things can work together across hospitals, airports, universities, and offices. So if your workplace already has way too many dashboards, good news. My guest today has no ambition to become tab number 18.

[00:01:47] He's got a much better approach. But enough from me. Please join me in welcoming Bert straight back to the podcast. So a massive warm welcome back to the show. For anyone that missed our previous conversation, can you remind them a little about who you are and what you do? Yes, absolutely. A thrill to be back with you, Neil. We had a lot of fun last time. So I'm Bert Van Hoof, CEO of Willow.

[00:02:14] Spent the last, well, spent about 20 years, not the last, about 20 years at Microsoft. And towards the latter part of my time there, I've worked on the intersection of IoT, AI, smart spaces, and the physical world, including launching Azure Digital Twins. So a lot of my career has really been about how we represent and understand the physical world in software and increasingly how software can help us operate that world. So that's ultimately what brought me to Willow.

[00:02:45] Willow provides what we call operational AI. I'm sure we zoom in on that a little bit more today for buildings and critical infrastructure. So we work across all kinds of environments like airports, hospitals, universities, corporate campus settings, and our data centers and large commercial portfolios. So I'd say for the long time, the challenge was getting the physical world into software.

[00:03:12] And what's really exciting now is that AI is beginning to reverse that relationship. Software can increasingly understand what's happening in the physical world. Reason about it and help us change the outcomes. So that's, in a nutshell, what I do, how I got here, and a lot of passion on the subject. That's as we discovered last time. Yeah, and that passion always comes through.

[00:03:37] And one of the reasons I was excited to get you back on that today is everything that you were just talking about there. I always try and get people thinking differently about technology in their world and work and lives. And when we talk about facilities management, it isn't usually the first area people think about when they hear AI at work. So what is it that makes building such a compelling use case for AI?

[00:04:02] And why is this moment, do you think, different from previous attempts at smart buildings that listeners might be aware of? Yeah, yeah. Fascinating question. At least it's that passion. So it directly attaches to our mission here at Willow, where we want buildings to respond to the people, purpose, and environments they serve. Right? That's something I always like to unpack because people live, work, and play in the physical world. Right?

[00:04:29] All the activities we do as humans are directly connected to the physical structures and the related infrastructure. So all buildings and the environments within those serve a specific purpose. Right? So that's where that triangulation between people, purpose, and environments is so important. And so when you say AI at work, people typically think about the continuation of digital transformation. Right?

[00:04:57] Which we've been into for decades. Digital innovation and digital transformation has lived in the world of bits, as we like to say. Right? Software, language, and data. Now AI agents and co-pilots today mostly live there still. Right? Data comes from web pages, social media, emails, spreadsheets, documents, code repositories. Right?

[00:05:23] We know all the frontier models have now indexed all the world's information that's publicly accessible to the internet. And then you bring that into corporate environments. But, you know, specifically LLMs who created all the breakthroughs, they understand language. Right? The LLM is large language models. So, but the next frontier of AI, which is getting a lot more attention as of late, is what I would say the world of atoms. Right? The physical world.

[00:05:54] And that's where intelligence is going to meet reality. Right? Buildings, infrastructure, equipment, grid. And so buildings are actually incredibly compelling environments for AI because they're enormously complex physical systems. Right? Operational AI is not just another digital layer.

[00:06:14] We see that as the unifying intelligence that's connecting the physical, operational, and the human layers of the built environment. Hence that mission statement as well. So, think about an airport, a hospital, university, campus, all kinds of environments we're active in.

[00:06:33] You've got HVAC systems, electrical system, water, elevators, lighting, occupancy, energy, work orders that are being executed, weather schedules, thousands of pieces of equipment. Some of them related directly to the building. Some of them related to the business operations, the activity that takes place within the building. And so all of that can generate potentially millions of data points every day.

[00:06:59] And all of that is in service of the activities that take place within those environments. Right? And that's serving the mission and purpose of the organization that uses those facilities. Right? That's ultimately what it's all about. So, we spent the last 20, 30 years collectively as an industry digitizing all of that.

