"You feel like you're tackling an impossible problem because you're looking at such a broad set of solutions or in our case, now we're looking at such a broad set of manufacturers, and so many devices in our building, and every building is different. How can you ever address that?
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 56 is a very enlightening conversation with Sabine Lam, Google's Building Operating System Global Lead for Real Estate and Workplace Services.
We unpacked what the building operating system is, the goals Google has that depend on it, and what needs to change about how buildings are built in order to.enable it.
I love Sabine's practical approach that seems to be necessitated by sheer scale and just not being able to put up with a lot of the BS in our industry because of that.
Mentions and Links
- Episode 29: Google's plan for smart buildings at scale (2:05)
- Xilinx (3:04)
- Mary Davidge (5:50)
- Portico (7:52)
- Kathy Farrington (8:23)
- How building digitalization is changing the Master Systems Integrators focus area (27:57)
- Trevor Pering (29:23)
- DeepMind (49:43)
You can find Sabine Lam on LinkedIn.
Thoughts, comments, reactions? Let us know in the comments.
- How Sabine and Google helped change the building materials supply chain and how that is happening in building technology now (5:35)
- The Google BOS mission statement (14:38)
- The three specific goals of the BOS program and how it's being driven by cyber security (27:44)
- Sabine's take on IT vs. OT. Hint: it's all on the IT network. (31:37)
- The role of the MSI and contractors in setting up the BOS and our frustration with data mapping (40:15)
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!
James Dice: [00:00:03] 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.
Episode 56 is a very enlightening conversation with Sabine Lam, Google's Building Operating System Global Lead for Real Estate and Workplace Services. We unpacked what the building operating system is, the goals Google has that depend on it, and what needs to change about how buildings are built in order to.enable it. I love Sabine's practical approach that seems to be necessitated by sheer scale and just not being able to put up with a lot of the BS in our industry because of that. Without further ado, please enjoy Nexus Podcast, Episode 56.
James Dice: [00:01:01] Hello, Sabine. Welcome to the show. Can you introduce yourself?
Sabine Lam: [00:01:05] Hi James. Thank you so much for having me on this show. I'm actually very much looking forward to this podcast and providing some good information here about what we do at Google, but as a introduction, I'm, I'm Sabine Lam. I'm a technology program manager at Google in the real estate and workplace services team.
So you'll, you'll hear me talk about REWS that's real estate workplace services. And I've been championing our effort, a program around building up rating system. With a team of different Googlers from different organization how to move forward and, and implement an infrastructure, additional building infrastructure infrastructure for Google portfolio.
So very much focused on our portfolio a Google solution for our corporate real estate. And really so that's what this bring me here today.
James Dice: [00:01:53] Yeah. Yeah. And we talked to Keith Burke had been back in November on the podcast. So anyone that has listened to that episode he's he works with you, I'm assuming that was back in episode 29.
I think that was one of my most popular episodes. So I've been wanting to get some, another Googler back onto the show sometime.
Sabine Lam: [00:02:13] Yeah. Yeah, we have a few of us on here, so he's definitely, he was focused on the digital billing ontology. I'm more representative the facilities and the real estate team on the risk side and kind of turning, turning to our networking and security and engineering team to provide a solution for REWS.
So yeah, that's, that's I worked very closely with, with Keith for sure.
James Dice: [00:02:37] Got it. So before we get into all that, we're going to unpack quite a bit which I'm really excited to do. Can you talk about your background? How did you get here
Sabine Lam: [00:02:47] today? Yeah, actually that, I think it's slightly different from potentially what you've heard so far.
Quite varied background. I graduated with electrical engineering degree in France. And came to the land of opportunities in the Bay area to work for Xilinx, which is a semiconductor company. So they, they make PGS, which are field programmable gate arrays, little chips with millions of transistor that you can program to do any function you want to do.
Okay. So it's really nice to drive innovation. You can install them and reprogram them. So it really is installed in a wide range of industries and technology. Right now, you would hear about them on the consumer market cars cloud. At the time I was focused on digital signal processing applied to video processing and telecommunication.
So it was all about creating a library of solutions or IP for compression, decompression transform. We developed a solution for 3d 3g based station at the time. So I, you know, and as I thought about it, I'm like, okay, here's our few buzzwords that are actually used today for building related buzzwords.
So, you know, transformation is one thing, but at that stability where you can reprogram the, the, the FPG is to do something different. So adjusting, and then obviously digitalizing digital, you know, digital building as opposed to GTL signal processing. But the understanding that technology can do amazing things.
Right. And so I think that gave me that view of the world, of this little tiny chip here can do amazing, amazing, and you know, has amazing capabilities. And and then sometime when I look at the billions of meetings and, you know, building automation system, like, you know, I think, I think we can do more, right?
So they, they, they always wanting more out of this in a process. So I worked for them for 12 years and During that time, I had three kids, so I got to the point where work and kids was too much. And I stopped working for seven years since you raised my kids and work on the other side of my brain.
So I focused on woodworking and glassblowing and we ran out of house and I bought, you know, build my cabinets and furniture. For concrete did all kinds of really fun kind of 3d things. It was, it was really, really entertaining and rewarding in some ways. Definitely not a good place to make a living.
It's really hard to make a living in anything that's kind of artistic related. And I quickly realized that I had to go back into if I ever wanted to go back in high-tech I kinda had to get back into it. So after six years or so, I started looking for a job and Took some classes for lead leadership in energy and environmental design, PMP, et cetera.
And it took me a while, but I did get a job at Google as an intern, of all things. So you remember when the movie, The Intern came out?
James Dice: [00:05:42] Yea, uh hu.
