46 min read

🎧 #101: Virtual Engineering with Hank

"Virtual engineering is an inevitability. It's not a, 'there may be a breakup of this vertically integrated stack, this monolithic stack that everybody owns.' It will happen."

—Zach Denning

Welcome to Nexus, a newsletter and podcast for smart people applying smart building technology—hosted by James Dice. If you’re new to Nexus, you might want to start here.

The Nexus podcast (Apple | Spotify | YouTube | Other apps) is our chance to explore and learn with the brightest in our industry—together. The project is directly funded by listeners like you who have joined the Nexus Pro membership community.

You can join Nexus Pro to get a weekly-ish deep dive, access to the Nexus Vendor Landscape, and invites to exclusive events with a community of smart buildings nerds.

Episode 101 is a conversation with Zach Denning, Co-Founder of advanced supervisory control software startup, Hank, which was acquired by JLLT earlier this year.


We talked about Hank’s founding story, how the product works and how it improves control system performance, why JLLT acquired Hank and how it fits into their broader product strategy, and how the Hank team approaches some of the challenges on the people side of implementing advanced supervisory control.

Without further ado, please enjoy the Nexus podcast with Hank Co-Founder, Zach Denning.

  1. Hank (0:38)
  2. JLL (0:40)
  3. Blue Ocean (50:14)
  4. A Guide to the Good Life (51:19)

You can find Zach on LinkedIn.



  • How the software layer improves upon the existing control system (6:42)
  • Building re-engineering (8:35)
  • Combating the diminishing returns argument (14:33)
  • Demystifying machine learning (25:11)
  • How Hank fits into JLLT's product strategy (34:53)
  • Dealing with skepticism and the desire to control things manually (42:19)
  • Carveouts (50:01)

Music credit: Dream Big by Audiobinger—licensed under an Attribution-NonCommercial-ShareAlike License.

Full transcript

Note: transcript was created using an imperfect machine learning tool and lightly edited by a human (so you can get the gist). Please forgive errors!

[00:00:03] James Dice: hello friends, welcome to the nexus podcast. I'm your host James dice each week. I fire questions that the leaders of the smart buildings industry to try to figure out where we're headed and how we can get there faster without all the marketing fluff. I'm pushing my learning to the limit. And I'm so glad to have you here following along.

[00:00:31] James Dice: This episode is a conversation with Zach debting co-founder of analytics and advanced supervisor control software startup, Hank. Which was required by JLL T earlier this year. We talked about Hanks founding story. How the product works and how it improves control system performance. Why JLL acquired Hank and how it fits into their broader product strategy and how the Hank team approaches some of the challenges on the people side of implementing advanced supervisor control. So without further ado, please [00:01:00] enjoy the nexus podcast with Hank co-founder Zach Denning.

Hello, Zach, welcome to the show. Can you introduce yourself?

[00:01:06] Zach Denning: Yeah, exactly. I mean a head of product sustainability with JLL technology. Cool.

[00:01:13] James Dice: That's a, that's a recent role for you. Can you talk about maybe start when, you know, start with your background educational background and how you sort of, move throughout the industry and a pretty diverse, diverse background.

Can you talk

[00:01:26] Zach Denning: about how you got. Yeah, yeah. Very very new role. Yeah, so we, you know, my background mechanical engineering and I always had a knack for software. So I did freelance in college and I thought I'd get into the industry, just doing something where I can combine the software with kind of my knowledge of, of mechanical engineering and HVAC.

And so kind of stumbled into building automation. I really liked it. I like the software side a lot and I was always looking for opportunities in software. No, I became a fairly accomplished at the software side and kind of realize there's only like two [00:02:00] paths in building automation to go. It's either you go into project management, you go into sales.

And as much as I like to software, it's like, okay, I have to choose a path. So I chose sales for awhile and it's very technical sales. And a lot of times it was integration. So I was really setting up a lot of integrations. You know, how do you connect to some legacy system on windows XP? And so it was really value-driven integrations that I stumbled into.

At that point I got really good. It looks around a little bit, started doing some project management. I'm spearheading my own projects. I started control division for mechanical contractors that was fairly large as a bay area in San Francisco and, and started kind of looking at controls a different way.

When we started looking at it more like energy services projects, where we weren't focused heavily on, I think where the industry was headed with the hardware side is like, what hardware can we sell? And how has that hardware and cap in a generator? For me, it was, how do we implement the least amount of hardware, right?

The best software and generate return. Um, I was able to partner with local utilities, get rebates and then also partner with some third party financier [00:03:00] to help finance the project. And we ended up with these great returns a year, less painful. On three or four of these pilots that we ran there, they were really awesome.

Went to work for a OEM for awhile, kind of learn that side of the business in partner channels and distribution, which was great. And then I brought it all together and, and started Hank at the time it was under a different name, but, you know, we really launched this premise that. Now you can take software and actually build a business around it in this industry, which I think is, is massively challenging.

I think a lot of people have tried it and it's gone, it's gone, I'd say in some cases very successfully for 11 other companies, but my thing was always like applying machine learning. Can, you know, can you take machine learning and can you actually apply it in the industry for more than just monitoring?

Like how do you get it to that kind of that next. And so, you know, I, I I went out tried to sell analytics like everybody else, except mine was going to be, you know, machine learning to generate a natural level, two audit went outside to sell it found that there was a very niche market that is willing to pay for that.

Long-term over, it's as honest as product and to kind of reel it back in, uh, one, the California energy commission grant [00:04:00] and the whole premise was can we take this machine learning that we've really used for monitoring an annual. And apply it to a building and not just in a way that it's doing little things, can machine learning, run a building.

So can a tune and run the actual functioning systems of an HVAC system, lighting, whatever Wells. So we could really get our hands on a place to now over the course of about a year and a half, we proved it was entirely functional. Not only that it saves mass amounts of energy and we stumbled it. Some huge breakthroughs with comfort IQ, where you can really expand on the industry and when it's done.

So, you know, grew from there commercialized and you know, really grew into what we know today. Got it. And the acquisition happened when a couple of months ago. Is that right? Yeah. Yeah. We uh, we required uh, late December, early January by jail technology. Can you circle back to what you're talking about with hardware versus software?

[00:04:56] James Dice: W how is the industry so focused on hardware? Can you [00:05:00] explain that a little bit more? And then why is that?

