Transform your asset management for building maintenance

Transform your asset management for building maintenance

Do you have an inconsistent asset register? If so, it is making your job much harder. The SFG20 Mobiliser AI solution is here to help you automate mapping your assets to SFG20 schedules in just seconds. In this webinar, Paul Bullard and Mike Talbot give you a product demo and walk you through how to use SFG20 Mobiliser to map your assets to SFG20 schedules so that you can easily identify which maintenance tasks you must complete to achieve building maintenance compliance. Let's jump right in and if you have any questions, let us know in the comments of this video.

Paul Bullard: Hi everyone. Thanks for joining us today. I'm Paul Bullard. I'm Product Director at SFG20 and today we're very excited to be able to present to you something that we feel is going to be absolutely revolutionary for our industry. I'm joined today by my colleague Mike Talbot.

Mike Talbot: Hello everyone. I'm Mike Talbot. I'm the Chief Technology Officer here at SFG20. So it's my responsibility to build our software products and I'm going to be helping by demonstrating a little bit of what we've got coming up today.

Paul Bullard: Okay, so thank you Mike. So today we're looking at a whole different way of how we mobilise an SFG20 system for your maintenance operation. So we recognise that in industry it's a real challenge actually forming an asset register or taking on an asset register. Now one of the challenges is there's so many different sources within an organisation. You may have your asset register in, it might be in your CAFM system, it might be in your BIM models, it might be on Excel spreadsheets. There's a multitude of different places, and this can be even different for different trades, for different types of asset in your organisation, for how you have to store it for the different types of regulations you're faced with.

Now from that multiple different sources of information, there's also the consistency of that data. So we recognise that the assets may not be named the same. They might not even be grouped or tagged the same in different systems. There's a whole plethora of different forms that those asset registers take and the consistency is something that we understand is very challenging.

I mean, my background is from the facilities management software industry, and many times I've seen a new CAFM system being implemented. And the reason the new CAFM system's being implemented was because the previous system, the data had gone bad. It had gone bad over time. The administration around the data hadn't been consistent. There'd been a definite lack of actually bringing that data up to date, keeping it in line, having that consistency of naming conventions, especially where different people are responsible for different aspects of your business.

So we've also looked at the data quality, and when we talk about that, we might mean the completeness of the data. How many assets do we actually have that are missing? How many are missing pieces of information? We might have a very large amount of data for one type of asset, but then for a different type of asset, we've got minimal amounts of information. That also then leads into the different types of data formats we have to deal with.

So if we're dealing with an asset register, we may have it in Excel spreadsheets, we may have that data available to us. I mean, the CAFM systems are able to export data to Excel very easily. They're all in different formats. There is no consistency around the different types of output that we see from these systems.

And also when we're managing it ourselves, if we've built our own asset registers using Excel, there's no guarantee that we've built it the same as someone in the next building. Now we can see that from the system. We have a COBie file down the bottom, and COBie is an export of BIM. COBie was introduced as an Excel format to make it accessible to everyone.

So a standard was starting to be formed through that process. But the challenge with COBie and the challenge with soft landings and all the government standards that are focused on getting you this great asset register through the design and build process has had different levels of success so far, which is why there's a lot of effort now trying to re-establish that principle.

But what we're also looking at is that many buildings are already built. So in that situation, we don't have the luxury of a BIM model to take that data from. We don't have that opportunity. So what can we do?

Once we've got our asset information though, the next thing is how do we actually maintain this asset? What do we actually need to do to each individual asset? What must we do? What does legislation say we must do? What's the best practice? What are the things we absolutely have to do to make sure this asset's good quality and it's maintained? We're prolonging its useful life, but we're also doing the things that absolutely must be done.

Now we realise that actually mapping that asset to an SFG20 schedule can be a challenge. There's over 1,500 SFG20 schedules. The basis that all of those are on can be different, that they can slightly vary depending on the use of the asset, the type of the asset. We recognise that that's something we needed to help industry with, and that's why we are here today.

We've also got the challenge of where your assets are in play. Actually, what happens on individual sites can be very different in terms of the criticality of that asset and actually what it's used for. What's its purpose? How does it fit in with those business objectives?

Now, you might have heard us use the examples before. If you're in a school in winter and the boiler's failed, the heating's not working, we have to send the children home in that situation. That's something that makes the heating systems an absolute priority for us. So maybe we go a bit beyond the statutory maintenance on those pieces of kit and we're actually delivering the maintenance that means that we can get a longer expected life out of those assets and make sure they're up and running all the time.