Sabine Lam: [00:05:42] The 40 year old guy that went to be an intern. That was me. I came in, actually neighbors of ours, Mary Davidge was working for Google as a Green Team Consultant. She is actually very well known for the workplace environment.
And so Mary hired me to develop a solution around healthy material. So at the time we wanted to make sure that every construction material in our building did not contain chemicals of concerns. So we created a database of those materials, like the paint, the carpet, the table, this and that, and and reached out to a manufacturer to understand what was the composition of that material.
And was there any chemicals of concern within, within that material? If there was chemicals concerns, we could not install it in our building. And so it was really just creating this database with you know, interaction between the manufacturer and adding their information and then the construction team who then could pull from that list of material to decide what to install within the building.
And the really cool thing about it is you realize that the industry did not know what was in there in that paint. Right. They had no idea and they had never looked into it and they had to go down to talk to their supply chain to find out what, what is this made of? And when they realize what it was made off and realized that I was bad for the environment, bad for your health, bad for all those different reason, they actually changed it.
And so that was kind of my introduction to Google, who like, you know, so much as in the high-tech world, summit conductor world, you zero in on one solution, you work as fast as you can on that one solution, and you hope to be first to market. With Google, you have that opportunity to look at everything.
And so we looked at every single material. Can imagine as a building, you have a million, you know, thousands of material? You look at everything and you see which industry is actually willing to make a change. And so we made a lot of progress with paints and with carpets and you know, some of the furniture, and right there just, you know, influence the industry to produce better material and create a better environment for the employees.
So that was called the Healthy Material Tool or Portico. From there, I moved into the team called, Technology Program Manager, which is I'm still in that team today. But we work with different business within REWS and the team I work with was the Facilities Team.
And so I joke about it, but about five years ago is when I learned how to spell BMS and understand it's back system. So you know, and work with the Facilities Team around what tool might they need to be more efficient with the operation of the building.
And around that time started working with Kathy Farrington, which you mentioned earlier, and Kathy has a lot more bench depth in construction, and operation and, building operation, et cetera. And so we started looking at how do we make our building better in many different ways and kind of focus on this digital billing concept and creating a solution for Google.
So Kathy and I, and a few others from a different organization created this program that we call, the BOS program, Building Operating System Program in kinda like bottom up, right. Decided to focus on it, propose a solution, or at least highlight where the concerns are, propose a solution and move forward. And then since then this program has grown quite a bit, you know, and we starting to have solutions that are worth sharing.
And yeah, that's where I am today. And you know, as I mentioned earlier I want to impress with, I don't have the depths and I'm not going to impress with what happened in the past. That's not, that's not me, but certainly for how we move forward, what's done today, and how we move forward and, you know, kind of moving OT towards this IT world.
There's a lot of interesting things that we're doing that I think I can share information on. So...
James Dice: [00:09:35] Yeah, I'm excited for you to do that. That's a really cool background. So I'm assuming you learned when you were going through all the different offgassing for lack of a better term of all the materials you probably learned that Google has quite a bit of pull in terms of the supply chain, right?
The scale of Google, right. Are you able to take those sort of lessons and apply them to what you're doing now with buildings?
Sabine Lam: [00:10:00] Yeah. So it's exactly the same thing. You feel like you're tackling an impossible problem because you're looking at such a broad set of solution or in our case, now we're looking at such a broad set of manufacturer and so many devices in our building and every building is different.
How can you ever address that? And I think the experience with the healthy material, it was very similar. We'll just address everything and then move forward with whomever is willing to move forward with you, right? And it could be at a regional level or a certain manufacturer, who understands what you want to accomplish, they totally buy into it, and they move forward. Some others, not so much, you keep them up to date, but you don't hold back. You just move forward with whomever is willing to move forward. And so that absolutely applies with the Building Operating System solution that we have today, you know, and our involvment with manufacturers and within Google as well.
We'll see how different regions move at different speeds, different construction teams move at different speeds, different facilities operation teams embrace the solution more than others. That's quite interesting. And so we're large enough that we always find someone who is interested and we have work to do, right?
James Dice: [00:11:10] We were talking a little bit before we hit record about how I feel like you guys are one of the first ones that are sort of pushing the marketplace a little bit and a bunch of different ways. But one of the reasons I'm really excited about that is I feel like there's been a lot of obviously inertia and sort of slow movement from, you know, vast quantities of people in our supply chain, in the buildings.
Right? And so once, once they realize that they can't get jobs, unless they start to meet some of these criteria, I think that's going to be an important driver for change in the industry.
Sabine Lam: [00:11:40] But Karen is quite important, right? When you talk to manufacturer and you have a 1 million square feet project at the end of the tunnel, do you want to work with us to be able to bid on this project or you don't want to work with us?
You know? And, and, and for us at least for, you know, for this team, that's leading the program, it's all about being technology agnostic. You know, w we just look for a solution, the latest and the best solution that that follows our, our criteria and those criteria are always evolving. So it's never a stable solution, even when you're in, you're not in forever.
Right? Well, we'll, we'll continue to evaluate the solutions against others. And so, it's an interesting position to be at Google, for sure. I recognize there's things. We can do that a lot of other company can't mislead do. Right. But hopefully we're using it the right way. I feel the company we work with and, you know, and we'll go through what is important to Google, but one of them being cybersecurity, it's a win-win situation.
Anyone we work with that he's going to focus on making the device more cyber secure is going to serve the entire, the entire industry industry. Right. It's not just specific to also
James Dice: [00:12:48] absolutely. Before we get into that. Can you talk, just like give people context on the actual portfolio itself? Like all I know is there's a ton of buildings.