[00:07:24] We've deployed building management systems, sensors, IoT platforms had a lot to do with that world as well. Meters, BIM systems and models, cloud platforms now, and digital twins. So, we've become quite good collectively at representing buildings digitally. But representation doesn't mean understanding. Right? So, a sensor can tell us that a room is 74 degrees. Right?

[00:07:53] A building management system can tell me that a valve is 60% open. But neither of those aspects inherently understands whether that's good or bad. Right? Why that's happening. What is likely to happen next. And what we should do about it. So, that's fundamentally different about this moment in time. Right? Where AI can begin reasoning across all of the systems and the relationships between them.

[00:08:22] And so, importantly, we don't need to rebuild the world's buildings to do it. Right? Most of the buildings that will still exist decades from now already exist today. Right? So, this is not like we're not talking about futuristic purpose-built modern buildings that have all these capability. The opportunity is really around making the infrastructure we already have dramatically more intelligent.

[00:08:48] And so, as I said earlier, most of the first wave AI has been about information, language, images, code, knowledge. Which I think one of the next great frontiers is what happens when intelligence encounters the physical world. So, that's kind of how I would frame that up to answer your question. So, yeah. Completely. That's the trick. Yeah.

[00:09:14] And at the very beginning of our conversation, you described Willow as delivering operational AI rather than simply just another dashboard. So, for people listening, what's the difference? How does AI help facilities teams make those better decisions instead of just giving them more data and expecting them to act on it? Yeah. Yeah. No, that builds nicely on top of the last question and answer. Right? I think there's a pretty simple way to distinguish the two.

[00:09:41] A dashboard essentially says here are 500 things happening in your building. Right? Where analytics might say here are 20 things that look unusual. Right? With operational AI, we should have the ability to say, well, there are three things that really matter. Here's why they matter. Here's what's likely to happen next. And here's what I recommend you do about them. Right? So, that's a very different next-gen approach to things.

[00:10:08] So, increasingly, it can help take that action and then verify whether it actually works or not. Right? So, there's a progression. What's happening? Why is it happening? What's likely to happen next? What should we do? And then, importantly, did it work? Right? So, a dashboard represents representation of reality. Operational AI reasons about that reality. Right? So, that's really a big distinction.

[00:10:35] I'm a big fan of a company called ProSci. Created a whole framework to say, hey, here's the things that humans will always want to be involved in or keep control over. Here's where AI can work with me, with the human. And here's what AI can do for me. Right? For the humans and for the environment. That's an interesting framework. And it's a fundamentally different proposition. Right?

[00:11:04] Today, a lot of what's happening is, you know, co-pilot became a concept, not just a brand. Right? People talk about co-pilots. So, AI is helping make humans better decisions. And there's the agentic part of it, which is a different thing where you can start to delegate certain aspects. But over time, as trust develops and appropriate guardrails are established, more of that can become more autopilot. Right?

[00:11:31] More of that agentic approach, multiple agents working with humans, with each other. And there's all kinds of governance aspects, et cetera, that come into the mix. Autonomy, by the way, is not the objective by itself. Right? The objective has to be better operational outcomes. Right? Hence, the term operational AI as well. But that's something we think about a lot and profoundly at Willow in specific. So, the unit of value is not an insight.

[00:12:00] It's an outcome. Right? It's measuring impact. So, it's really a whole cascade of there's the reality. How do we observe reality? How do we represent it appropriately? That's where technology concepts like ontologies come into the mix. What is the meaning of it all? Right? So, that's where a knowledge graph comes into the mix. Understanding. You can start to predict. You layer judgment on top of that, which can become quite complex. Right?

[00:12:29] And then, how do you take action? Verify the action. Learn from that action. And so, you develop expertise and wisdom. And that's ultimately how you get to more things that can start to feel like autonomy or smart delegation or orchestration as well. So, yeah, that's the profound difference between the world of dashboards, right? And the type of things we try to light up and that we see as truly next generation.

