Transcription
Welcome to today's practical webinar on promising AI tools from the German real estate industry. An overview. Um, which AI tools from Germany for the real estate industry will help us today? Uh, that's a bit of the subtitle we've chosen here. Why? Um, there are very few tools, uh, probably not just in the German real estate industry, but internationally, uh, that truly have an AI foundation. Of course, uh, many companies are now integrating AI into their tools, but, um, that's a bit different, because I believe that at many points it's difficult to add further AI functionalities to an existing tool. But if a company, I'd say, was founded three years ago, um, and, uh, then already directly uses artificial intelligence as its foundation, as its basis for what it builds or uses within the company, then, um, that's a different, that's a different topic, because it's somehow built a bit more modularly, also in one place or another. Um, and that's why today we want to focus only on the, uh, companies that truly have AI as their basis for what they do and, uh, offer something for the German real estate industry. Um, that means, under all aspects and conditions that we all have in mind every day, um, can act. If we now want to transfer an AI tool from the USA to Germany, then there are probably many things that don't fit yet, um, that still need to be made to fit. Uh, and, um, but with those that come from Germany, for the German real estate industry, those are rather the ones where I would say, you can already, uh, use them directly accordingly. Yes, and, um, you see here practical webinar. Can I mark this? Yes, I can mark it. Practical webinar, uh, again, I said it very briefly before, but, um, we want to, uh, show you, uh, what is already possible with artificial intelligence, um, in the German real estate industry. That means we regularly hold such practical webinars, uh, where it's really about, um, going into the tools, looking at what they can actually do, what I can do with them, uh, and then partly testing the whole thing with, uh, practical examples. Um, the goal of all this is simply to really bring AI closer to our, uh, to our industry. It's less about, um, really looking at sites. We have our own site webinars for that. Today, site will play a very small, very small role, but it will be just as big as, um, that of the other tools. But, um, it's essentially an educational format that we have here. Therefore, feel free to check LinkedIn regularly or our website, um, for the next, uh, the next episodes and the next webinars. At the end of this presentation, uh, I have also included a QR code for you. You can register directly for the next webinar, uh, there. Yes, and we take, I'll say now, and probably again at the end, uh, we always gladly, uh, take suggestions, ideas, and, uh, and, um, yes, uh, and criticism, uh, on board. Um, briefly about me, I've been part of the, uh, industry for, uh, yes, 7, 8 years. I was previously at BNP Paribus as a project manager for digital strategy, where I looked at which proptechs and startups could be, um, integrated and used there. Then I came to Side in 2022, uh, actually 3 years and 13 days ago, and, uh, now for about, I'd say, one, two, no, two years rather, uh, also as a lecturer, as a keynote speaker at ADI, CIA, uh, BDB, iWS, and one or two other, uh, organizations, and there I always show in a very practice-oriented way what is possible with artificial intelligence for project developers, for, um, real estate agents, etc. Um, why? Yes, I'm not a developer, I'm not a techie, I can't do any of that, and, uh, if I see code, then I could run away, and probably most of you feel the same way. Um, but I was still able to teach myself a lot and automate my daily work, and I always quickly see what we can use how in, uh, the real estate industry and what we can do with it. Um, and the implementation of this is super, super simple and easy. Of course, with all things, you might need an hour to look at it more closely, but then each of you can, uh, with the appropriate tools, really take the next step and, uh, build automation in the right direction, so that, um, s, that your own work or your own working time is used more efficiently, because in the end, someone else is working, namely the AI. And, uh, that's something I've personally noticed for myself. And therefore, from this, from this practice-oriented approach, that's where I come from, and, um, want to convey that just as well, because I think if I can manage it, then anyone else can actually do it, uh, who wants to deal with it a little bit, and that is, uh, yes, my full seriousness, and it's also getting easier and easier. So AI is getting better and better. Um, the whole thing is increasing exponentially, AI is getting exponentially better. That's something that a human can't really perceive. And, uh, I sit down every two months and think, okay, wow, uh, with