You guys are doing a ton of construction and there's also a lot of data centers. That's all I know. Just a little bit more context.
Sabine Lam: [00:13:04] Yeah, yeah, yeah, exactly. So we're saying Google is dang. Well, how big is it? You know, we're shy of 700 buildings. So that is not including data center. The data center is managed very separately from corporate real estate.
So again, I'll focus on corporate real estate. So, you know, short of 700 buildings across 54 countries 200 CDs close to 200 CDs for about 40 million square feet of real estate space. And I believe I mean, I've heard where one of the largest real estate owner in the U S. So that gives us quite a large footprint.
Some of it is leased. Some of it is owned. Some of it is full tenants. Some of it's just a few floors. It's really a mixed bag of types of buildings, size location. And then, you know, some buildings, we operate some building, the landlord operates. So there's just about everything out there for us to play with.
And the solution we're putting together is for the entire portfolio. So there's nothing, no building left behind as far as we're concerned. And, you know, they all have the concept. And the other concept is every Googler is important. We can't focus on one region more than another, you know, we could never go and say, you are more important in this region than you are in these other region.
And so we're really trying to be Google is big and equity and you know, all that stuff and it applies to our building strategy as well. That's quite
James Dice: [00:14:29] So let's kind of dig into it. So I want to start with like the overall sort of technology vision, you guys call it digital buildings. So like what's the, What are you trying to accomplish with technology in your buildings?
Sabine Lam: [00:14:42] Yeah. So I thought I'd pull out my mission statement cause you know how hard it is to put a mission statement together and then you never use it. So here's my chance to use it. I'll just read the sentence and then pull out like four main points that are very representative of what we're trying to accomplish. But it's to maximize the value of Google real estate investment by securely connecting spaces to enable data driven execution and tailored user experience. And so the very first one is the value of Google real estate.
And basically it relates to the ten billion dollar investment that we have in real estate and data center. And so how do we make sure that we get the most value throughout the life cycle of the building and avoid any kind of negative spend while we're optimizing the space and the health and the performance, right?
So that's kind of maximizing the value of that expense. Securely connecting spaces. I mean, security is number one. I would say that is what is moving this program forward. You know, the force behind it is definitely coming from the cyber security aspect of it. That is a problem that's well understood by our Security and Networking Team.
And they have the power to move it forward, almost ahead of how Real Estate Team is realizing the value of it. You know what I mean? It's kind of pushed by security more so than the operation team realizing how much of a difference we can make. So that is a securely connecting the spaces around connecting each buildings, the floor of the equipment.
And you think of it is, you know, it's not looking at one building anymore. We're looking at campuses like San Jose, where you have mixed use spaces and a multitude of spaces that are much more complex than just one building. And so how do you integrate the transportation program and the food and everything in a very secure way?
The other one is to enable data-driven execution. Of course, you know, Google, we like data. We make data driven decisions. And so how do we get access to those data for the entire portfolio, knowing that every one of our buildings is different? Pull this information, make it available in one place and make decisions out of that information. And then tailored user experience.
This is really the REWS aspect of it, right? Adapting the building for an optimal experience and productivity to the Googlers. So there's a mission statement and really four points. The way we're going about it is really again, delivering cyber compliant solutions. Collecting that data, and to get inside of the physical space and then supporting innovation.
And that goes into, again, two more categories. There's the campus vision from the Real Estate Team and then there's the operational efficiency. The campus vision is the space has to be adaptable. We have mixed use spaces. We want to use this space during the day and at night in different ways.
Right? Again, when you think about San Jose, there'll be a mix of public open space along with very private, you know work areas that are restaurants and stores. And I think art centers and you know, lodging area and an office, you know. And all of them are mixed together. And there's the concept of, we're spending a lot of money in those offices.
Why are we only using them from 8-5? Can we use now that space in the evening, but the space that was like a tech talk, tech center during the day, can that become like a theater at night, right? Or like, how do you change that space to utilize it 24/7, if you can, or at least more than one hour.
So the private space has to become a public space. This 8-5 pre-program of your HVAC system won't work. Your building needs to adjust and adapt to the way it's occupied. There's a lot around adaptability of the space, around minimizing the environmental footprints.
That's kind of related to sustainability and making the experience for the Googler just the fantastic experience so they can come and be productive and collaborate with other Googlers. So that's all like the campus vision on the operational side, it's around again, you know, controlling cybersecurity.
So you'll see the, you know, there's a lot of work around moving all those devices onto our networks. So we have visibility and manageability of all those systems, we don't believe in just putting them on the side and trying to ignore it, that we want to see them. And along the way, if we're going to see them, it has to be done. We have some requirements that the systems need to meet. And then the support costs. So everything around predictive real-time response, and predictive analysis, and forecasting, et cetera for the operation. So how do you use the data for optimizing operations? Yeah, so that's a, high-level like what is driving effort and, you know, the way to get there requires change. It requires change to the way we build our building. And that's complex.
James Dice: [00:19:37] Absolutely.
Sabine Lam: [00:19:38] You put one on that slide. I feel like we've used that slide for years now, where we have the vertical versus a horizontal.
And it's just a very simple slide, but it is a very real and still applicable slide that we are living by.
James Dice: [00:19:54] Yeah. Yeah. I mean, that's really helpful to hear like the outcomes you're going for. Especially before we dive into like the nerdy type of stuff. Real quick though, I've been writing and sort of digging into this 24 seven clean carbon-free energy goal that Google has put out.
How does that relate to the goals you're trying to accomplish in your group? Are they tied together in some way? How do you think about that?