We also look at the individual nuances of an asset and sometimes some of the SFG20 schedules may be written in a way that if something's fitted, you need to do a piece of maintenance, and if not, you don't. Now what we want to make sure is at the time that you're assessing your site, we can help you make those decisions so that when you are providing these instructions to your engineers, the engineers are only presented with things that are relevant to you and your site. So we need a means of taking that asset information, making those decisions, and then outputting the correct schedules.

Then we look at the automation of the whole process and where SFG20 Mobiliser, our new product that we launched at the beginning of this year, where that's structured around the principle of maintenance regimes, your maintenance regimes, where you've taken SFG20 schedules and you've looked at tailoring that for your site, making it individual to those site specifics that we just talked about.

Now, this is about how that asset data can be used to automatically structure all of this content. At the end, though, we're building the maintenance regime, we're creating that structure underneath and automatically assigning those important maintenance schedules to your assets so that that data can then be used for your operational purposes.

And that's when we come onto the next step. It's all very well doing all this mapping, all this identifying of which schedule needs to be applied to which asset, but then it's a handoff. Then that data can be handed on to your downstream systems. Now we've had a lot of success with our wonderful digital partner network where our partners are using our API to actually take SFG20 data into your operational systems and pass that information down to the engineer to carry out the maintenance.

Now our new product, our new technology that we've just shown you, allows for those mappings to be passed across to your CAFM systems, but also we can produce Excel outputs of the same information if perhaps your partners aren't ready at that stage yet.

Now all of this, the purpose is to create that fast start for industry. We recognise when you are taking on a contract as an FM service provider, or maybe as a building owner, you are looking to do a tendering process. Or maybe you are an FM consultancy helping out either of those particular areas. Now as a user of SFG20, we want to help you get the quickest start you can whenever you are taking on a new building or starting a new contract.

We want to make it as easy as possible to assimilate that asset data, assign the correct maintenance to it, and then move that across to operational systems. So we've built a whole new load of functionality that hopefully solves all those problems. It takes that data in any Excel format. It moves it through an AI mapping process in terms of assisting and providing you guidance and where key decisions need to be made around your maintenance, and then it produces that set of maintenance-ready asset data.

So now what I'm going to do, I'm going to pass on to my colleague Mike Talbot, who's going to give you a sneak preview of this wonderful new functionality. Thank you.

Mike Talbot: Okay. Let's look at how we can use asset registers inside Mobiliser. In the upcoming release, Mobiliser includes an asset registers function that allows us to pull in Excel files that contain an asset register, either from a COBie file, which is a standard format of BIM that allows you to get an asset list or just from an ordinary Excel file, and you can just import those things by choosing a file.

And very straightforwardly mapping through what are the columns in those files, which spreadsheets do you want to import, and so on and so forth. Very straightforward. COBie files even easier. You just point the file at it and it imports it with no need to set it up. Once you've done that, you end up with a list of all of your assets with the information that we've been able to glean from them, from that Excel spreadsheet.

And this is nice because what we can do is import multiple spreadsheets in different formats and create a unified view of them. So you can see I've got a whole list of different types of asset. We go in and have a look at one of those. So there's an air conditioning unit over here. We can see where it is, we can see its Uniclass codes, we can see other information about it and so on, and obviously we can filter these things. We can search for things, and so on and so forth.

But probably the biggest job is that mapping of this description of an asset into an SFG20 code so we can work out what is legally required for it, what is best practice for it, what you should do to maintain this thing. So in order to do that, the asset register system here identifies all the different types of asset that we've got coming in that we've mapped in from the files, and it allows us to then map those to SFG20 using AI to support that process.

Now by hand, this job will take hours. It took us 100 hours to map our current client's register by hand into SFG20 codes. So this is a tool to help with that process. And what we can do with this tool is we can just go over to the mappings here and I can show you what it's done. So it uses AI and what happens is it scans the description and some example assets of that description, and then it produces this intelligent mapping and that can automatically map some assets and certainly significantly reduce the time for the rest of them. I'll show you how.

So, as you can see, the AI here has come up with a series of questions it wants us to answer in order to work out whether or not this asset is the right type. So it's asking us if it's a split system or heat pump with a direct expansion. It's asking if it's located in an indoor room or space, and so on and so forth. And then it's automatically guessed the answers to these questions based on the information it's been able to pull from that asset description.