this tool, we've really reached the next big step. And, um, what I always do, and also today, is, uh, to really show you what's possible today, what are the real practical tools that you can use today, that we have available today. Uh, what I don't want to do is, uh, give you any future outlooks. Maybe a half-sentence might slip out, uh, about how things could develop. But, um, there are many, yes, webinars and a lot of information that are only about, uh, how AI could help us in the future. Uh, then it says that in three years, this could automate real estate valuation, uh, to a certain extent. Um, all of that is actually possible today already, and, uh, I don't want to make these future outlooks per se. I want to show you what you can do today so that you can use it directly. Uh, and therefore, the focus is on real tools. So, honestly, the first time I gave a presentation on this, um, I realized that the content, uh, that creating the content for a three-hour presentation wasn't the difficult part for me, but creating a presentation with, um, with, uh, Google Sheets or PowerPoint, and, uh, I had absolutely no desire for that. Nowadays, I use Gamma as an AI tool for that, which makes it much faster, but, um, out of necessity, the idea was born then, we actively look at, um, we actively look at real tools in a practice-oriented way, and, uh, that's fundamentally something that, um, that I would gladly continue like this, because that's how I would like to learn. Um, this is now the last, yes, theoretical part or the last theoretical slide of the, uh, of the presentation, and then we'll go directly into the tools. But there's one more thing I want to give you, or rather, I want to leave you with. Um, we hear the word AI on the news every day, um, at least once, I feel, because some researcher has done something special with AI again, and AI is always explicitly mentioned there. Um, what exactly has been done with AI there, I sometimes don't understand myself, uh, and I'm a bit surprised, um, that AI is needed for such things or, uh, yes, or also, um, AI approaches are being pursued. But okay. Um, but it's suggested to us a bit that AI is the "all-singing, all-dancing" solution on the one hand. So, um, I'm a real estate agent, for example, press a button, and my entire work is done directly by the AI. That's what's suggested to us. Um, but logically, someone found the tool for that. Uh, and there's no tool for that yet. Um, so it's not [laughter] like that, it's not an "all-singing, all-dancing" solution that we, uh, find, especially for the areas we are active in. Um, but through, uh, large language models like ChatGPT alone, and the connection of these large language models with our email inbox, with our folders in Drive, and with other, uh, with other tools via N8N, for example, an automation tool, like Make.com or Zapier. Um, through these tools, it is quite possible, uh, to build this "all-singing, all-dancing" solution practically yourself. Uh, as a real estate agent, I would still have to lock the door, of course, uh, and do the viewing, but theoretically not the rest anymore. Um, and that is quite possible even now. No one is doing that right now, or no one is building that up right now, um, because it's very, very extensive. But, but, um, with the possibilities we have right now, uh, one would break down this entire process that a real estate agent does, to, um, now let's just look at real estate valuation, and then break that down further into individual small process steps. With these small process steps, each of these process steps is already representable today via a large language model. So research the rental and purchase prices at this point, um, with access to public sources or, uh, partly also with access, um, to non-public sources, and then find out what the condition of the property is, um, by looking at the pictures, um, and then telling me on a scorecard from 1 to 10, uh, what condition is the room in, if you enter that beforehand, um, and then you can calculate and continue with that. But these are all small process steps, and all these small individual process steps are fundamentally already possible with AI right now, and, um, the tools we're looking at today also have an API, an interface, in most cases, and you could integrate them into your own process in such a way that in the end you wouldn't have to search for or open the tools, uh, nor would you have to open anything other than your email inbox, uh, because everything is already automated in the background, and that already works, that's already possible. Um, as I said, I don't think anyone is currently building, uh, this all-in-one tool, uh, and wants to build it. But if you want to, you can build it yourself, it's already possible. Yes, so. Then let's go into the respective tools and see what the German, uh, real estate landscape for the German, um, yes, no, what the German proptech