Sabine Lam: [00:20:22] So, yeah. W you know, one of the goal for 24 seven means that every hour of every day of every year, you need to know what your consumption is, right. So you can see, so you can eliminate it and you can source enough carbon-free energy nearby.
So within the same location or region at every hour of the day, well, how do we know our energy consumption every hour of the day today? We, we don't know, you know, I mean, and previous goals have been around in 2007. We talk about being carbon neutral. That was, that was an concept was to offset emission.
So at the time we would just know, okay, how, how much energy have we spent for the entire year? And you just go to your PGNE bills or whatever the landlord charged you add that up. And then you purchase enough carbon offsets and renewable energy to bring kind of your net operational emissions to zero.
So that was like, that was from 2007 on. There was a hundred percent renewable since 2017 and it was about reducing emissions. So again, that one was about purchasing, purchasing enough when you will energy too much the annual electricity use, but that was on an annual basis, maybe monthly basis.
Right? So we've been tracking consumption on a monthly basis. Some of it a little bit automated, a lot of it just, hello facilities. Can you fill out this form and tell me what your PGT bill looks like, right. And fill it here manually. Right. And if you don't enter it, somebody you can't do that. If it's 24 seven to automatically get access to that information.
And so the infrastructure work putting in place is about putting some hardware within the building that is capable of sending the data directly to our cloud on an hourly basis. With that, it also means negotiation with with a landlord, right. Because yeah. Like I say, we have lease, lease building.
How do we get access to that information? So there's a lot of effort around the, the clause that we include in our lease negotiation. And I think that's a very powerful one, too. And then the, the landlord should start becoming familiar with that, right? W we are now requesting that our space are cyber secure, that we get access to data and, and wherever possible that we operate our building, we follow it.
We have a lot more opportunity if we can operate our buildings or at least our floors. And so we have, we organized kind of our portfolio in different models. So there's a model that is fully operated by the landlord. There's only so much we can do, but at least, you know, here are best practices. Please keep our Google are safe and to follow these guidelines.
Right. And, but there's nothing we can really enforce is not ours. And we can't see the data. So it's just a high level conversation. And then there's models where we can operate our Google floors or operate the entire building. Okay. All models, whatever we're not operating, we're still asking the landlord to give us access to some level of data.
I think energy data, you know is not a complex one that the landlord would argue against where it becomes more controversial potentially is when we start looking at HVAC data and you know, and others where I think it comes across as a threat, as opposed to an opportunity we have, you know what, we're looking at our entire portfolio.
We might find things in this building. We want to understand how a building, how well our building is operating and likely we can return information to the landlord saying. Maybe you're, you're doing well and you're not doing well, you know, can you, you know, improve here or there, but yeah, absolutely. The sustainability goal, you know, sustainability has always been a core value to Google for, for decades.
And I think now it's required requiring technology for us to pass it. You know, we're claiming 2030, we got to get stuff in place in time to, to prove it.
James Dice: [00:24:05] Yeah, I think, I think this is an area where, you know, the Biden administration has come out and said, they're going to try to do this in federal buildings as well.
This is an area where it just my opinion, my assessment of the smart buildings, nerds, don't quite grasp the connection between what they do and these 24 seven goals. So if that's you out there, like start wrapping your head around this. Cause it's, it's definitely coming.
Sabine Lam: [00:24:28] It's one also that you can easily measure.
And so I kind of have mixed feeling about it because I feel just saving energy for your building. It's not a, there's much more complex goals. So you look at it. I mean, you've seen the three 3,300. Yep. Yeah. And so are you addressing the $3 a square foot and itself? The ROI of it is not mind boggling.
Excuse me. The ability is a much better story than energy saving, I think. And so you want to do well for the environment. As far as Google, Google goes, the saving on the energy itself is not a driving factor. I could never go to my CFO and say, Oh, look, I'm saving 30% of 0.01% of your spending. You know what I mean is the right story.
That's what we are trying to achieve. And, and we do. And it's one that's quite easy to measure. The one that's really hard to measure for us is Google our productivity. Right. We measure it.
James Dice: [00:25:24] How do you measure it?
Sabine Lam: [00:25:28] Well, Google has, I've been working very well from home.
No, I, you know, that we do, we've done a lot of surveys, so we do a lot of surveys. And so when during work from home, you can look at line of codes. You can look at mental health a lot of conversation between manager and Googlers. And so it's hard to measure it with data, but you certainly can measure it based on people, mental, mental, health code, you know, released things, reviewed, et cetera.
And, and we measure that back back when we're in the office, we've been measuring it throughout the time where we were at home and adjusted our approach as a result of those surveys, everything, you know, is that we're trying to understand and probably the best solution and. Yeah.
James Dice: [00:26:15] Cool. And real quick, before we kind of dive in, what did you learn from this, this measuring during COVID, besides, you know, the Google alerts were productive at home?
Sabine Lam: [00:26:27] I, I actually don't think I'm allowed to talk about all those aspects, so I'd rather not, but, you know, I think different jobs do better than others working from home. And the result is we're going to have a mix, makes model for returning to the office. So we certainly learned that we don't need to be in the office five days a week, but we still want to be in the office.
There's plenty. That needs to be done. You know, we we've heard her. Sundar's kind of talk about it a little bit then in the it's evolving as we learn more, I think we're really looking to probably flexibility, but
James Dice: [00:27:03] yeah. Evolving evolving, evolving.
Sabine Lam: [00:27:06] Never always
James Dice: [00:27:07] Yes. Yes.
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 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 email@example.com lab. Start online. All right, back to the interview.
So I want to dive into like the, so you talked about like the outcomes you're trying to enable.
