In our Podcast series “Sh*t You Wish Your Building Did!”, Memoori explores the intersection between technology and commercial real estate through interesting conversations.
Using ChatGPT to Augment the Experience of Smart Building system users. Instead of naming their dogs, Bitpool has put ChatGPT to good use, connecting it to their API to create their own Chat Bot for Building Data. When the user initiates a chat with their platform the Bitpool API sends all of the visible time series and named data to ChatGPT for analysis and response, the user can then engage in dialogue with ChatGPT to gain more understanding and insights from the data.
We sat down with David Blanch from Bitpool to discuss how they did it, what the results were, and why he is super bullish about using artificial intelligence to enhance users’ abilities rather than replace them. Artificial Intelligence has the potential to totally shift the paradigm of UI and UX. Generic data dashboards could be replaced with dynamically generated data representations based on the specific user’s requirements at that specific time. That is quite a profound concept in the context of commercial building data, where there are so many different stakeholders (Facilities Manager, Portfolio Manager, Investors) who need different data presented in different ways.
Don’t forget you can now Subscribe to ALL our Podcasts on Soundcloud, iTunes, YouTube and now Spotify!
Transcript of ‘Using ChatGPT to Augment the User Experience’
James: Hello everybody, welcome back to Sh*t you wish your building did this is the podcast where technology experts tell you how to make your building smarter today we’re speaking to David from Bitpool who have put chatGPT to good use creating their own chatbot for building data so stick around for this interesting conversation where we discuss a real-world application of this hugely influential new technology. If you’re serious about making your commercial building smarter then you’re in the right place every episode we talk to an expert in their field and discuss practical business advice on how to implement technology in buildings and if you like what you hear don’t forget to subscribe to the channel you’ll be notified when we publish new content so let’s go.
David welcome to the podcast. Great to have you here maybe you can start by just telling us a little bit about Bitpool and what you’ve been doing with the chatGPT.
David: yeah cool man um so so bitpools are building data Cloud platform um we specialize in getting data out of buildings at extremely low cost and putting that into people’s hands and um our chat GPT Journey really came about with all of the the hype in buildings with everyone you know uh writing poetry with it and uh writing job descriptions and what have you and um we thought we’d put it to some good use and and feed it some building data and see see what came back from it um you know we we often you know experiment with new technologies I won’t say that we’re uh experts in the field we sort of just came across chat GPT and thought well how could it help our product.
James: Obviously it’s interesting that people are playing around with it but it’s great to hear some kind of real-world experiences and you know I came across your LinkedIn post which I thought was super interesting and so you’ve basically been feeding it you get data from your API the bitpool API um and then it’s been you’ve basically created um a chatbot right for your for your application is that is that right?
David: yeah I mean Deadpool goes a little bit further than that in the sense that with our product we always wanted it to be simple and for anyone to use you know one of the premises of our product is it’s not just for engineering Specialists or um you know energy professionals things like that it’s like anyone can get into the platform and use it and create some value out of their data so when we saw chat GPT we saw the opportunity to bring that into that world are supposed to to help people understand what they’re seeing from our platform it wasn’t as much about us pushing large chunks of uh building data to it it was more about being that assistant for the user to say okay well this is what this is what I can see and this is what I can tell you about what you’re seeing um you know where we’re super passionate at the moment about this this problem where there is a um a diminishing skill set in the industry and we’ve got to try and be able to bring people in who aren’t from our industry and and help them understand what they’re seeing so they can actually participate in the conversation you know otherwise as an interview so it’s sort of isolating people by virtue of how complex buildings can be and the technology we use yeah exactly I mean and so it’s like augmenting the user experience and also then the ability of the user to be able to kind of get a deeper understanding of this data right.
James: You’ve obviously been seeing now how your users have been interacting with it like if you’ve got some use cases.
David: Yeah it’s it’s interesting because when you talk to different people we don’t have a large subset using it of our users there’s and there’s users using it in different ways which I’ve found interesting um the most common uh comment that we do the feedback that we get at the moment is it’s taking a lot of the workload out of it um I suppose for everyone to understand so chat GPT and that that API there’s there’s more iterations to come and there’s a certain level of accuracy that you get with this place but what we’re finding from the users is it’s currently it’s taking the workload off them having to mentally understand what they’re seeing in front of themselves um you know from from my own personal experience you know it’s been coming back telling us about missing data sets so one of the most common things that happens with when we’re looking at energy consumption let’s say in a building uh we may look at it in you know the last week in hourly intervals but the actual interval data is 15 minute or five minute interval data and there may be data missing from that subset so what’s been interesting is when I’m looking at a dashboard or or an energy consumption graph and I’m looking it looks normal to me there’s actually missing data in that in that visualization that I don’t even know about until I look at the the data behind it so to speak I don’t know if that’s making sense but it’s it’s sort of picking up the things that you just wouldn’t see as the user um which has been amazing for me to say to be honest.
James: Okay and can you also then use it to like query the data so viewers then you as the user write in okay tell me I want to see data between this specific data between this time and this time scale is that how yeah.