[00:12:58] And I suspect we'll have many people listening who have not thought about this before and completely unaware of the scale of what we're talking about here. And how every building generates millions of data points every day. But, of course, data alone doesn't prevent a boiler from failing at 2 a.m. on a Sunday morning. So, how do you turn all of that information into actionable intelligence that teams can actually trust? Yeah, we've been leaning into this nicely, right?

[00:13:28] So, I think the key word there is context, right? Yeah, yeah. Imagine I tell you a sensor is reading 78 degrees, right? By itself, that tells you almost nothing, right? What room is it in? What equipment serves that room? Is the room occupied? What's the purpose of the room? What's the outside temperature? What should the temperature be? What's the air handler doing? Is that normal behavior for the equipment?

[00:13:56] Did something change since yesterday? Was maintenance recently performed, right? So, all of those pieces of context. So, once you understand those relationships, that individual data point suddenly starts to mean something, right? So, data acquires meaning through context is an important concept here, right? So, that's fundamentally what technology layers like, ontology, knowledge graph, digital twin, spatial temporal world models, all complex sounding terms.

[00:14:25] But they serve technologically. Each layer serves a purpose within that progression of representation, meaning understanding context and dynamics, right? That's the live data coming in. Reality is different every second and minute of the day, right?

[00:14:43] So, it creates an understanding of the relationships between spaces, equipment, systems, sensors, operational information, and governance and stewardship, right? So, one of the things I learned very early working with digital twins is that the relationship between things can often be more valuable than the things themselves, right?

[00:15:08] So, AI can now reason across those relationships and context at a scale that simply wasn't practical before, right? So, then there's another important piece, which is trust and trust building, right? In the physical world, being clever just isn't enough, right? You have to be trustworthy. If AI generates a bad paragraph when you're doing a document or something, that can be annoying.

[00:15:34] If AI makes a bad decision about a hospital or a critical piece of infrastructure, that has real consequences. So, people need to understand why a recommendation was made. The AI needs to be grounded in real operational data. There need to be not only appropriate guardrails, but understanding of governance and stewardship, right? That's where the human interaction is important or how that's configured and set up.

[00:16:00] When an action is taken, the system should verify whether the expected outcome actually occurred, right? So, that's also how we should think about autonomy, right? That's not simply switched on. It should be earned. You're seeing that progression, like, even with concepts, Excel driving, right? Like, you go to an autopilot, you watch it all the time, you know, just simple lane changes as that now has evolved

[00:16:27] with some of the players into something quite different. And it's still mind-blowing to people when they get that experience now in a Tesla or Waymo. But you get to that pivot point where there's enough verification and trust that builds up where, you know, evidence starts to show that it can be safer than human driving at some point, right? So, it's a similar progression in this world, right?

[00:16:57] That's where it's also about the physical world. So, there's a ton of interesting overlaps there. But, yeah, the trust has to be earned. Autonomy is not always the answer. It comes back to the with me, for me, right? It's what are the mind-numbing activities that are proven to be capable of thoughtful and accurate execution by agents, and at which point do you let that go. But that's certainly all part of an evolution here.

[00:17:28] Yeah. And predictive maintenance is something that we've been talking about for years now. But the reality still inside many organizations is very much reactive maintenance still. So, what do you think has changed to make predictive maintenance practical today? And what kind of business outcomes are customers seeing? I think it'd be great to bring to life what we're talking about with some examples here. Yeah.

[00:17:54] Well, predictive maintenance, as you say, isn't really new conceptually, right? What's really new is our ability to make it scalable, right? If you look at it, historically predictive maintenance often required specialized centers, sensors, bespoke models, data scientists, expensive integrations, and then somebody still had to interpret the results, right? So, the economics may have made sense for jet engines and turbines,

[00:18:21] but it's much harder to make that work across tons of thousands of relatively ordinary assets in buildings, right? So, several things have changed. We have far more operational data available. Cloud infrastructure makes it a lot more accessible. Digital twins, knowledge graph, associated anthologies give the data context, which we touched upon extensively already. AI dramatically reduces the cost of interpreting what's happening.