Let's talk about this, these three sort of specific goals that sort of enable those outcomes. Right? So, and I'm kind of keying in on an article that we'll link to, that you wrote for automated buildings or you and your team did for automated buildings, which has the slide that you referenced earlier on it.
So whatever order you want to tackle them in, I want to dive into those sort of three, because I think these are the sorts of changes that you guys are driving on a project by project basis, sort of around the world right now, it seems like.
Sabine Lam: [00:28:19] Right. And so the reference we're making between the vertical and horizontal is, you know, the systems are on their own.
They're siloed. It's really hard to integrate. And if you, even, if you integrate it for one building, it's really hard to redo it at scale. So we're not, we're not gonna spend too much time about that. I think most people understand that the problem. When we look at horizontal infrastructure, it means that the bottom layer of the stack is all those devices.
And so what we're putting in place is you know, we're saying those device, we're trying to be as technology agnostic as possible. We want to get raw data from those devices directly. And we want to make sure they're cyber secure. So we have very clear standards around performance-based standards that is shared with the Construction Team that specifies what we're trying to accomplish with those systems and the IOT security requirements.
And so the IOT security requirement are quite complex. There's actually a list of a hundred plus tests that we do for every single IP connected devices. And so in order to automate it, Trevor Pering I'm sure you'd love talking to him, by the way, Trevor Pering created this platform called, DAQ, Device Automated Qualification. And that platform allows you to create tests for those devices.
You connect the device to your laptop and run those tests. So anyone has the ability to create a test of their interest. We have a hundred for Google, but maybe another real estate owner has, you know, other things they want to test. They have the ability to write that code for that test and automatically run it on the device.
So at this point out of the a hundred tests, some of them have to be manual, about 40% of it is automated. Manufacturer can use this DAQ solution and pretest device if they want it to. We're not quite there yet, but that is, ideally what we want to enable. It is on GitHub, so it's public and anyone can use it.
No one has jumped on it and want to do it themselves. They tend to send the device to us and we have qualification lab that go through the kind of tests and have it, you know, all the way through security review, pen testing, and all kinds of things we're doing before the device is considered qualified.
And there's two types of qualification, one is cyber compliant and one is what we call Smart Ready. The Smart Ready device are capable of sending data to cloud and that's using MQTT you know, in the UDI format. So the device is smart enough to send the data to cloud directly and cloud IOT.
And from cloud IOT, we take the data into our data platform. Okay. If the systems, you know, a lot of IP devices don't necessarily have the capability to send data to cloud, they can also connect to a gateway. And now we have smart way to gateway. So think of the sustainability, the meter is connected to a smart rate and gateway like a, you know, I'm not going to give names, so you connect to a gateway and the gateway sends the data to cloud.
Technology that is not, you know, doesn't have that capability still can connect serially to another and send the data. So, okay. So one layer is the, you know, the bottom layer is the hardware and there's a lot of conversation with manufacturer and, you know, most are very happy to work with us because we are teaching them, at least sharing good practices that are valuable for them.
The network is converged networks. So we are putting all those devices on the Google corporate network. We don't believe in to OT. We believe that we have a networking team that is very strong at supporting the IT network. And as a segment for it, that will be for the OT, operational technology.
But that same team will be able to manage both and kind of push the technology to really enforce IT principles to the operational technology.
James Dice: [00:32:02] That doesn't surprise me at all. And you've probably learned that there are people in the industry that want to keep them separate and, you know, thinking that way.
Sabine Lam: [00:32:09] I understand it, right. I mean, our IT team, they look at the device I'm like, seriously, that this is version for Android. How is that happening? You know, there are things that is just so far from the IT world. There's so many things that have to improve that it is. It makes sense to put in a, if you want to ignore it, that makes sense to put it on a separate network. The problem is then somebody hacks it, you don't know it. And that's a problem for us. You know, we think as far as PR goes, whether someone hacks the HVAC system and doesn't touch Google data, it doesn't matter. We've been hacked bottom line we've been hacked, and it's a problem.
And so we want visibility of all of it and access to the data as well. And we believe that the industry is right for implementing those IET kind of principles. Task word, you know, a little Post-It by the side of your laptop, not really a good idea. Here's the technology requirement and here's how we're going to track passwords for every single one of our buildings. We're putting like very clear solution, remote access and manage network removed, you know, 3G, not okay. There's a lot of things that are not okay and what providing solutions for the operational world, while being quite aware that it's an industry that cannot meet all those IT requirement right away.
Right? So, very pragmatic about how we can get there. And so we, so you add a layer of security. If the device cannot be secure, then the network has to do more. And then as you send the data, then there's more layer. There's many, many layers of security to provide the secure environment. Not one thing is enough.
But yeah, we work with vendors that can actually get there. But certainly narrows the pool of systems we can pick from, so it's kind of a natural standardization for our portfolio. Where as right now, total snowflake. Every building has something different.
It, I think naturally will be a little more standardized towards solutions that can meet the requirements and it's not all of them. Right. So
James Dice: [00:34:14] be good. Okay. So that's, that's the device layer. What's the, what's the next layer up from
Sabine Lam: [00:34:20] the device layer? The networking layer connectivity layer was the second one.
And I said, it's software defined network on, on faucet. So you know, it on the same network. And then the data layer. So that's when you hear about the digital building ontology, how do we send the data in a way that we can normalize it and tag it so that we can make sense of the information?
This data layer normally has, like the time series data and, you know, information of the registered devices and kind of relationship between all those devices. We can also augment it because it is within our environment. We can also augment it with metadata around, you know, and it's in this building and it's in this location and, you know, we have API to our facility facilities, API, HR, API potentially, you know, our badging.