[00:05:03] Zach Denning: Yeah. Yeah. It's, I mean, it comes from like traditional sense, right? Like you started with pneumatics and like the sixties, seventies, and then progressed into electronics and, and controllers and everything.

It wasn't. And I think they never really evolved outside of that. Adjacent industries. You look at like, you know, computers in the nineties. I think the industry, our industry is still stunted in the same way that IBM compact and allow these other companies where it's the assumption that a vertical stack is still plausible in this industry.

It, it works in some cases, there's no doubt about it, but I think that's company specific, not industry specific. You know, and so I think they think this out in that. Some people figured it out the hard way with IBM and compact. And some of those companies kind of learned the hard way, but you had, you know, windows and Dell.

And I think Dell was the recognition that hardware, I, if you build, you know, the cheapest best hardware and you apply one singular software interface to it You can build an entire [00:06:00] stack uh, separately, you can see some new office stack. And I think that we're going to that district distributed architecture is coming.

I think a lot of people don't want it. You know, I think a lot of people would like to say that it's vertically integrated and that everybody can do everything, but I don't think it's plausible. And I think that, you know, in a lot of ways, the OEMs traditionally are still focused on hardware. That's what they.

No, I noticed it five or six years ago. We still talking about IC hardware and Ethan net and buildings. And you go talk to owners and they don't know what you're talking about. And so the value delivery is, is, is very broken. I would say, between what our industry and building automation values versus what the end user, the true consumer of the product, which is the financial benefit is proceeding.


[00:06:45] James Dice: With Hank, you said, okay, let's develop this software layer that improves upon control systems. What were the improvements that you were hoping to make to a control

[00:06:54] Zach Denning: system? Yeah. Yeah. So originally when we got in, it was, Hey, we'll optimize it. You [00:07:00] know, we'll leave everything in place and we can optimize.

And we did that and we did it with, with some level of success, you know, five, 10% total energy savings in buildings. That was kind of like the rev one or alpha. That we went after with the grant and we prove it in a couple of buildings that yes, you can do the traditional sense. And this is like a thing, like maybe what like a building IQ would do, for example, in that sense that yes, you can get in and you can optimize that points, but what you find in the buildings, and this is always what's really interesting is, is, you know, when you look at like, autonomous car, for example, the car has to be able to drive successfully and efficiently from point a to point B one stop sign.

Effectively it asks you to drive up the RPM manages it it's miles per gallon. Stop, break. It has to do basic functions once it can do those functions, then the sensors necessary, which are already in building. To take it from Sacramento to San Francisco. For example, then that becomes a software problem.

That's a machine learning problem. Now you're talking about tolls, Steve's [00:08:00] traffic, all these factors of how do I get from point a to point B extremely efficiently, but I can't do that if I can't get from one stop sign to the next efficiency. And so what you learned in these buildings is you can't just optimize the top layer.

You can't just grab the set points and say, Hey, we're optimizing set points and you're going to save all this energy because the underlying logic, the sequencing, or sub-routine. Most of the time, what we find is they're not optimized to work in the first place. So you have to start kind of at a base layer.

You need to re-engineer the building, and then you can apply machine learning to the top of it with a lot of success. Got it. Got it. What do you mean by

[00:08:35] James Dice: re-engineered the building? Can you break

[00:08:37] Zach Denning: that down a little bit further? Yeah. So, I mean, you, you already have the equipment in the building, you already have the building automation systems.

So you have all the control points. You have all the sensor readings and you have the physical equipment and the space that's already doing something, but does that physical piece of equipment functioning in the proper way? And that's what we look at and say, no, it's not, let's say the economizers open before the cooling starts, right.

If you're not [00:09:00] getting free cooling anymore, and you're just engaging your company. And that's a, that's a sequencing issue. And traditionally that's been tackled by design engineers, design engineers, come in and say, Hey, we're going to redesign and retro commission. They retrocommissioning, they turn it over to a building automation contractor.

What we've done is been able to combine those two functions in Hank, where, you know, we have he's on staff. We have engineers that can take a look at the system to say, this is the best way the system should run. We actually rewrite our own sequences and those sequences are interchanged or intertwined with machines.

So Nigel it's in the sense of like suppliers, hands control, you know, we're not just taking control of and optimizing machine learning. It's actually the machine learning gets more intertwined in the equipment where maybe optimizing a VFD seed or something of that. Okay. So, whereas

[00:09:43] James Dice: I think if we look at the rest of the industry, you know, traditionally you might have a commissioning agent or a design engineer, like you said, come in and do retrocommissioning.

They might use an analytics tool, like an FTD package, right. That might be separate a separate effort. And then you [00:10:00] also might have a, so I call this advanced supervisory control where you have a software, like. Sending supervisory control commands down to the underlying systems. Might have another vendor come in and do that.

And you're saying, okay, this makes sense that we combine these two processes together in one software layer. Is that right?

[00:10:18] Zach Denning: Yeah, exactly. And I think that that was part of, you know, when you look at like the JLL JLC, the big vision is you have all these disparate toolsets. You have all these disparate services without a common out.

And if they do have a common outcome and owner, it's like so obvious gated for an owner of like, how many people do I need just to make my equipment work. Right. Right. Like when we look at, when you look at software, right? Like this is the one thing with our industry, I think is often missed, is it's still just software and it's still just technical.

Right. Our industry loves to believe that we're special, like building automation special cause it's software and technology and buildings and it's not it's just software and [00:11:00] it's just technology. I mean, you have, you have tinkers at home playing with raspberry pies that sometimes it's in case there's more functionality than what.

In a multimillion dollar class, they office highlights, right? Well, they're paying a hundred dollars per square foot, you know, an annual leases. So when you look at it, we're actually a technology category that's significantly far behind the curve. And so when we look at like putting a project together and what we do conditioning is, is technically just cute seeing in software.

So what it's basically saying is the guy who wrote the software, didn't write it properly. And a third party needs to come in with third-party. And they're paid separately to QC somebody else's software all the, while the guy who wrote the software is getting paid to revive his software, that didn't work in the first place.

So it's a really weird industry when you actually look at it from a 10,000 square foot, 10,000 foot level, it gets really odd.

[00:11:51] James Dice: So let's break it down a little bit further in terms of what the product does. So typically, are we just talking about HVAC data? So we're collecting HVAC. [00:12:00] Primarily is that all that data, all the data is

[00:12:04] Zach Denning: collected.