[00:18:52] So, generative AI makes the resulting intelligence much more accessible to ordinary operators, right? And agentic AI increasingly allows us then to connect the insight directly into operational workflows. So, I think there's an important distinction as well between prediction and operational judgment, right?

[00:19:14] The facility managers doesn't really want an algorithm that tells them that pump has 73% probability of failing. They need to know what's wrong. Why do we think that? What happens if I do nothing? How important is this relative to everything else I'm dealing with, right? This is where impact and impact scores come into the mix as well. What should I do next and when should I do it?

[00:19:42] So, prediction without context, understanding consequence and action doesn't solve the operational problem. So, when you put all those pieces together, you can move from reactive maintenance towards predictive and ultimately much more proactive operations, right?

[00:19:58] So, that means less unplanned downtime, fewer emergency interventions, better utilization of technicians, longer asset life, lower energy consumption, better capital planning. All these things start to connect into each other. So, again, that's a much broader picture. And so, yeah, there's fascinating results with that as well.

[00:20:23] What we see now too is, I don't remember when we talked about this last time, but people do, when we talk about capital planning, for instance, people do facility condition assessments, right? And so, those are traditionally very expensive exercises where you have humans running around with clipboards and iPads and basically inspecting assets and building up a condition index across the facility.

[00:20:52] Now, that can be very expensive as mentioned. And then a big report gets generated out of that and has useful data to get done every three to five years, something in that vicinity versus with a live living representation like a digital twin that has always up-to-date information about it, right? Where you can just be a lot better informed in real time.

[00:21:23] Now, a facility condition index could cover things like I have a hole in my fence or my tree on the parking lot died. That's not something we monitor on the willow side. But for all the assets where it really matters and where we now have live data, we can easily cover those kind of components as well. So, yeah, that's the whole gradation then of reactive maintenance into proactive.

[00:21:51] Now, sometimes those assessments will say, well, this is the reasonable lifespan of this type of equipment, right? But that equipment may still be humming just fine. We monitor it on an ongoing basis. There is all the predictive ability to that. So, why would you proactively replace that piece of equipment? You may squeeze another two, three, four years out of it, right? And it's doing just fine.

[00:22:19] So, those all become useful, interesting concepts where you can also save yourself a lot of money without necessarily risking unintended or unwanted downtime, right? So, it's a broader spectrum of this whole reasoning ability, understanding all the relationships, understanding all the objectives, understand the purpose, understand the risks, right? And not just looking at pure data points.

[00:22:48] So, yeah, I think that kind of covers it. And we see that in a lot of examples, whether it's refrigeration, where in retail settings, if one of those units go down, a whole rack with food in it, you know, there's all kinds of regulations that if that unit can't hold its set point, it goes out. The clock is ticking before you have to throw the food.

[00:23:12] You can have the same situation in distribution of meds and vaccines and things like that, or the content is even more valuable. So, these things translate across industries as well. And so, we have lots of examples of that. There's things like water leaks or other liquid leaks. You know, obviously, predictive is the best, right? But also, quick detection is very important.

[00:23:40] If something gets detected timely, then it may be a very cheap fix with a muddle mop up versus it goes on all night. But now you have a flooded space and now you have to remove all the furniture and the big blow dryers come in and you have to do mold prevention and a $300 problem escalated into a $30,000 problem overnight. Right. So, yeah, so there's a lot of richness around that topic.

[00:24:07] And that's a big trend in general as well around the move towards condition-based maintenance, which ties into everything I just covered there as well. And another concern with AI in many areas is that frontline workers can end up with yet another system that they have to learn. And many of these individuals on the frontline are not technologists and they feel a bit overwhelmed by you must use this now.

[00:24:33] So, how do you design AI that genuinely reduces complexity for those frontline workers rather than just adding another layer of technology for them to manage, which they may feel is unnecessary? Yeah, look, I mean, this is something we've lived through many times where people would tell us operators say, I come in in the morning, I open 17 tabs in my browser for all the systems I need to monitor. Yeah.

[00:24:58] And then, you know, we always make the plea that we don't want to be tab number 18, right? And so, and that's not a fictitious number. I mean, I could quote real customers that some of them then would say, oh, that's understating it. We got more than that. But so it's an incredibly important point because facilities teams already have too many systems.