David: yeah it’s it’s you know what what ends up happening is you enter into a more of a conversation you know with you know the bot so to speak so it always so when when you first open chatGPT or the vehicle GPT in this instance um we push it a heap of data before you even know before it even starts the conversation so so chatGPT knows what it’s about to comment on and talk about and then you enter into this conversation with it where you can go deeper and deeper as you feed it more information so you know one of the common things is as soon as you open up our GPT window it will come back and tell you what it’s saying and then you can say okay give me some more insights to that data and it will drill down and drill down and the more you interact with it the more it can refine and hone in on on what you’re actually looking for um which is quite cool and then you can sort of ask it you know you can ask it questions completely out of context to that then as well you know you could say okay well um tell me why I’m seeing a peak demand issue as an example and then you say okay what could be some of the remedies around demand issue and can you write me a specification for that or can you give me some steps to follow and it will come back and do that um you know that’s really cool comments in the in the LinkedIn post was someone asked oh could you get it to respond as if it was Romeo or Juliet and we actually asked her to do that about the building data and it came back and rephrased everything for them it’s quite cool so um yeah that’s sort of some of the context of how we’re using it’ll say it’s very experimental like all of this stuff that you’re seeing in the Market at the moment um I don’t think we know where it’s going to take us but if we don’t experiment and try then then we would never know either
James: I mean that’s really powerful it’s sort of not just about visualizing and displaying the data but also as you said like providing remedies to it okay suggesting how that might be fixed and then also writing a specification for doing that I mean that’s that’s definitely super interesting.
David: it’s super interesting and I think there’s a there’s a whole heap of use cases that we don’t even know I mean uh one of the interesting things that I look at especially with our platform so we support QR coding um so we put QR codes on equipment and we often talk about the service technician on a service call um you know 10 o’clock at night and they’re in the plant room and they’re trying to understand what’s happening when you have an assistant like this that’s as powerful as it is and simply getting it to start to show you where the problems are it’s extremely powerful but I think like anything we we don’t understand how to wield that power yet um as an industry yeah it’s like I said it’s very experimental.
James: What’s next for you guys I mean they’ve said that they’ve already gonna release a new version of chapter GPT Google is going to come to Market with their version I guess um are you going to take advantage of those?
David: I mean we you know we’re on a journey of Our Own in a lot of Senses um the next step for us is obviously as as that API improves we’ll continue to invest in that and bring that um you know that technology to that platform but we’re also experimenting with a lot of visual AI as well so AIS we can create visualizations um we think it may get to a point where the traditional building visualization could be completely disrupted where You’re simply asking her to show you something rather than building anything at all um and you know that’s a very future state but if you sort of take any uh any Improvement whatsoever in this technology for it to be able to create a dashboard or create a report um I think it’ll get down to the point where the you know the the facility manager the service technician is simply just asking the technology to show it um what they what they need to see so we’ll create that on the fly so we’re very we’re very bullish on on all of the technology but um we’re more bullish on AI not completely replacing people like there’s there’s really two trains of thought at the moment um from from what you can see in the market which is you know some people take this approach that yes we’re going to automate people out of existence or there’s going to be this heavy level um level of automation um and then there’s another side to it where you know there’s a group of people who believe that it will augment you know human capability and if you look at you know things like our mobile devices and our computers and everything that we use today that’s sort of already happened and we sit on that side of the fence that we think this technology can can augment people’s ability um especially when you start to see the the skill shortages that we have in our industry and and how people are trying to catch up and understand what’s happening in buildings as it is um today so yeah we’re very bullish on that side of the fence which is this technology can certainly help people in buildings be more proficient understand what that understand large volumes of data a lot faster you know things like that.
James: Yeah I mean it it’s uh it’s really really interesting to see um and how people could potentially use it uh I mean and again that that ability and everything what like what you’re doing you know that you’re adding this you’re adding um capabilities for the user it’s not really at this stage about replacing people is that it’s about giving them the better information to do a better job and actually potentially use less people to do the same or less brain power to do the same work correct?
David: Or give more bandwidth you know a lot of people will say okay we use this technology to to do less work or provide more bandwidth you know if if this could help someone uh do you know a factor of five times more work than they would usually do um that’s a great outcome as well yeah so uh I I really believe that yeah we’re only going to see this technology Advance more and more I mean the great thing is we have a lot of conversation about this with people and if you think about it did you even know about chat GPT maybe 12 months ago um you know no one had heard about it um or you know some had been following I mean we’ve been following it for a while but it wasn’t prevalent like it is now the the lesson I think people need to learn from chat GPT is that there’s going to be disruptive technologies that come in and uh really change the game and and the building industry um has traditionally been very slow to adopt anything than new so I think the one thing that’s certain is that the industry is going to be disrupted and it’s going to be very disruptive over the next five years.
James: Right and that’s that’s uh an opportunity right not just uh um like a barrier like and obviously knowing that like if we know that this is there’s going to be more um Innovative Technologies come to Market you can start thinking about strategies now about how you you deal with that.