Anything we can get out of the building automation system. So traditionally, yeah, I would say like, you know, half the buildings we work in have windows like. So, it's not like they're going to say control anytime soon. You know, I would love, I love it. You know, I'd love if we guys say control projects, because then you have the whole idea of like, I can turn down how much I turned on the lights versus the shade versus the HVAC.

Right. And you get to like the BTU equations and it gets really sophisticated and fun, but you know, the average building in the United States, dizziness, CO2 control. So yeah, it's a lot of HVAC and that's not to say though that, I mean, we've gotten into very advanced, like chill beam projects with slab heating, cooling she'll beans.

Very sophisticated stuff that, you know, we actually appealed to design engineers about cause they're like you did what, and we started sharing data with them and then it gets very interesting as to the conclusions. The machine learning makes on a, on a day to day. Okay.

[00:12:56] James Dice: And then once the data's collected from HVAC or primarily [00:13:00] HVAC, w what, what do you do with the data and then what sort of commands does it send back down to the system or updates?

Does it send

[00:13:07] Zach Denning: back down to the system? Yeah, we're, we're, we're traditionally in full control of the billing end to end. So I think that's a good place to start is, you know, everything is still running in the building, how it used to, but we are overriding everything. So it isn't full override end to end override And that is something to note is like, we're functionally, rewriting how the building operates.

Almost every time now you can't do that in like a train package unit. You can't like take control of the compressors by any means. But for the most part, we get pretty deep into these buildings and pretty deep into the engineering. We, we, we do a lot of the normal stuff that analytics do. We take the data, we scraped the data you know, push it back to the cloud, run it through a digital twin.

That feeds out, you know, optimal running conditions, whether that be a VFC speed down to a set point, for example. So we do, we do a lot of pretty in-depth machine learning. It's not just really. There might be some functional stuff. This is running standalone over here. That's just functional control or [00:14:00] sub-routine dress.

So that a lot of different names in the industry, but just functioning programming that runs actually on prem on the edge. And then we have corrective actions usually come from the, the AI force and the true machine learning force. And there's a lot of corrective actions and those will be corrected down to like cracking the VFC speed if it, if it doesn't make sense.

So it

[00:14:18] James Dice: could be sequences schedules set up. I call them the three assets you could modify. Any of those are taking full control of all three.

[00:14:27] Zach Denning: Okay. It is a true I building automation, software replacing. Yeah. Yeah. Okay. What

[00:14:33] James Dice: do you say about, I think there's this line of skepticism for these types of tools coming from what I would call like the controls geeks.

That's basically, and I say geeks endearingly, of

[00:14:44] Zach Denning: course.

[00:14:44] James Dice: Oh, they say basically I call it the diminishing returns argument. They basically say like, well, if the system were to be optimized the way it's supposed to be.

[00:14:55] Zach Denning: Then

[00:14:55] James Dice: the system like this, that sits on top of it, software like this, that sits on top of it.

[00:15:00] Machine learning is not really going to get us much else or much savings on top of that. What do you say to that

[00:15:06] Zach Denning: argument? Yeah. Yeah. It's an interesting argument and I had to think it is one word challenging, and I think in some ways it, it does have some signatures. Right. So I used to be one of the geeks who like got ahold of title 24 sequence and programming it.

And then I didn't want it to understand it. Like, why was it written this way? And, and I have a few colleagues in industry that I'd reached out to that are pretty close to the whole title, 24 scene. And I, I begged them for stuff. You know, you, when I started this, it was like, what do you guys want? You know, I want it in my platform, part of my functional program before it gets released anywhere else we found we can deploy software a lot faster, the way we deploy.

And traditionally, I don't have to wait for an OEM to produce an update, but then I've taken. And so we, we can be six to 12 months ahead of the market. A lot of time, the way we deploy, we deploy a lot faster. And so I did it right. And I think one of the things to recognize is in the industry and where it's headed.

The sophistication of software is off the charts [00:16:00] for what one is comprehensible by your average BMS programmer. And two, it is incomprehensible to most building engineers and people to actually have to run the building. When you look at like the title 24 seconds, it was written five years ago. There's what is it?

Over a 600 different tuning set points and a hundred thousand square foot. 600 and that 600 split between I localized controllers, the distributed controllers and like an air handler controller plan, rooftop controls, that's 600 different tuning points. And none of them are documents of course, because like our initiatives and believe in, in, in software and software.

Right. So it's all undocumented. You tune it, however, And for everybody has their own way of tuning. I'll turn this knob and that knob and there's 600 knobs, right? Imagine that borders are coming up and I'm tuning all of them. And I say, it's done. And I walk away. I used to do this. What will happen is either somebody tunes in Navision suppose they're not supposed to, or the knobs tuning don't work for different seasons.

And I promise you. And you're going to have to come back and reach into those nods, but at the time you come back, somebody asleep those not so much for [00:17:00] the current season that they don't even apply realistically for the next season. And so, as a result, it goes off the rails and you, you have to be retrocommissioning or recommissioning, or essentially just rebuilding software that was once built.

And I think that that's where the industry falls down in a lot of ways is not recognizing the fact that everything has to be tuned. The nature of the world, especially with this kind of software, is it static software? And the problem with static software, especially in this context is it's tertiary systems that all rely on each other that all have different tunings, that points.

And so it gets you good, more sophisticated. And so we look at projects less than six months post install, and we're able to drive 15 to 20% overall building energy savings. If you give me a building that's a month or two after, sir, it may drop the 10%, maybe 15. So there maybe there's an argument there, right?

The way in which you look at equipment at different times of the day and how it operates differently. We're just not taking enough into account anyways, from title 24 or from HVAC operation in general to really encompass the entire energy gamut of what a piece of [00:18:00] equipment. I mean, you look at like, you're building pressure.

That's an exhausted supply fan pressure. That's the supply back, you know, supply air, that's the heating and cooling aspect of the equipment. You have all these different energy types that are all interrelated. And what people are trying to tell me is like you've written enough code to understand all those, all these different conditions throughout the day in order to optimize.

Yeah, it's not possible. Yeah. I like to say to people that bring

[00:18:26] James Dice: up that argument, like show me that building, show me that building that's optimized that doesn't need a software layer on top of it. Yeah. How many of, how

[00:18:35] Zach Denning: many buildings have I been in

[00:18:38] James Dice: and audited and crawled through the ducks and looked through the BMS on that?