[00:25:22] And they have building management systems, work order systems, energy platforms, sensor platforms, drawings, documents, dashboards, which we just covered as well. So the answer just cannot be another dashboard someone has to stare at. So, and historically, people have been forced to understand the way machines and software organize information. And with all the abilities that we just talked about, we now have the opportunity to change that, right?

[00:25:50] You, you no longer need to know which application to open. Today, you need to know what an obscure BMS point actually means. You need to know where the manual lives. That's a bigger problem than people may think, which is why we ingest all these source documents, including operational O&M manuals. You need to know which database contains which answer, right? AI gives us that opportunity to invert that relationship.

[00:26:17] But the technician should be able to ask, why is the third floor hot? Or what are the five things I should worry about today? Or simply what changed overnight, right? So those are important things where these new abilities can allow you to do that. So the technology should work out where the information lives and how it fits together, right? That should go through these questions. And that's why we see, we saw things when they were introduced, like a daily briefing in the copilot.

[00:26:47] It became a very interesting entry point for people, right? So don't ask an operator to spend the first two hours of the morning searching for problems in the 17 tabs, right? Let the building tell them what changed, what matters, and where they should focus their attention, right? So that's really what we're doing. And that's really appreciated by people. They can also go on discovery journeys of things they would want to know, right?

[00:27:17] They can tailor this and switch it to their needs. Then turn that into briefings that can also be sent around to other people at a specific schedule. So all of those are super popular features, right? So this is, again, let the AI work with me and for me. So AI gives machines the opportunity to learn the language of people now, right? And ultimately, the promise of AI shouldn't be another application. It should be fewer applications you have to think about.

[00:27:46] So I think we definitely have already arrived there in a big way. I'm curious, if we look beyond preventing equipment failures, where else do you see the biggest opportunities for operational AI? I mean, it could be areas like energy efficiency, sustainability, workplace experience, reducing operational costs, or something entirely different. Where do you see the biggest opportunities here?

[00:28:11] Well, I actually wouldn't think about those examples that you're given as completely separate problems, right? A building is one interconnected physical system. Occupancy affects ventilation. Ventilation affects energy consumption. Equipment operation affects maintenance. Weather affects load. Energy prices affect operating strategy, et cetera, right?

[00:28:38] So all of those things ultimately affect carbon, comfort, cost, right? So historically, we've put those things in different software systems because that's how we've been able to manage the complexity of it. But the building doesn't know that we've put energy maintenance, comfort, or sustainability into different applications, right? They're all part of the same physical reality. So there are good practical examples, right?

[00:29:04] Why heat and cool is space as though it's fully occupied when nobody's there, right? That's become a very popular feature. And agentic comes into the mix because you start to command and control, et cetera, and you save a ton of energy doing that. And there's reasoning, right? How do you determine some things occupied? You want to be anticipatory so you can look at schedule systems in a hospital that can be an EHR system like Epic.

[00:29:33] Is there a patient in that bed? Is there a patient scheduled to be in that bed? And in education, you have classroom scheduling software, right? Which classes are taking place where in a regular office setting you would use Microsoft Office or 365 or Google Calendar, et cetera. So all of those are signals, but then that can be augmented with real-time occupancy sensing and people count.

[00:30:03] And so it can be both anticipatory. I'm going to make sure that the room is prepared for something that's scheduled. But if for whatever reason that thing was canceled and nobody shows up physically, we can detect that through other means and then still do a setback, et cetera. So we see situations where there's simultaneous heating and cooling different parts of a building because two systems aren't aware of each other.

[00:30:30] Can I dynamically adjust ventilation based on actual occupancy or air quality measurement? Can I detect abnormal water consumption before a major leak is detected? Can I shift electrical loads across periods of peak demand? Can I optimize equipment schedules based on tomorrow's weather rather than yesterday's static schedule? Can I take a look at the actual level, right?