landscape can already achieve for the German, uh, real estate industry. Five, six tools today that we'll look at, uh, where we'll also go in directly. And, um, the first thing I definitely want to show you is, uh, Alpha Prompt. Um, like other tools we'll look at today, Alpha Prompt is a, uh, a tool where I would say, uh, it, um, uses AI exactly for what AI is for, or what AI is made for, and can do exactly that, or, um, addresses exactly what AI does best. And that is, first of all, to extract data from, um, from documents. And, uh, Alpha Prompt does the following. Um, I can upload thousands of documents, hundreds of documents, thousands, sometimes just three if you only have three or, or want to use them, and then these documents are automatically read by several different AIs that are behind it. And these files are then automatically extracted. Um, the connections between the files are checked, and then or between the documents, uh, rather, are checked. Uh, and then a structured data room or a structured data set is created from it. All of this sounds relatively theoretical. Uh, let's move on. We can also look at the, uh, website of Alpha Prompt here. Because what is all this good for? Well, if I have a portfolio, um, with 15, 15 multi-family houses, um, and I want to check it, then it's about, uh, extracting the necessary information in a short time. With large transactions, sometimes, um, several lawyers from both sides are involved, who essentially look at all the information, read everything, and move forward with it. And, uh, Alpha Prompt structures these data sets first. That means, maybe it's not a transaction, uh, but maybe it's the asset manager who has 15,000 documents in some folders, where it takes two hours to find something specific. Um, but with Alpha Prompt, I can simply connect my folder, or I, um, or I provide the files in a different way, and these files are then read, whether it's a PDF, a Word document, whether it's a scan, or whether it was written or typed at some point, it doesn't matter. Um, these files are then read and structured. That means, for one property out of the 17 properties in there, I have seven craftsman invoices. Um, which is the youngest, uh, which is for building for apartment 3 or something. Yes, and these are all things that Alpha Prompt can read directly. And, um, we'll also take a quick look at that and go into the, uh, into the overview. I need to open it here and go directly into Alpha Prompt, because this dashboard, it's still in the development stage, but depending on when you're looking at this, uh, you'll probably already have the option to access this dashboard directly. Um, I was kindly provided with a test dataset, and, uh, what you essentially do here, I'll go back one step, is first, uh, upload the documents, if it's in a number where I can upload the documents. If I have 15,000 documents now, then I probably won't sit here and drag them all in and wait 5 days for them to upload. But, um, if I have 200 documents for one or two properties, uh, then I upload them here. Um, I just need to, yes, drag them from the computer here or connect my dataset, my folder, and then, now I've just missed it, then I click on classify here, and then the AI first looks at all these documents and can first assess what it is, what it's about. And, uh, then all these documents are stored here in the data room, and they are already sorted and organized accordingly. For example, with the warranty bond, uh, we have several, uh, documents for this property here. So these seem to be different, uh, house numbers here, and there we have several documents that can all be assigned accordingly. So first, the first thing is always this assignment to a, uh, to a topic, to an, uh, to an overarching category. And, uh, there I now have construction measures and maintenance, and immediately see everything that belongs to it. Uh, I can also look at the whole thing and then see here, uh, the corresponding, uh, the corresponding explanations, the corresponding data. Uh, and also a direct summary, um, of what's included in the end. So when is it from, uh, what is the client name, what is the contractor name? Depending on that. So this template for what is, uh, extracted, I can of course, uh, I can also customize it myself. So, what are the most important documents in there? Um, and then everything is put together like this. Management, yes, management invoices, which we also have in there, data quality, etc., and I get a very, very structured data room first. That's the very first thing. And in this structured data room, I can now do various things. One thing that, um, is still in development, as far as I know, is, uh, I can ask questions here in the Due Diligence area. Um, yes, in this area here, I can ask questions. Ah, no, that actually works in the version I have here already. Uh, and can ask questions about specific documents or can ask a question for