I've actually been at that point for a significant period of time. It's

[00:18:46] Zach Denning: not very many. When I say, see, the other thing they have to look at too is like the long-term the long-term right? So we're not just like, Optimization and control layer. Like we offer full end to end support on the back end. It's all [00:19:00] layered under a one-time or not one time, but a monthly fee, a fixed monthly fee.

So you look at that too, and they say, well, why do you be software? So it's not just a software, it's a service, right? If any, one of our engineers in our buildings call us and say, Hey, I'm like one foot on a ladder holding on to a dog and I'm looking at a unit right now and I just need it open. We open it for them.

Not only do we open it for them, we tell them the reaction. We've worked with him as if we're an onsite engineer. So we've predicated our model on the fact that our software runs so well. You won't have software issues in your building. Very, very minimal. You'll have so few that we can actually help you support all the mechanical issues in your building.

Which is something where if you go into that building automation industry right now, and you look at some of the service contracts, do they offer hardware support? I mean, we have engineers telling us like pipes burst and they're like, what do I do? You know, like we have one guy where like the domestic hot water pipe burst on it.

It's like, it's like pouring, pouring on him, you know? And he goes, what do I do? And one of our guys who's like find a domestic hot water system. It's probably on the floor is probably in this closet. And he runs and within 15 [00:20:00] minutes he had to shut down. Thank God I can call you guys. And I'm like, who else would he have called here?

The DMS. He had no idea what he's talking about. It's very interesting service.

[00:20:12] James Dice: The physical issues, right? Because a software that sits on top of a control system, it's only going to be as good as the physical components that it's commanding open or closed or on or off. Right. So can you talk about like the value of fixing physical things?

[00:20:29] Zach Denning: Yeah, absolutely. I think this is one of the things that I learned early on doing like machine learning, driven analytics, and again, it's applied. Like I never wanted to build something that was like fluff or, or, or not necessary. Right. So like when I incorporated machine learning in it was, can I go identify a problem before.

You know, can I look at the capacity of an air handler and based on the VFC speed and the flow going through it and the capacity, can I identify that one of your compressors is out before you know it? Now it may be 55 degrees it's meeting set point, everything looks great. [00:21:00] But you have a compressor out or have the compressor that's going out.

Maybe you status on both compressors and you're like, it's working fine, but you're 20% down on capacity from the previous standard, these exact conditions. So there's a problem here. And those are the problems that we try to dig in and, and get to with our customers is there's going to be something that's happening here soon.

You need to go check it out. When you look at buildings and how they operate today, and you look at like why the three pillars, right? IQ comfort. Those are the three pillars we built the product around. And we have them staggered, obviously like IQ first comfort, second energy, third. So, you know, IQ overrides everything followed by comfort.

When you look at like, what are the problems causing for? What are the problems caused by that? Don't allow you to achieve those three pillars. When you start to learn is like 70% software and 30%. And we looked at that across millions of square feet and that ratio is never broken down. Unless sometimes it goes up to like 80 or 90% hardware or an 89% software and 10% hardware.

And so it's a very interesting ratio of, yes, things need to get fixed in the buildings, but oftentimes it's the software that needs to get fixed [00:22:00] first. Not discounting the fact though that, you know, with what we do, we can get in during an audit phase of a building for a new customer. And we can tell them like, Hey, and we did it for one in a huge economizer wall and we're like, it's not worth.

And it was cause they did a new Mac retrofit and forgot to retrofit the it was literally like a barn door size economizer and a 200,000 square foot building one air handler. And, and so we, we quantified it for them, you know, you'd probably be losing in the upwards of, you know, I think it was like three or $400 a month and that's not seasonal obviously.

And so we, we asked them work up a, a payback period. Even as part of our service, we help them review proposals on that to ensure that they got the work done correctly, but you have it full end to end support of property management and asset management from an engineering perspective. So do you deploy

[00:22:47] James Dice: it like a, like a tune-up first and then, you know, let the

[00:22:51] Zach Denning: software to control.

Yeah. Yeah. I mean, when, when, you know, we generate significant returns from the software, right? So [00:23:00] 80% of all your energy issues or 70%, whatever are contained in this opera, there's significant upset. We have very low payback on our products. We have very good payback month over month on the satisfied, like extremely good in terms of, of CRE.

But you know, when you look at like what the why, right. I think that's what the most interesting things is, is like we identified this air handler I in the bay area, we were kind of project identified in your head, good size, they're handling a hundred thousand CFM they're handling and. And it had these old, these old blades that they basically regulated the pressure.

So it didn't have a VFC. And so, the mechanical contractor is the one that we've worked with in the building recommended they put VFDs on it and gave them you know, they gave him the cost. And so he brought the cost up and said like, what's the payback? Well, our system is running analytics. On that unit saying, Hey, it's over pressurized and this is how much money you would say if it wasn't over pressurized.

Well, we can't control it over pressurization. We're doing our best with the software. And so there's this savings that's left over. And what we ended up finding was it was an 18 or 19 year old unit based on the runtime and the [00:24:00] conditions of the unit. It only had about four years of life in it. And there was a five-year payback on the VA.

So the unit really wasn't, it didn't really need to do that much. And so we went back to the owner and we're like, save your money, go spend on a brand new unit. And everybody just looked at us and they're like, you never do that. You know? Like, why would you do that? If you have use are the best payback and, and when you looked at it and you say like, not in this instance now So we provide that level of financial detail down to our customers too, so that it can make a really qualified, like it's not, you know, you hit 200,000 miles on your car and your, your drive train gone.

You know, that's what our industry is used to. If it's quantifiable evidence as to why or why not, you should invest in HVAC.

[00:24:37] James Dice: Hey guys, just another quick note from our sponsor Nexus labs. And then we'll get back to the show. This episode is brought to you by nexus foundations, our introductory course on the smart buildings industry. If you're new to the industry, this course is for you. If you're an industry vet, but want to understand how technology is changing things.

This course is also for you. The alumni are raving about the content, which they say pulls it all [00:25:00] together, and they also love getting to meet the other students on the weekly zoom calls and in the private chat room, you can find out more about the course@courses.nexus lab. Start online. All right, back to the interview

I want to circle back to a couple of the things you've talked about so far and kind of demystify them and mostly around the machine learning aspect of this.

[00:25:19] Zach Denning: So can you start.

[00:25:22] James Dice: By, by just saying, what is this, what does machine learning mean in this context? And and maybe start also with, you mentioned digital twin, and I feel like starting with a digital twin is the first piece of this. What do you mean by digital twin? And then how does machine learning interact with digital twin?