[00:30:54] So increasingly with technologies like Willow's active control, we can move beyond simply identifying the condition and actually change how the building operates within certain defined limits, of course. So that's important because the real opportunity isn't optimizing every system independently. It's optimizing the outcome of the whole, right? That's why I reacted to we have to think about it in that way.

[00:31:20] And so I also think that the traditional separation between like sustainability and economics is increasingly disappearing, right? Wasted energy is wasted carbon, but it's also wasted money. Operation AI gives us the opportunity to attack all of that simultaneously. So, yeah, that's the way to think about it. Yeah, 100% with you.

[00:31:45] And as buildings will inevitably become more connected and in many cases increasingly autonomous, what new skills should facilities managers need now? And how do you see the role evolving as AI takes on some of those routine monitoring and analysis, etc.? How do you see that evolving? Yeah, I mean, first of all, I should say I don't think the story here, which is always a concern these days,

[00:32:14] it's about replating facilities expertise, right? And in many ways, I see it as amplifying it. We all know the best building engineers have extraordinary institutional knowledge. Typically, they can walk into a mechanical room, hear something that doesn't sound right, put their hand on something, feel something that's not right. So, they know the quirks of equipment that's been operating for 20 or 30 years. We do have an aging facilities workforce.

[00:32:43] That's a problem. We know that. So, a tremendous amount of that expertise is currently locked into people's heads. So, AI gives us an opportunity to capture, democratize, and ultimately compound that knowledge. So, at the same time, I do think the nature of the job changes. Today, a lot of facilities work is about monitoring, hunting for that information, diagnosing, and reacting.

[00:33:10] So, AI can increasingly take on some of that continuous observation and routine analysis, right? And that moves the human role towards prioritizing, validating, optimizing, orchestrating, ultimately more rewarding stuff, right? And have that ability with me and for me concept to ultimately towards something even more important, setting intent, exercising the judgment, right?

[00:33:40] Judgment is very important. And deciding what good actually looks like, right? Because AI can optimize against an objective. The humans will still have to decide what objectives matter because that can be, again, a pretty complicated picture. And AI has amazing capabilities there, but that will have to be attuned to the specific environments and where these experienced operators come into the mix as well.

[00:34:08] And, yeah, that's all the trade-offs. Should I optimize entirely for energy? That's a myopic focus that's often not healthy. What about comfort? What about resilience? What about equipment life? What about safety? Those aren't purely computational questions, right? So as machines take on more observation and routine decision-making, I think the human role moves towards setting intent, exercising that judgment, and managing those trade-offs.

[00:34:37] And in that sense, the more capable the technology becomes, the more important human judgment becomes, not less. So, yeah. I'm going to have a bit of fun with you now. If I pull out my virtual crystal ball, I'll ask you to look ahead. It's going to be fun now. If we were to walk in a commercial building three to five years from now, and I appreciate it is a crazy question because the level of change we've seen in the last three to five years. Yeah, I think it's going in weeks now instead of years. Yeah.

[00:35:07] It really is. But, I mean, what do you think will be happening behind the scenes that most occupants will never notice, but which will completely transform how buildings are operated and maintained? How do you see that evolving? Well, I mean, yeah, that's a good one and loaded. And as you said, right, like three to five years is crazy.

[00:35:31] But, well, I think the interesting thing is that the building may not look futuristic at all, right? Like it may simply or will simply work better, right? Behind the scenes, those thousands or millions of signals will continuously be interpreted, right? We're already there in some ways. But any equipment degradation will be identified before failure.

[00:35:56] All the predictive aspects we just touched upon will be democratized and in full display, so to speak, or deployment. HVAC will respond dynamically to occupancy. We're doing that today, right? But these things will become just commonly built in. It will respond to weather. It will respond to changing conditions. It will respond to air quality issues.

[00:36:26] Energy loads will be continuously optimized. That's going to be a big thing. You know, the whole issues around grid and data center build out. We all know those stories at this point. In the US in particular, there's a grid that has lots of needs. Buildings are going to be important to that. In Europe, in specific as well, we all know the geopolitical situation and all the unique needs there.