the entire portfolio. Um, a simple question would be, what are the current rents for all properties in the portfolio? Uh, and what were the current rents, uh, when we, um, when we first rented it out, for example? Yes, I have the option here in Due Diligence. That means, I no longer have those two hours, uh, that I need to, um, yes, find the answer to a question from my 15,000 data sets for a property that I need here, but I can simply ask the question here, uh, get the answer, and of course, also get the link where the answer is included in the documents. That helps immensely. And on the other hand, I also have various dashboards here. Um, these dashboards summarize what, uh, could be particularly important for me or what is particularly important for me, because I have set it up accordingly. In this case, for example, we have a Lease Portfolio Overview. Uh, there we see, this is the monthly cash flow, this is the monthly rent. We have one unit in there. Um, but of course, it works the same way if we have 15, uh, units or, or hundreds or thousands. And, um, this overview in these dashboards, I get it here, tailored exactly as I would like it, with the corresponding revenue mix, lease expiry, timelines, and whatever else I can imagine that is sensible for me here. You see, you can open and add many things here and then look at them. That means, within a very short time, we have a very large, comprehensive overview of the portfolio, of the properties, of, uh, the transaction, and of all the documents that are included. And Alpha Prompt enables this because, of course, there is an AI behind it that reads all the documents individually, can relate them to each other, link them together, and thereby, uh, yes, compile these overviews, uh, exactly in the way I would like them. I can then of course also transfer them directly into my system. There is an API, an interface too. Um, then you essentially only upload these documents to Alpha, and then you say exactly which information should be read into your own system. Um, and then it's extracted there accordingly. Uh, and I have a very, very good overview, or alternatively, you use this, uh, this overview that we have here, uh, on the online platform, through which we then get quick access, can ask questions, um, etc. This is, um, one of the best ways to use AI, because, uh, with this data extraction, uh, there is the risk of hallucination, which you might have noticed yourself when using ChatGPT, which you definitely have. Uh, we have this risk significantly less here, because, uh, the AI only accesses data that it sees, and it usually understands it very, very well. At the same time, we also always have, uh, now specifically with Alpha Prompt, uh, here also the overview, let's take a look. Yes. Uh, we at least also have an overview of the, uh, of the confidence score. So, how sure is the AI that this data is correct? Um, there are different AIs behind it that check this repeatedly. That means, we have, uh, corresponding security, um, regarding this data basis. Yes, this is something I can definitely recommend, especially if you work with, uh, with multiple, uh, with extensive datasets. Yes. Um, exactly. And then we move on to the next, the next tool, uh, that we want to look at. IMOl, uh, honestly, is something that is, uh, currently, uh, in the first test phase with some test customers. Um, I had the opportunity to meet the founder, or one of the founders, and we exchanged ideas. And essentially, uh, I've placed OML behind Alpha Prompt because, uh, the foundation, uh, of MOML is relatively similar to Alpha Prompt. Um, we just upload files, uh, up to 500 files in any format, they can also be photos, etc., in any, uh, any types, any document structures. And these files are then used, firstly, to build a platform, uh, where I can ask questions about the files. Um, but on the other hand, that's what, uh, where, uh, im L ML is currently working very intensively on. On the other hand, I will also be able to generate an exposé with InDesign templates directly. That's currently being planned. Um, we'll take another look in a moment. That, uh, this, we don't see that right now, but, uh, the further functionalities are already very clearly visible, and, uh, this is primarily a tool for brokers. Um, I get from the owner, who has, who perhaps has a folder on his computer, uh, where it says House XY, and then there are simply, uh, yes, 500 files in there, with various invoices, with everything else. Um, and he simply sends it to me and says, "Yes, please create the exposé." Um, and then we can go into sales. That's of course what we don't want. Uh, therefore, uh, it's set up a bit differently, uh, a bit differently at OML. We can simply upload these files, and the AI does the rest. What we are seeing right now is the website. Um, there you already have a, uh, a short impression, uh, of what it looks like. It's primarily about