These are like the two buzzwords of the decade and smart buildings, but I feel like, I feel like you can demystify both of them. What you mean

[00:25:46] Zach Denning: here. Yeah, it sucks that we even had to like start using the term digital twins, because like everybody else is using it. I guess we used to call them. Yeah. Um, Their equipment models.

We still call them models internally, but [00:26:00] I'm stuck on models.

[00:26:02] James Dice: I'm going to my models until the day I die. I'm not, I'm not going to go.

[00:26:08] Zach Denning: It makes a lot of sense. You know, it's like, I guess in some contexts, yes, it is a digital plan, but it's like, I'm not a huge fan of buzzwords. And I usually try and steer clear of a lot of ways because.

Buzzword is marketing dollars. That's all it relates to who has the most marketing dollars wins the buzzword, but the real value it's like, where's the value. I think when, when we look at digital twin, it's a combination of alarms and machine learning models. That point to the best operation of a piece of.

The ideal operation. So if you look at the machine learning models that we have that make up our digital twin, we may have around six or seven in a VAV, for example. So VBS or sedan for just a damper Danford about the day of the damper valve and some sensors, right. We may have six or seven that around that VAB.

And we deploy them to either forecast they condition that's going to happen so that we [00:27:00] can react to it. Or we use it to optimize to a condition that we know is the most efficient possible. The third way we use them is for alarming to say, in a mechanical sense, I can't do anything here. I can't optimize how a damper reacts.

It's just reacting. It's just controlling the one outcome. But what I can say is, is it controlling properly to that outcome? So I can use it in a reactive sense as well. So I can use it in a forecasting sense. I can use it in a control sense to optimize, and I can use it in a. Reactive sense to say, are you doing what I thought you should be doing?

So you can use actually to police itself. And that is our digital twin. It's actually self policing. The only way ours started was we started on an analytics platform, right? So we had already built like digital twins, five years. And optimize. We correlated them to costs. We built a financial infrastructure.

We needed to say yes or no on, should I replace it or repaired or upgraded. We built a lot of those things into the platform before we even started down the road of control using machine learning. So let me repeat, try to repeat

[00:27:59] James Dice: that back to you. [00:28:00] So you're, you're taking a piece of equipment and then you're taking the system that's you know, basically system.

A bunch of equipment that's network together, basically. So you have these systems system of systems. So you're creating a model of each of those systems and how they operate. And then the machine learning is basically saying here's how it could be operating, should be operating or is it heres the problem with the way it's

[00:28:24] Zach Denning: operating?

Yeah. Yeah. It might even operate in the future, you know, like what are we looking at when it comes to the future of. You know, this piece of equipment, you have an HVAC is extremely reactive in nature, which is also one of the reasons why you drive up energy for all the you know, all the naysayers, the machine learning.

It's like, it's, you, you may be the most, the best reactive software you can write, but it's still reacting. What, if you could write forecast that software of what it should be doing are going to do not currently doing you take the data of what it's doing, any forecast, what it should be doing. You started doing that.

And you can see these there's drastic changes in it's like flat-lining. I think that was one of the most [00:29:00] interesting things that, you know, when we, I reached out to, through my colleagues, go in and review our data and I'd be like, like the energy flatline, like you don't see like these crazy peaks and valleys and like it's so stable.

And I think that those are the outcomes. Cause when you're forecasting ahead, you're not always reacting to the situation and you don't have really wavy, dramatic. Which leads to less pay w savings to the demand is something that we're starting to get into. Now let's look at even like demand is like that that's demand and demand is a result of, of reactive control.

Got it.

[00:29:29] James Dice: Okay. I'm gonna throw you a curve ball here and talk about pricing, not, not for you to reveal the pricing of the product or anything like that. I'm not asking for proposal, but I'm wondering. The average, you guys, you guys primarily target CRE type of customers, right? So I'm thinking about the average, like transactional building, you know, I'm going to sell this thing in a couple of years.

How do they think about payback of this product? Because I feel like they're not sitting here going, [00:30:00] man. I really need a SAS tool to help my control system work

[00:30:03] Zach Denning: better. You know what I mean? How do they think about

[00:30:06] James Dice: the ROI of a

[00:30:07] Zach Denning: product like this? I think in a lot of cases, it's a different decision maker that we've appealed to over time with Hank and, and we go to asset managers and going up we work with everyone else in the supply chain, your BMS contractor, HVAC contractor, mechanical engineers, design engineers, property managers, building engineers.

We work with everybody across the entire supply chain to provide value, but we sell to the building owner. So it is, it is a pretty cut and dry situation. It's a financial. And a lot of times it's an effects decision, not a CapEx, which is what the industry is used to. And I think that that's one of the challenges that building automation will now continue to have.

It'll be very impacted by solutions like this is, they're always looking at cafes. You know, it's always a CapEx solution and we're looking at an OPEX solution now. That's what says it. And I think that, you know, looking at the industry from that angle changes a lot and especially a lot for owners who are looking [00:31:00] at quarterly financials, They're saying, can this pay back in the same quarter?

And the answer is absolutely. I can make you go cashflow positive this solution in a quarter that's unprecedented, you know, like lighting as a six month payback. That's a two to three quarter payback before you start seeing cashflow. And that's like the biggest no-brainer black and white should do solutions still.

And, and we can beat that by an entire. Okay. So it was a pretty

[00:31:21] James Dice: easy, it's not about increasing NOI and it's not about decreasing NOI. And then trying to justify that with increased comfort or whatever, it's just positive and alive. To begin with that's the argument.

[00:31:36] Zach Denning: And then you add on you start to compound the intangibles of better comfort.

So your engineering staff that can do more, you know, they can actually fix the equipment that they want to. Now they're not as focused on comfort all the time. Indoor air quality. Give your, give your tenants peace of mind that we're exceeding the ASHRAE standard for title 24. In most cases, by 30 to 50%, that's exceeding when we walk into buildings, most of the time they're 20 or 30% below the standard, [00:32:00] right?

Yeah. So that's like a 60% improvement overall. It's significant. And so you look at some of the intangibles and I think it gets very interesting. You even look at equipment degradation five to $7 per square foot for HVAC costs for complete roof replacement, asset value, energy savings. If I go say 30% energy on your HVAC, I reduce its consumption by 30%.