So we can compliment the information with a lot of. PII information that we could not share with a third party vendor. So that's another reason we want it internal to our company, right? There's a lot of solution out there with this horizontal layer. We're not the, you know, again, when I was trying to sell anything, we're doing it internally because we need to have access to data that is internal to Google and not share it with any third party vendor.
Also, we don't want to pay a license for the rest of our life for such a large portfolio. And we think we're pretty darn good with understanding big data and analysis of that big data. So it's an interesting area for us to focus on. But the way we look at it, you know, it's, it's again, just so you know, similarly to me, not having a strong background and 30 years of experience in building automation system, the per the team that is creating this ontology or creating the solution.
As much more of an it data modeling type of background, but we're surrounding ourselves with really strong leader in the industry. Right? So we don't have the depth, but we're working very closely with people who do have the depth, but because we don't have the depth and we are not locked into one way of doing things, I think we can be very creative about how we address it slightly differently.
And I mean, I, you know, we don't, we don't have to talk about Keith the general deontology you already have a podcast on all of that, but when you think about it, he came up with this and a couple years, and it's at the level where haystack four is just about doing the same thing now with the relationship.
Right. And it took haystack tend to, I mean, that's, that's quite clever, right. To be able to get to a solution and there's, and we want alignment between the, between the different ontology and taking approach. But I'm like, I'm quite impressed. Like in two years you came up with a solution that, you know, the industry has been working on for a while.
And and you know, they're each learning from each other and each progressing with each other. Since let's, that's really good, actually, I look at it as healthy competition, right? You want more than one solution and you always take the best and learn from each other and kinda keep, keep progressing.
And so I think, you know, DBO was inspired by brick and haystack and potentially haystack with inspired by breaking DVO and you know, and others. And that's how we get to a better better solution. All right. So on top of the data layer, then we have a, an API and that's where you run the application on top of it.
And the application can be there in-house, or it can be developed by third party vendor. And that's again, I think on the application level, an area where the industry has much more knowledge around. You know, potentially folds to action or different things than, than we have in-house. But you know, we did a lot, we can use machine learning in some ways and kind of come up with things that would just kind of throw the algorithm.
You know, we throw machine learning on the big date, on a lot of data and all the anomalies show up and it happened to be exactly the same as what somebody wrote through, like, you know, pre-coded I'll go with them right. Of, of finding faults. So it's quite interesting the way we look at it. So what we've put in places standards, we have qualification lab we are using open standards, you know, I'm Kijiji is open DDMI. DBO is released. DACA is released. There's nothing we're doing that is proprietary. And I think at one we have it, we share it and it could be implemented on any platform.
It's also not technology specific. It's not specific to GCP the same thing. The same horizontal platform can work on it, us or, or others. Yeah. So, you know, the, the very first part of the device, connectivity, that data, that data platform, I call that the infrastructure, that's what we need to get in place.
And the first two bottom layers, what we need to put in our building during construction. So new construction is very much following some of the standard existing building as well. If we can go back, we, you know, we want to make sure that those devices are moved over to our, to our network. And so we are also addressing our portfolio of existing buildings with a similar approach.
James Dice: [00:39:18] Okay. So is this the, when you guys say building operating system, is this, these. I guess four layers you've talked about, is that the building operating system?
Sabine Lam: [00:39:27] Yeah, pretty much. I think it's the three layers that makes it the building operating system. Okay. And then, and then the application runs on top.
Yeah. Yeah, it's a group of, yeah, we look at it, it's a program. It's not one solution. It's a program of many products and combination of product to create a solution. And
James Dice: [00:39:46] yeah. Yeah, very cool. And when I read that article that you wrote, it's very clear how much open source is a part of your approach, because like you can do a control F search for open source in that article.
And it's like at every layer and multiple places. So that's really cool. I think we're going to be able to benefit as an industry a lot from what you guys are putting together. Right. Can you talk about the. The bulk of that article that I wanted to dig into a little bit more with you was this changing role of the MSI.
So given the context of what you guys need in your building, what are you asking of service providers and contractors to help you get there?
Sabine Lam: [00:40:29] Right. Right. So, you know, the MSI, and again, I won't pretend to know what MSI I've been doing, I will just tell you what we want them to do. And so you know, as I use the word evolving, but it is kind of definitely focused on the security and data driven focus where we're trying to move away from OT protocols and replace it all with, you know, communicating with MQTT UDMI.
And so forget BacNET, forget BacNET SE we need the MSIs to understand MQTT. We need them to understand how to register devices in cloud. How to verify that the data is communicating into cloud and you know, that the data is available on the cloud. And so I kind of break it into three categories.
There's the security aspect of it. They are, in our world, they are responsible for reviewing the drawings and ensuring that the technology selected is qualified or at least qualifiable and potentially even they are the one qualifying those devices. So they work very closely with our digital billing consultant. The digital billing consultant understands what we're trying to accomplish with these buildings in the use cases.
And they kind of propose some solution. The MSI is the one who understand our standards really well and ensure that the solution that has been selected can at least be qualified or they can qualify, or we send it to a lab and we qualify it ourselves. But the result has to be that the solution is qualified.
And then that solution is capable of sending data to cloud. On the data modeling side of it they are the one that are creating our billing model config file. And so they are the one that take the information from either the BMS or others and creating a billing config file.