[00:36:52] Things like the EPPD, the Energy Performance for Buildings Directive that's unfolding in Europe. So that will be an important aspect and where the buildings and associated infrastructure will play a key role. You know, we joked in a shared book. It's called the Active Efficiency Collaborative. But demand is the new supply, right?

[00:37:15] The cheapest electricity you can get is the one you don't consume or that you can reply it into the system, so to speak. So that's important. Water anomalies will be identified before they become damaging leaks. All kinds of things we touch upon. Work will be automatically prioritized based on consequence rather than simply alarms or schedules. We're already going there. But I see all these things as just becoming common sense. Everybody's doing it the new way, the right way.

[00:37:45] A technician may arrive already knowing the likely problem, right? We do all the pre-diagnostics now and stuff, the history of the asset, what they need to do, bring the right tools will become commonplace. Increasingly, any routine interventions will just happen autonomously without, well, within the proper guardrails. And with AI verifying that the action actually produced the intended result.

[00:38:11] But all of that, I think, is in many ways just still half of the story, right? If we look that far out and seeing other trends that are already emerging with our customers. Because we're still defining the building largely by the systems that make the building itself operate, right? The physical structure. Traditional smart building technology tends to understand the assets that are part of the building.

[00:38:39] Air handlers, chillers, boilers, electrical systems, lighting, water, elevators, meters, sensors. So essentially all the equipment attached to the building, embedded in the walls, sitting above in the ceilings. And all of those things are obviously important. Then we can get a lot of savings and carbon improvements and all of that. We'll continue to put that on steroids. But they're not the entirety of the physical world inside the building, right?

[00:39:07] So if you take a hospital, the physical world of a hospital doesn't end with the air handler serving a patient room, right? So inside that same physical environment, you may have smart patient beds, medical equipment that may be rolling around. Hand hygiene monitoring systems. Autonomous cleaning robots and their charging stations. Refrigeration for meds and vaccines, which we touch upon.

[00:39:35] RTLS safety and security systems for some of the workers and the nurses. Pneumatic tube systems. Many other pieces of operational equipment required to actually deliver healthcare, right? So those are traditionally thought of as smart building assets. And they often belong to completely different technology stacks and different departments. But in the physical world, they're all participating in that same environment. So that's important.

[00:40:04] At Willow, we don't think the digital understanding of a building should stop at the equipment attached to its walls and the ceilings, right? We can bring those mission-specific assets into the same physical world context, understand what they are, where they are, how they relate to the spaces, systems, and the other assets around them, right? So there's another important distinction there as well.

[00:40:27] Like we deliberately talk about environments in our mission statement as well, plural, because a building isn't really one environment, right? Think about that same hospital. A patient room is one environment. An operating room is an environment. A waiting room, a laboratory, a kitchen, the parking, and on and on, right? Then you could bring that to any building pack in different industries. They may all be part of the same building, but they serve very different purposes, right?

[00:40:55] So each environment has a different combination of space, building systems, operational equipment, people, and conditions. So what good looks like is effectively different in each one, right? So an operating room and a parking garage may be a part of the same building, but they're very different environments serving very different purposes. So intelligence has to understand those distinctions, right?

[00:41:21] And so once you truly understand the physical space, the building system serving it, the operational assets within it, and the purpose of that environment, then you can begin connecting all of that with business operations, right? So if we stay with the hospital for a second, what procedures are scheduled tomorrow? What operating rooms will be required? What equipment needs to be available? Is the physical environment ready?

[00:41:48] Are the right conditions being maintained, right? Is plant maintenance going to affect tomorrow's clinical schedule? Can cleaning and maintenance activity be orchestrated around clinical operations? So the question really stops being simply, is the air handling operating correctly, right? It becomes, is this physical environment ready to do what the hospital needs it to do? So that's a fundamentally different question. So I think that's really the crux of it all.

[00:42:17] So you're no longer simply optimizing the operation of the building. You're using understanding of the physical world to help optimize the operation of the business that's taking place inside it. So I think that's, that's a big thing. You can, you can apply exactly the same things or ideas to, to airports, right? Just don't understand HVAC and occupancy, understanding flights, gates, passenger flows, anticipate where people are going to be, what those environments are going to need.