I've I've subsequently reduced uh, the wear and tear by 30%. Now all of a sudden you're looking at like Rav improvements and what do I get out of that? And it's still kind of an intangible, but I think it's gonna become more prevalent as time goes on. Yeah.

[00:32:36] James Dice: And, and what happens in a situation like this?

When the savings aren't there? Like, you know, we've all been to the building where it's like, it looks like, it looks like there's going to be savings here. And then you get to the point where it's like, oh, there's just this reason or that reason why it's just not going to happen. What do you do then? As far as the.

[00:32:56] Zach Denning: Yeah, that's been like the interesting question since leaving [00:33:00] California and progressing. So last year was our first time that we went nationwide. We've never been to a lot of the buildings. So it's, it's also a first, I think, prove proving grounds that we were doing on a technical sense. Building automation.

2000 miles away, 3000 miles away. And we never stepped foot in those buildings and some of them are wildly successful, but it was also the first time where we saw energy prices as low as like 5 cents per KWH. California average, 9, 10, 15, 20 in the area. It's crazy. And also we thought five, 10 cents.

And that's where we started. That's what the intangibles has to become tangible. Now that's where you have to look at it and say, look, I only have one engineer in the building. He's running around with his hair on fire. I really need two or three, but I can't hire him. And it happened. I can't hire them.

What about comfort? You know, you look at comfort and if the building's running, you know, out of shape, then comforts out of shape and indoor air quality is out of shape. It's not just one thing. Then you can point to those intangibles. You can also look at what's the age of the control system. Are you getting pressured to upgrade your control system for security purposes or whatever [00:34:00] the reasoning is today as to why you would upgrade it to.

And we usually work around those issues as well. Where, like, instead of upgrading, you just go with us, we'll hide everything, you know, put everything behind a VPN secure connection. You can have access via normal dashboard on the web. You don't have to worry about some crazy IP address that points to a Comcast router in your building anymore.

You know, this is more traditional SAS. So I think that those intangibles we have sold around not having paybacks most of the time the product goes cashflow new. So then you're just looking at like, what are those intangibles? Are they enough for a, for a buying decision? Got it. Yeah. I think more and more

[00:34:35] James Dice: buying decisions are headed that way to start, start quantifying those intangibles.

[00:34:39] Zach Denning: But we're

[00:34:40] James Dice: definitely, it's definitely not evenly distributed. It's one of those questions that it's like, it depends on where you're at and what type of business it is. And most of the building, it is

[00:34:47] Zach Denning: all that. So cool. So even in the understanding. Yeah.

[00:34:53] James Dice: Yeah, absolutely. So let's, let's zoom out to

[00:34:56] Zach Denning: JLL for a minute.

So how does, how does your

[00:34:59] James Dice: [00:35:00] company that got acquired fit into what JLL is trying to accomplish?

[00:35:04] Zach Denning: Yeah, I think that was like a big part of our acquisition is like, where are we going to land? I think at the end of the day we knew we were like an acquisition based exit for us. Like we knew we were going to get acquired.

But we wanted to make sure we landed in a spot that made the most sense that we could expand. And really gain market share. You know, when we were approached by JLL technologies, it, in my eyes, it was like the Google of prop tech was the master vision and the master vision being, we have. All these disparate tool sets in the industry right now.

And what we're trying to do is build an umbrella to house them under so that there's a lot more internet connectivity. And that's what we want you guys to be. You guys can be the sustainability piece, obviously have huge sustainability goals. We want to get, you know, net zero by 2040. We, we really want to push on the sustainability picture.

Cause there's just a huge opportunity here. And we've seen it right. So we saw it firsthand. Um, And it was like, you're right. But what about the technology play? You know, we we're so used to in like commercial real estate of like, you know, [00:36:00] how is software, software only solution going to drive? Within the JLL portfolio.

And I think that that's where we got really hooked into JLL technologies was that we could see the end game. It's definitely there of, of there, there needs to be a Google. There needs to be more software. I mean, Microsoft, all these guys in the nineties came about for a reason is because they found a way to house a solutions and be one unified cloud.

[00:36:24] James Dice: Yeah, I like that umbrella term. Yeah. Stitching all these different products together because in the end, like you said, when someone wants to buy de-carbonization, they don't want to buy advanced supervisor control plus FTD plus independent data layer. Plus all the acronyms we love on this podcast, people don't necessarily want

[00:36:43] Zach Denning: to buy those.

Well, they don't understand them. Yeah. That's the other side is it's very hard to understand. Totally.

[00:36:49] James Dice: I don't know what that means about this podcast then if that's already, those are the things we talk about, but we're obviously not just talking about those things. We're talking about enabling these, these

[00:36:58] Zach Denning: outcomes.

So [00:37:00] yeah, I mean, at the end of the day, like a lot of the acronyms, even back then still play a role. Right? A lot of the things that you did still have to play role, like there's fluffy interoperability in buildings. There's lots to be these things. For an umbrella to exist, but it's, it's the why? Like why should the umbrella and exist in this industry?

And I think there's a very good why now, whereas before it may have been a little bit harder to establish the why I think the why is pretty obvious, the reluctance to change is still there, but you know, they, I think they will supersede that reluctance in the payback, the definite benefit. Yeah. Yeah, it makes perfect sense.

[00:37:35] James Dice: Okay. Back on the product a little bit more, I want to talk about in our foundations course, which is like all on my mind right now, because we're going through it with, with 56 people right now. And it's like very, very intense uh, for, for all of them that are learning about smart buildings for the first time.

So the first thing they think about is. Who right. Who does this impact, this new technology. And so I'd love to sort of dive into this a [00:38:00] little bit, because I think it's interesting that the product is named tank. It's named a person's name, and yet it's a software product. And. It impacts a certain who, which is a building engineer, a building operator like that user persona.

So I'm wondering, and throughout this conversation, you've been talking about alarms and the ability for this person to get alerts about how their building's operating and opportunities. So I guess w where I want to start with that. Is that the primary user of the software, because when you have the supervisory control layer, you know, machine learning layer, typically you talk about it, not needing users, like it's supposed to automate stuff, but what you guys have is like, almost like an analytics platform also.

So you have something that the building operator can log into and they're, you know, are they, what's their sort of role in making the software successful?

[00:38:55] Zach Denning: Yeah. Yeah. I think you know, even where the name was born from when we got into our first [00:39:00] we got to our first building, it was a privately on building.