And it just happened to be in a Yammer format in our world. So how you map this information to our digital being ontology, putting the proper format so we can ingest it into our data lake. And that's a role that is a 100% MSI role, very, very manual today. And it has to change. Like, this is not possible you know, it's a one-time, potentially it works, very error prone. So I think that's the biggest gap and I think the industry realized that that, these data mapping and modeling is the biggest gap. And and the hardest thing to maintain during the whole life cycle of the building right now, we're throwing MSIs at it and then I realized they have to stay. Once the building opens, the MSI always needs to stay around, right, to update the information as we go. So you know, model the building validate that the information is valid and match the as-is, right. So the functional description is just aligned with the as-built and trying to come up with a methodology to ensure the quality of the solution and lasting set of information is accurate.
And then the onboarding that's the concept of this. So yeah, security, and then onboarding is going to registering the device to cloud, but being familiar with cloud, understanding of how you provision those systems on our network, how you onboard those devices onto the cloud platform, things like that is also kind of their wall.
And so you didn't hear me talk about the application itself. It's all about provide the data in a format that I can do something cool with it. Don't worry about the cool part, just do the first part.
James Dice: [00:43:50] Got it. You've mentioned all the labor that goes into mapping points, so is that something that you guys are able to drive with these sort of manufacturer standards where we can start to have self identifying, self modeling type, sort of interoperability, sort of machine to machine interoperability? Is that kind of where you're headed as far as forcing standards?
Sabine Lam: [00:44:12] So the model itself, we're not discussing muddling with the manufacturer themselves. It's more once the device is in place...
James Dice: [00:44:19] This is like a pet peeve of mine. The industry, I feel like it's something, when we're talking about interoperability, we're not putting enough onus on the manufacturers themselves to basically say you guys need to adopt the modeling standards and then basically self-describe your stuff. Right?
Sabine Lam: [00:44:36] Right. So we are doing it in a way where we say, if your device is, what we call smart ready, you know, you take the data and you send it to us in this unique format. Universal Device Management Interface, right? Which is not BacNET. But there's very few company or devices that are capable of doing that.
Right. And so, the honest right now is definitely on the MSI to do a lot of the work and it's manual work of data mapping, you know. Here's what you get and here's how we want to call it, and how we want to describe the type of device, and that function of the device, and how it's connected, and what data points are for it, it's not the manufacturer doing that for us.
James Dice: [00:45:15] Yeah. And I think a lot of my frustration it's like industry level, obviously, but I've also done a lot of that mapping myself. And so the frustration is like, I just don't want anyone else to do it anymore.Shouldn't have to do this. And that's what I was telling you.
Sabine Lam: [00:45:29] You know, I think my background, in digital signal processing, do you think when you have millions of transistors, do you think people manually do things? Like, no, you have an abstract level, things are automated, you don't have people typing hundreds of lines of codes, right? You just have a library, it fits in, it knows what it needs to do and it's fully automated. And so, I'm appalled when I see what's going on, like what? And is that expectable? This is not acceptable and very quickly, it's just not scalable for Google. So it's not possible. Not only, it's not acceptable on top of that, I can't use it because our portfolio is global and it's 700 buildings and absolutely I cannot put enough people, you know, there's not enough people in the world to do it. But, you know, I don't think everybody in the world wants to focus on manually creating this config line.
James Dice: [00:46:22] Yeah. Yeah. I get emails daily about people hiring integration engineers. You know, I love the emails because I love to see that the companies admire you know, the Nexus Network are growing and that's amazing.
It's just like, it's tough to know that, that's the work that we're hiring for. And I know there are very few people out there that are looking for a job and able to do that work as well. It's just such a huge bottleneck.
Sabine Lam: [00:46:50] And I think they're doing that because they're solving for one problem for one building and it's not reproducible.
Right. And so, yeah,
James Dice: [00:47:01] so that was just fascinating. Look at all those layers. Thank you for taking us through that. So you mentioned at the top of the BOS, is this API that then feeds the different applications. Can you talk about like what applications are using these dashboards, like reporting tools. FDB like, like what's the, the application sort of stack look like that what's the what's Google's app store
Sabine Lam: [00:47:29] not great. Well, cause here it is right. The, it would be. I think the temptation is to develop the application. I don't care about the application if you don't have my billing data. So please use my engineering resource to solve the hard problem of ingesting the data. And then I know what the, you know, we know what the use cases are and w it looks great to have a dashboard and full detection, all that stuff, but honestly, there's a lot of solution out there and I, and it is not the hardest problem.
And so just like everybody else, I think I was listening to a podcast about this shiny object. Well, this shiny object is absolutely taking your engineers away from the hard problem. And so that was kind of a, definitely a hard to prioritize even within, you know, within Google and with our engineering team, please ignore the application and the pretty dashboard focused on ingestion, focused on getting this DVO working on automating these onboarding process and getting data from my entire portfolio, not just from one region in the data platform.
Okay, then I know we'll do a lot of stuff. So this, to provide an excuse to the fact that we don't have a lot of application. But we have looked at machine learning, an anomaly detection with for fall detection and in area where in, in buildings where we have data, the service provider is pulling data from the data Lake and you, you identify the type of area you're looking for and so you search for a certain condition. Well, yeah. Yeah. So, you know, as opposed to machine learning, it doesn't know anything.
It's just finding what is functioning differently in one VAV versus the other. That's how it pulls it. Whereas the other one is saying, you know, if you, if the, you know, if it's hot and cold at the same time, You know, send me an alarm or something saying I'm wasting energy anyway. So, so we're doing both at the same time when we can compare it.
It's quite interesting. I think one that is very interesting and it's been published it's outside of our group. It's from it is from DeepMind, the industrial adaptive control platform. Basically the deep mind has released this kind of real time adjustment and optimization of central shield and hot water plants.
And so it has, was primarily implemented on data center, but now we are reporting it to corporate real estate. And so that's AI driven application. And the beauty of that is not only you analyze it, you actually turn around and overwrite the systems, which is what we want to go.