research, analysis, and then corresponding execution. A lot more will definitely be added here with corresponding workflows, um, that are built with it. That means, uh, the data comes in, is validated, and then I can consider what, what do I want to do with it? Yes, you will perhaps, uh, be able to enrich this data, um, with information that we find from the internet about the buildings, and, uh, we do a document analysis. Uh, GDPR compliant, it is anyway. Um, workflow builder, that's probably the, uh, the core in the near future. But now I would say, uh, we'll take a look at Emo ML, uh, once, how it looks. And, uh, I also kindly received a login for that. So. And, uh, I myself uploaded some documents yesterday. We can also do that, uh, as we go. But I have to tell you now, what I upload here, it all looks something like this. It's, um, absolutely fictitious data. This will now be a fictitious rental agreement that I had generated by an AI. Uh, none of this is true. Yes, and if I now say, okay, these ten documents that I have here as an example, for that I would like, uh, now you probably have good access or see what I have in here. We might have to blur it at the end, the, uh, I'll take this data here. So. And we'll upload the folder. Uh, and create property. Yes, we first need to upload the data. Now the data is uploaded, and we'll upload everything accordingly here. Um, exactly. We are currently only uploading the documents here, and then we'll see, uh, what are the documents that I, uh, probably need to be able to really market the whole thing, uh, and what documents do I have, and where is the delta in between? Um, and what information do I actually have? And the documents have been uploaded. So, I tried this yesterday with the same, uh, same property. No photos were uploaded, otherwise it would look as nice as Musterstraße 10 here. But, uh, it's still taking a bit of time now to really look at all this data. The document types are currently being determined, and, uh, the API needs a few minutes for that. Therefore, let's go back to Home. Um, although it's already finished in one place or another at least. I already have an address. Yes, and it's now creating exactly these folders and telling me, well, in the energy certificate folder, there is now, uh, the fictitious energy certificate that we have, right? And, uh, first of all, this assignment is of course a big help. So, data room processing completed. I don't know if it's completely finished or not. Therefore, um, I'll take the, uh, the sample property, uh, that we were provided with here. And, uh, firstly, I can now chat with it directly. So I could, for example, say, uh, what is the property tax for the property? So, I enter this, send this message, and, uh, then these documents will be searched directly. Um, and I get the result, the annual property tax is €1250.50, uh, for the corresponding period. And how do I know that? Yes, from these property tax, uh, property tax documents that were provided to me or that I, uh, uploaded for it. That means, directly queryable. Um, in the overview, I now see, first of all, that we have an incomplete document dataset. We generally say that we can start marketing if we have 37 documents or at least their content. 15 are complete, 26 are missing. So now I know immediately what I need to, uh, ask the seller for again. And, um, I now see in the data room what we have here, but the document monitor, that's probably the crucial thing. We are missing, whether we need it for this property or not, is debatable, but we are missing the building permit, we are missing floor plans, we have the energy certificate, we have the tenant list, we don't have the deposit list. For the rental agreement, it's only partially complete. Um, here too, there's this confidence score. Uh, so we are 99% sure that this file is of the type rental agreement. Um, and then we see, uh, that it's only partially complete, some things are still missing, and that, um, you can then look at it directly here. And then, uh, if we have all these documents, then the next step is to go into the exposé, uh, coming soon, yes, use the exposé agent and simply press the button and say, I want nice templates, templates, I want a nice exposé from it, uh, that might be a bit more special, um, and that is automatically filled for me. In other words, the process is, I simply upload my data, I'm told what's missing. Um, I will probably sooner or later also have the option here to say, via the workforce, please send the owner an email, uh, and tell him that we still need the following documents, they are missing. Um, it will probably also be a click, and then, uh, we'll go directly into the generation of the exposé, and that with InDesign templates, so not with Gamma, for example. With Gamma, I build the presentation here. You could also create exposés with that, but, um, InDesign is still a bit more individual and, and nicer, and that