David: Correct yeah I think um you know I’m excited to see what things look like in two years time yeah um you know if I take just a short to you Outlook but I think there’s a lot of challenges in the industry as well at the moment if you look at um I think we’re talking on the phone about you know energy and energy pricing decarbonization there’s a lot of work to do in buildings and I think Technologies like this can certainly help and speed that process up um but I always think you’ll need that human in the loop you know uh so do you think that’s sort of people like the sorry I was just gonna say do you think that’s like a misconception then um about yeah this kind of focus on job replacement yeah I think people always want someone to feel empathetic about their problem right.
James: This Focus then the people have on like AI being about replacing jobs is kind of a misconception is what people are kind of missing the point a little bit all right?
David: Yeah look do I think AI is going to replace some jobs I think it will um for sure um but there’s a big misconception you know or and at least this is my my opinion um people love human interaction so when there’s a problem um and and you know with our customers on a daily basis I’ll talk through problems with customers and it’s not so much you’re solving the problem but it’s the journey that they go on with you to solve the problem so you’re sitting there you know they learn something I learned something and it’s great and there’s there’s two parties who are empathetic maybe to each other’s cause we there’s an emotional connection uh with it when you’re talking to a chatbot there is no emotion you know behind it you know if if you’re successful a human being shows that emotion that there’s excitement there’s some high fives or whatever you’re doing but when you’re talking to a chatbot it’s pretty raw um I personally don’t see that but I’m also cautious of my words because I can be proven wrong very easily um but I really do believe it’s in human nature to want to solve problems together and thinking that that’s going to be replaced easily um you know yeah yeah it could be a misconception.
James: I’ll just go back to something you said before about the data visualization I mean that’s like really really interesting because I think again it’s sort of we get fixated on dashboards right like that’s the way that data is presented at the moment that’s probably that’s how it’s going to be but like again like this could be a create like a totally different and much more intuitive system of just aren’t creating it on the Fly giving you exactly what you want at that time.
David: Correct yeah it’s um you know we’ve been experimenting with mid-journey if you don’t know what mid journey is but even just for creating imagery for our Graphics in these systems we’re using mid Journey um and just experimenting with that so we’ll ask it to create a power meter an image of a power meter or um an air handling unit or something and it will create that imagery pretty accurately but like I said if you take any form of improvement on that technology there’s it’s not within the realm that it couldn’t create that visualization for you um now if it can correct the visualization for you and this is where it gets really interesting is it’s going to create the best version of that visualization for you to understand the problem that we have in buildings at the moment is we have Engineers that create these visualizations and assume that the user will be able to understand and consume the information um but that’s not that’s often not the case that’s very often not the case especially when you have a non-technical user who’s trying to consume and gain Insight so if you picture a world where you could ask something to create that for you and if you said basketball I still don’t understand and it completely created in a different way and you kept going backwards and forwards with that conversation inevitably you would get a better outcome you know you would get you as the user being able to understand uh that system a lot better in my view yeah especially when you talk about multi-building you know if we talk about is there any form of scale if we said okay there’s 50 buildings with you know you know 10 000 points in a building or something like that that’s a lot of data to consume an engineer and Design um but yeah there’s there’s also the financial aspect of that so if you think of what it would cost to build those visualizations and for someone to manually do that if we don’t have to do that anymore we can reinvest that money in other areas of the of the building um into you know Green Technology or different Technologies to improve the building so yeah it’s a it’s definitely a rabbit hole you can go down it’s happened again.
James: Like going back to what you said I think with those with the different types of user right like everyone learns differently everyone and obviously within the organization everyone has different functions within that or different they have a different perspective right so to be able to give them a visualization based on their level of understanding and you know where they’re coming from I think that’s incredibly powerful so it’d be really interesting to see what you guys do with that.
David: If you think about it too we don’t do that right now you know we just create the static visualizations and we say that’s how it should be now we we’ve introduced you know UX and UI design and things like that to try and make this more simple but if you look at the purpose so if you look at the purpose of why someone would invest in technology in their building it is actually to answer questions it’s to gain knowledge but the systems themselves are designed to only provide that knowledge to particular skill sets of people ideally we don’t want that ideally we want anyone to be able to consume information in the way that they can best understand it and if if AI helps push that that direction I think that’s a that’s a great thing um you know and that’s why I’d like to still go back to that companion piece I picture you know like a portfolio manager about to walk into a meeting pull openly at their mobile and just asking hey can you quickly show me our carbon emissions through the portfolio or something to that effect and this assistant has simply builds that for them and shows them and I think you know some people will say oh you know that’s really Pie in the Sky that’s that’s really out there stuff but I just don’t think it’s that far away uh to be fair and that’s what will disrupt the industry.
James: yeah David those awesome talking points thanks for sharing uh your experiences um I mean if anyone wants to find out more about what you guys are doing what’s the best place to do that?
David: Just hit the bitpool website uh reach out to us on LinkedIn we’re always happy to chat uh but yeah hit the website and reach out we’re happy to happy to show you guys what we’re doing.
James: Great thanks a lot.