Right. You know, I talked about the, I talk about the layers as a one way, but the reality is we also come back down. Right. But the solution we're providing is bi-directional where the idea is we analyze the data on and we turn around and make those updates directly onto the controllers and, you know data set points or whatever we need to do.
Real-time so, so this one is is one that that has been, you know, part of the, some of the announcement saying that we're, you know, providing solution, this that's a control platform is one of them that was the WAPs and is being shared with different vendors and building measurement, software providers and, you know, data center operators and, and others.
But that's the only one that we have. Publicly announced and selling and sell sold on our side. Yeah. There's dashboard there's the sustainability team is interested in viewing the energy consumption. So we do a dashboard for them. But you know, that'll be kind of the next step in my mind. Again, if we need, we need our billing data.
We need to prove that our stock is actually working at scale for many different types of building. And then we can, you know, we can go and have a lot of fun with the applications. Yeah.
James Dice: [00:51:20] How do you, how do you think about, and maybe you haven't yet based on that answer, you just give the employee facing applications.
So this is a very hot area right now, a bunch of companies getting invested in and bought and, you know all of the different, maybe you guys have your own already that you're planning on integrating yet. So how do you think about that piece of it?
Sabine Lam: [00:51:41] Has the employee controlling the environment,
James Dice: [00:51:45] the environment, but also just like there's these you know, that's one capability of that type of app.
Others are just like wayfinding and access control through your phone. Engagement amenities, like those types of things that,
Sabine Lam: [00:52:01] so that there is a lot, you know, there is work outside of this particular effort with wayfinding. You actually, you know, you just need to know the layout of the building and some data from, from your location.
A big thing at Google is privacy. So tracking people is a hard thing to have approved for them from legal data privacy and, you know, employee policy, things like that. So we're trying to try and minimize that as much as possible. But yeah, the concept of, you know, where's your room and how you get to these, to these conference rooms based on where you are.
Yeah. We have that solution internally. A lot of it as probably ops to end. Yeah. And then, you know, the, the badges, it's really hard to get information on who's in the building or not in the building. Again, it's kinda too much of tracking of the employees. So where, where quite we're doing it very consciously and very carefully, and making sure that our employees and, and, you know, are okay with our approach.
Anytime we do anything like that. But as far as, you know, getting the employees to adjust, you know, the, the space to their preferences, it's certainly a long-term goal, but you talked to facilities, they don't like that. You know, maybe not, you know, I'm going to come in and I'm going to. You know, blast the heater and then you're going to come after me.
You get a blast, a cooler. I don't think it's, you know, it might not be great. So yeah, I would say honestly, we haven't solved any of that. It's a be great. I don't know if it's actually applicable to be confirmed.
James Dice: [00:53:35] Yeah. This is awesome to hear. I just, I just feel like there's so much going on in that space right now that I, that not enough people are saying like, wait, like privacy or wait, do employees even want this?
Like, there's, there's a bunch of things that I'm just kind of like skeptical about that whole.
I mean, even like the space, you know, if, if, if, if we can just provide a consistent temperature within the building. Yeah. Just do that right there. I'll wear a turtleneck. You wear a t-shirt I'm happy. You're happy.
Everybody's happy. Just keep this consistent airflow. And we're good to go.
Yeah. That too is then like do that only when people are there. Right. Only in the rooms that people are there on that step too. And like, if we get there, like we're still so long from getting there that it's like, okay, tell me when we're there.
And then now we can talk about something cooler than that.
Sabine Lam: [00:54:34] No, yeah. It always makes me laugh. When we talk about facial facial recognition and it can, you know, decide what mood you're in and what coffee you're going to need that morning. And I'm thinking, I don't know what coffee I want, or if I will copy right now, how can the machine does it?
So, yeah, we're quite, you know, I think we're quite pragmatic and, and, and again, we can't, because we're looking at it for the entire portfolio. We just can't jump on one quick solution. That sounds fun. It has to be proven. It has to be there's a lot of, lot of testing, a lot of functional testing of the solution.
You know, what is the business case? Like we go through like quite a thorough process before we put something in our building. At least we try, you know, I don't know everything about all buildings, but that's the goal.
James Dice: [00:55:22] Absolutely. Yeah. It's been fascinating. What are you, what are you looking forward to the rest of the year?
Sabine Lam: [00:55:28] Oh, well, can you go back into our buildings?
Getting, you know, leaving my house cybersecurity priority. Number one, we got, you know, we have a lot of support in the area, so there's something we're going to be working on towards the end of the year and next year. Okay. 2021, you know, there's only six, seven months left.
It's almost over. What do you mean 20, 22?
You get, you're going to answer that, that one's next.
So yeah, 20, 22 is, you know, th this, the stack that we were talking about we've proven it in a small area. Let's, let's make sure, like we can turn to the world and say, yeah, it's working. And here is, you know, half of our portfolio is on it and that's success.
And if we can do it, then anybody can do it because we don't have an easy portfolio. So that would be fun. And then, yeah, start focusing on some cool applications if we can. Yeah. I mean, it would be a good time that we've solved the other problem.
James Dice: [00:56:28] All right. Well, the next Google show we do, we'll have to talk about implementation.
You guys have it all designed. You'd pilot that now it's time to scale it up. And I'm sure there will be, that will have its own, you know, lessons and strategies and all that. So we'll just keep going with this Google series and just follow you guys as you go. That sounds fun.
Sabine Lam: [00:56:50] Sounds great. Well, thanks for the opportunity.
James Dice: [00:56:52] Absolutely. Thanks for coming on.
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.