You know, it was a hundred thousand square foot privately owned building and they gave us like a pop-up and one of the opening. And it was like concrete fleet. We popped up a table and we sat there and we were running the building and people would come by and they'd talk to us. And it's like, what do you guys do?

Like you guys are the HVAC people doing the HVAC thing. And I brought my whole staff out. So it was like five of us at the time sitting in this pop-up because we wanted to see and feel and experience it. Like, what does this thing actually do? How does it actually feel like, you know, and they'd come by and they say, Hey, I feel more comfortable.

And that's when we, we kind of stumbled across the comfort picture, comfort equations or. You know, the grant period was like, we can actually prove comfort. Like that was never even part of the equation. This can improve comfort, but then you started asking questions, like, what is it? We started calling it a virtual engineering, cause they didn't have any engineers in that building.

They had a floating engineer that would come around every now and then, but they didn't have an engineer and it's like, listen and if it was his name and so I just Googled the top 10, most trusted names and Hank was like seven. And I was like, how about, Hey, and people loved it, you know? And so we went from like a CIA [00:40:00] spinoff names to like Hank and, and it worked out really well.

No, when we look at it, like you're, You're absolutely right. The end user here is a building engineer. And when we look at all these different flavors of dashboards that building industry have access to, it's expansive. And then all the knobs that they can turn in on these dashboards is even crazier. You know, how many different set points do I have the control and what does it actually mean?

Why would I do this or that during the day? And we've tried to consolidate as much as we can, like literal comfort sliders. You can, you can actually go into Hank and just set the comfort. And when you set the comfort, you can see how it impacts energy. And you can say, wow, if I increase comfort here by 10% or 20%, I'm going to decrease energy by 5% or 6%.

I can actually know... we make it. It's a relative platform now that is pointed towards the outcome that they want. That is, that is a challenge though, right? Because you're going from being able or wanting to, or, or just historically changing all the knobs to that's being done already. Now you can focus on

like, hey, we have like 20 things that [00:41:00] are broken and they require a lot of hands-on intensive labor that's very specific, right? Like shimming motors or finding out why compressors aren't working to their full capacity. There's these really advanced mechanical issues and that's where we really work

with building engineers on is, you don't have to sit behind a console staring and turning knobs all day anymore. Instead you can really focus on the things that you're really good at. And we can get you out there and working on those things more to improve the building. We can also help you with payback analysis.

Like, we can actually get you a budget or help you get a budget and get that piece of equipment fixed or get it repaired. And we can be an advocate with the owner with you because they're the ones that put us in the building. And so I think in a lot of ways, our head of sales and I'll take it from him, but he coined the assistant to the assistant engineer.

You know, this is, this is giving you the ability to do better at your job. It's extending your capabilities drastically. And I mean, the flip side of the equation is, I mean, we all know what's going on in the trades right now is they're hurting pretty badly. And building engineering is [00:42:00] no exception. I mean, I graduated school with a lot of building engineers, even they're saying it is like, there's just no more showing up.

The baby boomers are retiring. Nobody's here to replace them so they're being stretched. Budgets are being stretched. You know, owners don't want to spend money on an HVAC unit. You know, what does that do? You know, it's like an insurance policy and we can help them out with those.

[00:42:19] James Dice: I mean, that's a great answer. I think that's where it has to be headed, which is get, get these manual tasks off their plate get these things that they shouldn't be doing off their plate, so they can actually provide the, do the things that are that are higher value. What about the engineers that feel like they need to keep the manual stuff?

And let me, let me give you an example. This building I've worked with probably five to seven years ago. There was this guy and he just really felt like it was his job to decide when the HVAC was going to turn on in the morning. I don't know why he felt that way, but there's a lot of buildings like that out there that like, they view it as [00:43:00] their job to show up in the morning and turn stuff on.

How do you, how do you sort of overcome that inevitable skepticism or pushback around giving control to the, to the AI?

[00:43:12] Zach Denning: Yeah. Yeah. I mean, I, I, one of my buddies is chief engineer down from Cisco. He said when he was assistant chief he floated around to a few different buildings. He went to a building and the chief engineer every day, he would go, you know, he had, he had an hour lunch.

He liked to work out, you changed your gym clothes, he'd run upstairs, he'd flip the chiller off and he'd go to lunch. And then when he come back, he'd flip it back on and he's like, why did he do that? And he's like, I noticed that when I used to leave, it would run for half an hour for an hour. And so I started saving energy and I'd flip it off.

And he was like, all right, like, you know, if it's just a different mentality, you know, and I think it's that mentality is like, you have to, like, everybody is afforded a significant level of control. It is their building. Their job is to run the comfort and the HVAC equipment to comfort. Like we're, we're trying to [00:44:00] achieve the same outcome, but HVAC is still their job every day.

They're being judged on it. Their whole career is built on. Does the HVAC work properly or does it not? So you can understand that there's a significant level of, there should be right. Somebody came in and they said like, Hey, I'm going to take over your whole job. And when it comes to this part of your job, I'm gonna take this part of your job and you have to trust in me, I would say.

There should be a lot of skepticism. No, what we do is we try to get them involved early and often it's still they're building, they're still running the building. We need to know the nuances of the building. Like what are those new, what does that floor that you just don't know? Which one is that, you know, it's always this floor, it's always this piece of equipment.

It never works like great. Let's tackle that one first and we've come up with a we call it an onboarding plan, but it's a strategic plan to work with the onsite staff to ensure that it's onboard in a way that makes sense of their current workflow. We train them on the software. We get them very integrated because they're an integral part of us being successful in the building.

You can't just walk into a building and grab whatever mechanical drawings they have grabbed the building [00:45:00] automation and say like, let's go to work. Inevitably there's 30% of the things that you're gonna have to guess on. And if you guess wrong, you're going to cause more complaints and more harm than good.

We learned that very early on and so like we are providing a service that helps them and helps the owner and helps the property manage. You know, the three people running the building or managing the building in that instance, and we have to ensure that it is going to help everybody. And there's a big focus on our service in the beginning.

Alignment with beyond on-site staff. When it comes to the platform itself, they have the exact same overrides we have today. We went over, I dare say, pressure, go for it. You want to override the DFC, go for it. If you feel there's a certain scenario where this needs to happen. Sure. If you do better than we won't get an alarm, if something happens that goes wrong and we get an alarm and it comes back to our support staff, we may call you and talk you through it.