And so it's decided that this is a room air conditioner. So it's an AC unit type one. It's inside that building and it thinks those are the questions. You can see it was pretty convinced that was the right one. It thought it might be a split system or heat pump. And we can see immediately from this, all of the different stats on that.

So this has got seven legally required or critical tasks that need to be done. We can go and view that schedule. It's mapped it to, so we can just go and make sure this is a room air conditioner and what you need to do with a room air conditioner, and how long those things take, and so on and so forth.

And over here, if you don't want to bother looking at the whole schedule, it'll tell you just that short brief overview of that. So what the AI has done is it's taken the description, it's mapped it against the thousands of different schedules in SFG20. It's then worked out how to differentiate the top most likely ones from each other.

And it's given us a ranking of how likely it thinks each one is. As you can see, it thought three of them were possible. And it didn't think any of these were possible really, but they could be there in extremis. And then it gives these questions, but you can change the answer to these questions.

So if you did know it was a part of a heat pump system, you can just click that answer to yes. And it changes the stats down here. And now it's 100% sure it's a split system heat pump. Switch back to 'I don't know'. And it's gone back to a room air conditioner. So you can choose one or more of these schedules here, or if nothing matches there, you can manually map it.

Or you could say, I'm just not interested in the SFG20 codes for this particular schedule. And that's a pretty powerful tool because once we've done this, we can convert this into a maintenance regime. So if we come over here and look at my maintenance regimes, I've got this test maintenance, test register maintenance regime over here.

And as you can see, this has got my medical dental clinic in it as my one facility. And the way we built that is very straightforward. We go over to the asset registers tab here. This is a tool that our professional services people use in the first instance. And what we can do is we can just take that register that has been set up and we can just synchronise that register into this regime to create the structure of it.

And when we do that, we could choose to organise it by the physical structure of the building. So the facility, the floor, and the space that contains the asset. Or we could do it by categories of assets, so we could have all of our air conditioning units together, all of our fire prevention systems together, those kind of things.

So either way we can do it either of those two ways. This one's been built up. SFG20's typically been a book in the past, a way of describing the maintenance needed for a particular structure or building. But what we want to do is to make sure that you are applying the minimum maintenance necessary to get the job done.

So that means potentially, even though you're using one of our schedules, those schedules have optional things in them at times because it's about the particular equipment that's fitted. So something may not be fitted. And we can do some work with that. So if I go in here, you can see it's built it up to these various different floors.

We've got some systems here. We go into the first floor. We can go and find, for instance, in this case, let's go for the pharmacy dispensary and we can see immediately what's inside that location. And then we could go in and have that as the book, tailoring as we needed. But a really important way to think about this is probably you are thinking about your maintenance tasks as being maintaining assets.

So thinking of it not as a book, but as a list of assets and which assets cost the most to maintain, those kind of things. So we can also go and do that over here with the assets tab. And here you've got all of those assets again now, but now we can actually see what tasks we need to perform on them and how long it will take to maintain it.

So we can see a complete list of each of our assets and how long they take to maintain. So I can sort that list so that's the fastest to maintain ones. And then I can sort it in the other way. And you can see we've got these HVAC ducts taking five hours, we've got over here some switches taking a long period of time to maintain and so on and so forth.

So we can do that. Very, very straightforward. But then what we can say is, okay, well look, these switches, I've actually mapped them to something else just so I could do this demonstration. We could just say, I actually don't want to do any of the amber tasks on those particular switches, so I can just turn off all of the amber tasks for that particular category of thing.

And it'll map that stuff through, recalculate the whole of the results, taking out the amber tasks there, and leaving only the four red tasks, significantly reducing the amount of time needed to maintain those assets. So it's a very powerful system and it produces an output that can go directly into a CAFM and actually set this stuff up.

We can also tailor individual things as we map them in. And say, for instance, if it's a piece of equipment where something isn't fitted, we can just turn off that thing. We can see where optionality comes in quite easily by looking at the types in the register. So if I just go and say, find some pumps like so.

Then we can see we've got cold water and hot water pumps here. So if we go down to the cold water pump, we can see it's selected circulating pumps general, and you can see that there are two options on those for green tasks: if pulleys are fitted, and let's say we know that they're not, and if drive guards are fitted, and let's say we know that they're not, so we can just turn those two things off and then we'll end up with a tailored version of SFG20 that doesn't tell people to do tasks that don't matter.

If we go back to the hot water pumps, we'll see that's been mapped to secondary hot water circulating pumps, and they have two optional things: bolts, pulleys, couplings, and belts were fitted and drain and draining dish were fitted. Let's say there's no drain and dish.