comes in here. And then I have, um, a first major process step of what I do as a broker, for example, already, uh, already completed, and, um, can work with that for now. Yes, that's what im OML is made for and intended for, I can only recommend it, take a look at it. Um, a lot is still being built, but, uh, it makes a very, very good impression on me, indeed. Yes. Um, exactly. Then, uh, let's look at Alago next. Alago is a tool. Let's go to the website, we have it, we have it right below, we don't need to go to it. Um, Alago is essentially a tool for documenting construction projects. If I, as a project developer or also as an architect, or possibly also as a private individual who is having his house renovated, um, am on a construction site and look at everything and see, okay, uh, this is the current status, I need to document it. Um, we need to move to the next step. Uh, what should everyone pay attention to? Um, then it's always very, very time-consuming, uh, to really document everything accordingly. It would be easier, of course, if you provide it as a kind of digital logbook at the end, and, and do it, if you can then trace everything. But, um, if I were to create a digital logbook for this without Alago, then I would, uh, yes, I would always have to enter a lot, I would always have to, uh, have the notes displayed and incorporated as comprehensively and clearly as possible. Clear. Um, and that's not necessary with Alago at all. Let's look at the website and see what it offers. Simply to get a picture of Alago here with the corresponding screenshots. Um, by the way, it has also just completed a million-dollar pre-seed round and is, as far as I know, one of the, uh, one of the, uh, yes, larger AI, uh, startups, but also already in our, uh, in our industry, um, and it simply solves a very, very simple problem. So I can upload a transcript, um, that I, uh, recorded during a meeting with a, uh, with a, uh, worker on the construction site or during an inspection, uh, where we discuss everything and check everything. I can upload this transcript, it will be automatically, um, extracted, and the most important topics will be automatically extracted. Um, but I can also immediately, uh, record a voice message if I'm on site and say, um, yes, this and that needs to be adjusted like this and that. And everything is then presented to me clearly and concisely, so that the issue of, uh, yes, documenting the, uh, the construction process is no longer the big pain point that one might have, um, had before. Yes, um, I actually don't have access to Alago at this point. Um, but I have access to, uh, yes, these, uh, these, uh, overviews of the respective tools. Um, and I want to show you that very briefly. Um, I can open an assistant here with Alago. Then I just click on the fact that I, um
I believe you are hearing this now too, I'd better move on. Uh, then I'll just click on start recording, can end this recording, uh, and can add the whole thing as an entry. And then the transcript of what I've spoken will be automatically extracted. And I know exactly with the current date, maybe a bit fast in the last step, uh, with the, uh, with the current date, what was discussed, why, how, um, and what is the next step behind it? Yes, that is a very simple way to work with it. Um, I also have the, uh, the transcript assistant, so if you are in a call and there is an, um, an AI Notaker, which I absolutely, or the principle, I would absolutely recommend, uh, there is a Notaker included, and with this Notaker, you create a transcript, then you can upload this transcript directly here accordingly. And that will also be added and processed accordingly for the construction documentation. And, uh, it leads to the fact that you then immediately see the next steps, the to-dos. It leads to the fact that you have a perfect overview of everything, um, where you, uh, yes, um, where you still have to work on. And sorry, someone just raised their hand, that briefly, uh, that was the first time, which is why I, that just unfortunately brought me out of it briefly. Um, and you have an overview of all individual steps in the planning, during the construction, um, of the object, yes, or also during the renovation, for example. And the whole thing is again a use of AI, as it makes the most sense. Uh, because, um, you only have to say something and speak it, um, as you probably would anyway if you are on the construction site. Uh, and then you get a very well-structured protocol of what still needs to be done, what needs to be done, what in the future, uh, yes, um, one might not need to consider further, etc., and have this construction protocol. This is something that otherwise takes a very long time and, um, is not really, um, it's annoying, let's say it as it is. It's a bit more annoying. Yes, that's it for the topic, um, Alago. What we will now look at as the next, as the next tool is also a, um, yes, a