Why did you do that? Why did you feel like you need it to like, where can we come to a common ground? What can we do? You know, we are in every term of it, like a virtual engineering staff. If they're to support everybody. [00:46:00] Cool.

[00:46:00] James Dice: How about we talked about who, and that would be one who there's another, who that's often involved, has their hands in the pot here which is like a service provider, some control system that you've probably played that role before.

And there's a, there's a technician that's responsible for showing up here once a month and doing PMs and keeping. You know that building automation system running. So obviously we interject this virtual engineering team. We interject this overlay software. How can you kind of bring those teams along and, and in a new change like this?

[00:46:34] Zach Denning: I mean, John Dell found a way to survive with windows and he was wildly success. You know, he, he recognized the changes in the industry and he adapted very fast, fast, or anybody else could. And that's why we also used ELLs. You know, I think that there's a certain adaptation that will have to happen.

Just virtual engineering is an inevitability. It's not a, a, it may happen. There may be a breakup of this vertically integrated stack , this monolithic stack that everybody owns, it will happen. You know, and [00:47:00] I think understanding that and understanding that, you know, we're not a hardware company. We are a software company.

If I go find a broken controller, a broken sensor, There's one one end point per thousand square feet. And there's three associated sensors to that. So there's three sensors associated per thousand square feet. I'm not providing them. I'm not replacing those. That's still entirely your system. If it's an Allerton building or Siemens building or Johnson building is still their system.

No intention of replacing components. That's not my business. I'm just trying to make it all work better. So when I go find just like a building engineer would, when I go find a broken sensor, broken controller or an opportunity to balance it, this. I put it out there. Hey, you found the system. This is why, this is what it's going to do for you.

And you should probably go use X. Who's going to be the guy there, path of least resistance. So in a lot of ways, again, I'm just providing a virtual engineering service. I'm not, I'm not ruling out or changing. You know, a lot of what the industry is used to, which is the hardware side. So I'd say the services are going to change In some ways you know, we are running the [00:48:00] software and the building.

Now there is no need to do software maintenance or software upgrades, or a lot of those things, but there still is tons of improvement. I mean, we, on average we find 10% of all controllers are failed in the building and nobody knows that's a five to $10,000 project is waiting in the wind on, on every building that we get into.

It's just the understanding of where the industry is going and where can I drive value in this kind of like new. New market or evolution of the market that we're in. So do you see

[00:48:26] James Dice: a reduced need for controls service contracts then? How does that, what are you thinking

[00:48:32] Zach Denning: there? Yeah, yeah, yeah. Yeah. I think that, you know, control service contracts are for the exact reason that we talked about before, which is you know, the, the skepticism, right?

The skepticism is you have to have controlled service contracts right now because you have to be able to change the software. Yeah. Inevitability. Right right now that is entire, entirely an inevitability. Like you will have to change it seasonally. It will not operate cross seasonally unless it's operated and efficiently, you can run everything at maximum cooling and heating and I'm sure it'll [00:49:00] survive seasonally.

If you want to operate through its utmost efficiency, that's not plausible, but you know, in a lot of ways, I think that that's where the. Here. There's a lot of those kinds of software driven changes during the service contracts. And I don't think that, you know, under our control, those aren't necessary, but that's not to say that things don't break customers don't know.

I mean, one of the funniest things we get like suppliers from sensors, which in a, in a, in a zone. And everybody goes like, oh, that's a monitoring thing. And it's like, no, it's a control thing. We use it for control purposes. It's actually a very, very valuable set point in terms of machine learning and forecasting.

It's almost one of the most valuable set points that we have in the system. And we find that 20 of them are broken. We'll tell owners they have to go replace them. Like you want this level of comfort. We need those replace. That's a project. There's plenty of projects to be found if there's that's the hardware side of the business, and this is the software.

So if you're able to divide those and figure out what service needs to be had here, then, then that is still where the service logs. Cool. [00:50:00] Alright,

[00:50:00] James Dice: thanks. Zach has been fun. Let's let's end with carve-outs. I'd love to hear a book podcast, TV show, whatever you want to share with the audience that recommend that they check out it could be personal and

[00:50:12] Zach Denning: professional life.

Yeah, the, the book blue ocean was the book. I talked a lot about category creation. There was pretty, pretty fantastic when we read it for our business to understand a lot of the things I'm talking about, like virtual engineering and where this goes, where it's derived from the necessity to create a new category.

This is the evolution of building automation. And everybody talks about that, but it's siloed. It's so disparate of like, we're going to evolve. We're going to evolve, you know, optimization and it's like, you're missing all the core components though. It's just value delivered to the end users, these three pillars of, of people that run and managed building.

And so I think that that was like, the biggest thing is like this all-encompassing category that we've created, which is virtual engineering. And then creating the connotation around that. It's a very positive thing for everybody can be [00:51:00] perceived as scary, but it's also like creating the messaging and allows that was derived from, you know, a lot of the, the big things they talked about in blue ocean.

Cool. Yeah. I think we need a lot

[00:51:09] James Dice: more of that sort of thinking. I haven't read that book, but

[00:51:12] Zach Denning: Thinking about picking it

[00:51:13] James Dice: up now. Um, I'll share mine. So this is one of my favorite books. So you said that's your favorite books? It's called a guide to the good life, which is it sounds very gimmicky and it sounds like self-helpy, but it's really not.

It's like an introduction to stoicism, which is a 2000 year philosophy that came from, you know, ancient the ancient world, basically. So this is just like, modern day explanation of, of that philosophy. And I think it's really great for really just managing the day to day BS that we all have to deal with mostly in our own minds and just kind of being happier people.

So, we'll share that one in the show notes as well. I definitely

[00:51:51] Zach Denning: recommend people check it out. So. Cool. Well, thanks Zach.

[00:51:56] James Dice: We'll check out blue ocean. We'll put that in the show notes and [00:52:00] we'll link to everything we talked about as well with Hank and, and JLL and all that stuff too. So thanks for coming

[00:52:04] Zach Denning: on the show.

Awesome. Yeah, I much appreciate it, James. Thank you.

[00:52:11] James Dice: All right friends, thanks for listening to this episode of the Nexus Podcast. For more episodes like this and to get the weekly Nexus Newsletter, which by the way, readers have said is the best way to stay up to date on the future of the smart building industry, please subscribe at nexuslabs.online. You can find the show notes for this conversation there as well. Have a great day.