[00:42:45] That can influence cooling, cleaning, maintenance, energy, many other operational decisions, right? So you can extend that into educational settings, et cetera. So, you know, they're, they're all, all, all these systems are participating in, in the same physical environment for the same underlying purpose. And so our software systems have historically seen those as completely different domains. And in the physical world, they're all participating in the same environment.

[00:43:15] And so AI gives us the opportunity to finally reason that way. And, and that comes back to, I'll close it out, a long-winded answer to a far-reaching question. But it comes back to the will of mission, create a world where every building responds to the people, purpose, and environments they serve. Those words are quite deliberate. I know that's where we started, but people who is actually using these spaces, what do they need?

[00:43:42] Purpose, what is the place actually there to accomplish, right? A hospital exists to care for patients, an airport to move people, a university is there to educate, a data center to provide reliable computing infrastructure, right? And then the environment is plural. Many distinct physical spaces in which that purpose is actually carried out, each with its own systems, operational assets, people, conditions, and requirements.

[00:44:09] So if you only understand the equipment attached to the building, you understand the infrastructure, but you don't yet understand all of the environments or what they're there to accomplish. So five years from now, I think the really interesting building is simply one where the HVAC runs itself. It's one where intelligence understands the physical environments in which the organization operates, the building systems, the operational equipment, the people, and ultimately the purpose.

[00:44:38] And then it will increasingly help orchestrate all of that. So that's where building operations start to connect with business operations. So ultimately, that's what we mean when we talk about creating a world where every building responds to people, purpose, and environments they serve. And I think that's where we're headed on that journey into the next three or so years. A lot of that is life today. We're here to democratize this, and we believe ultimately all buildings will have those type of capabilities.

[00:45:10] Yeah. Wow. Exciting times ahead. And we've got such an in-depth conversation today and still only scratched the surface of what's possible. So where is the best place for listeners to find you or your team online? And dig a little bit deeper on anything that we've talked about today. Where should they go? Yeah. Easiest place is go to our website, willowinc.com. So there's a lot of new content all the time.

[00:45:36] There's a very rich blog there as well that people can look at very specific deep dives as well. It keeps being built out. So that's definitely a good place to go learn about what we're up to and some of the concepts that we touched upon here. So I also remain pretty active on LinkedIn. So that's probably the easiest place for people to find me personally and continue on if the conversation for those that are interested.

[00:46:05] Well, we did cover so much there from why AI should be used to support frontline workers, why it isn't just another layer of technology for them to manage, and also how operational AI can help teams cut through noise, prioritise faster, and take action with more confidence. And anyone looking for the return on investment here, I think the benefits and cost savings capabilities of leveraging predictive maintenance and other aspects of operational AI really speak for themselves.

[00:46:35] So I will have links to everything that you mentioned there. I encourage everyone listening to go check it out. Go check out yourself on LinkedIn as well. I know you're quite active on there. But more than anything, thank you for joining me once again. Really appreciate your time. Thanks for having me, Neil. Always a pleasure. I think Bert's argument reminded me that autonomy should be earned through evidence, context, and verified results.

[00:46:59] A clever recommendation is useful, but a hospital, airport, or university still needs to know why it was made and whether the action actually worked. Operational AI offers facilities teams something practical. Fewer tabs, earlier warnings, better priorities, and less time hunting through systems for answers. And the human role refreshingly remains central because people decide what looks good,

[00:47:27] what good actually looks like, and balances comfort, safety, resilience, cost, and energy use. So a massive thank you to Bert, as always, for returning and translating intelligent buildings into a language that the rest of us can understand. And you can find Bert on LinkedIn. You'll learn more about Willow at willowinc.com. But over to you. Can your building already hold the data needed to work smarter? But over to you. Let me know your thoughts.

[00:47:57] I'm easy to find at techtalksnetwork.com. But that is it for today. So thanks for listening, as always. Bye for now. Bye for now. Bye for now. Bye for now. Bye for now. Bye for now.