So using these tools, it's very quick to make a version of SFG20 that actually fits the assets which are in place, which is a primary reason for using this technology.

Lisa Hamilton: So my first question to you is, how accurate is the AI recommendation?

Mike Talbot: Well, I mean, I guess it's all dependent on the data coming in. So we really want to take Excel spreadsheets that you've got, the data you've got and do what we can with it. But what it does, it does a three-phase process and it's pretty good at finding that list of things and then it has some rules based in it that says it has to be pretty sure it knows what it is before it'll map it. Otherwise it asks you to answer those questions and choose one, but that's just a massive saving.

Imagine how hard it is for a person to know our 1,500 schedules in detail. By having this approach, you know, we just significantly reduce the time and we can automatically do a significant amount of it and it's as accurate as the data provided to it can be accurate. If your description is 'boiler', we probably don't know if it's a steam boiler, an electric boiler, a condensing boiler. If it says condensing boiler, you're done.

Lisa Hamilton: Absolutely. Thank you so much. So my second question is, some people in our audience, perhaps they're working in healthcare, you know, it's not uncommon for them to be maintaining thousands upon thousands of assets. So you've just showed us about how AI is used to clarify an asset, but what about the majority of the assets? How much of this process is actually automated?

Mike Talbot: So what we do is we pull in the data, it can be tens of thousands of assets. That's not a problem. And we identify the types in those assets, and then that's what we map. And then we can generate a maintenance regime containing all of that stuff, and then make it available to a CAFM so someone can actually go and do the work.

So, yeah, I mean, it'll handle very large volumes of data. You know, clearly this is what we're releasing initially. We have lots of plans to continue to improve mapping old and inconsistent Excel spreadsheets further and further to make it better and better. And there's lots of ways we can do that, but I think it's a very useful use of generative AI. So we do use OpenAI's ChatGPT models to do this. And the reason why that's so good is they're very good at reading things like our schedules, summarising them, and then making inferences about mapping types. So, yeah, I mean, they're going to be capable of doing a lot more than that, but just at a starting point, it's saving hundreds of hours of work.

Lisa Hamilton: Yeah. And I suppose the more assets you've got to maintain, the more you stand to benefit from using this technology. Yeah, absolutely. Right. Okay. How does AI improve the process over doing this job manually?

Mike Talbot: So when we, how it works is, and this is what you sort of have to do as a person, is you read an asset description and it also finds a couple of examples of that asset in your asset register, and then it has to search first to find a list of schedules to do. Now, if you have to do that by hand, I mean, it'll take you a couple of, I mean, we've got some great tools for that, but it would still take you a while to do that interpretation. Now, AI is doing that, it's basically like ten people at once working on it, ten of them at a time.

So they're taking, they're splitting the whole thing up and they're doing that first bit. Searching through, then it pulls back those schedules, and then it has to really examine now which of those schedules is better and rank them, and then come up with those questions. So each of those questions is uniquely written based on the asset type description and the schedules it pulled back.

So yeah, I mean those jobs, I mean, it's just a huge thing. So the ability of it to write questions is fantastic. And then we have another pass where it answers those questions or says it doesn't know if it can't work it out. And that's just a really interesting use of large language model AI.

Lisa Hamilton: It sounds really exciting. It sounds great. Groundbreaking.

Mike Talbot: Yeah, I mean, I honestly, there's a new tool in my toolbox, in my team's toolbox, which is this ability to do straightforward comprehension, synonyms, all those kind of things that used to be very, very hard to do. It would be very difficult to do that, and it's become much easier.

So we've been able to sort of tailor that and pull that right into the core of what we're doing with the asset mapping stuff that you've just seen there. But I think the other key thing is we've also re-engineered our platform to be not like a book and to be like an asset list. And that's an incredibly important part of this as well.

So combining that AI ability with our ability to express things as assets rather than just as schedules. So really actually thinking about for this asset, what tasks need to be done. And I think that's very important. I did notice in the chat, someone was asking about some of the zero-use stuff, and we have a whole platform in there that goes through and identifies all of those places and lets you make those risk assessments, work out what your zero-use timings and frequencies are for vast waves of assets in one click.

And you know, in the future, yeah, we're going to be doing more AI stuff and the reason why we can't just do that straight away is we really need to understand what is the difference in a hospital versus a prison versus an office building versus a residential for those things. And we're going to get the AI building up those kind of things. So even more of it will be done automatically in future.