Proptech from the German real estate world, from the real estate industry, where I would say, again, what AI can do very well, very well, broken down very simply to a specific activity, um, which many of you probably have or had or will have, namely, um, property management and the, um, speaking and the exchange, the contact with the respective tenants and, um, um, if the faucet, yes, uh, is dripping, who takes care of it, then you get a call or an email. Um, it's probably like this, then it's rather the call, and you want to record that relatively quickly. You want to be there for the, um, tenants with you and be able to say, well, um, we can now, uh, we can solve the problem together now. Um, of course, it makes sense to have someone on the phone who can immediately initiate the next steps. On the other hand, if you have very large property management, um, or if you have many properties in your portfolio, um, then it is not necessarily, um, efficient to have 20 people sitting on the phone, um, who are actually just recording something, or a damage, for example, recording, um, which an AI could also record for you, because perhaps you have already done it, that you speak with, uh, Chat GPT, with Google Gemini, with other tools. Um, and they understand you very, very well, and just as well, an AI-supported customer service can also be on the phone in the end. And that is exactly what, uh, what Managable AI does. Um, it was actually recently taken over by, uh, by, um, Kazavi, and Kasabi, Fasilio, and IDWell, um, are very, very easy to connect with Managable AI, by the way. Now I've gone one step further here, and then it occurred to me, for some reason I couldn't embed the website directly in the presentation. That's why we'll just go through it like this. The basis is simple, um, you have, there is a phone number. This phone number, uh, is called by your tenants, and behind it lies an AI, uh, lies an AI that speaks with your tenants, asks all the necessary questions that you need to know to assess whether it is urgent? Uh, is it important? Is it, um, short-term? Is there still a question, do I need to act, uh, as a human? What is, what is the next step? And finding out this next step, that's what the AI does completely here. And we'll go into that in a moment. Um, that actually sounds very, very good. Uh, what does "very, very good" mean? I think, um, it sounds better than better than if I were sitting there on the phone. Um, and that simply saves a lot of time, so that those who really take care of the problems no longer have to bother with the phone, um, at one point or another, but can solve the problems. And such an AI agent, who is available on the phone for the, for the tenants. Yes, increase in employee satisfaction. Okay. 48 hours saving potential per 1000 residential units. Um, these figures, they won't come out of nowhere. Actually, I do believe that one can, um, that one can break it down like that. This is also something. You can test it for 14 days for free, um, where I would recommend you try it. Um, go in there and, um, so. And this is the dashboard when you are logged in. And in this dashboard, um, you first have an overview of the inquiries that came in. Uh, I had one here, for example, an inquiry that came in. You can also test the whole thing well by simply calling your own phone number that you have here, um, and speaking with the AI. Um, this is the overview. How many inquiries did we have? How many, uh, yes, who are the top 10 callers? Who has contacted us most often? Um, when did the calls mainly come in? Um, but what is much more exciting is this inbox here. Um, here there was a damage report, uh, from Kurfürsten 50A. Uh, here we see the, uh, the transcript again, which was recorded from the call behind it. We can also download the whole thing as an audio recording. We can also listen to the whole thing again here. I don't know if you can hear that now, too. Yes. Uh, I, I don't know if you can hear it. We can also, uh, set up these flows ourselves again, but here you have your dashboard, your inbox for all, um, messages that come in. And the whole thing is then, of course, directly linked in Kasavi, in Fosilio, in IDWL, which leads to the process being created directly there accordingly. And these flows here, um, you determine what the AI should actually do, what should happen. You have, for example, for these, um, yes, these categorizations, into which it is automatically, um, sorted and entered what the inquiry was actually about. Uh, you have the option here, um, email sequence, call attempts, incoming, outgoing, exactly. Um, you have the option here to adjust the announcements accordingly. I don't know if you can hear it, as I said. Let's try it. I hope you heard it now, otherwise you had 10 seconds of, uh, silence here. Uh, in principle, you simply enter what the AI should say. Um, who should speak? There are various, uh, yes, voice frequencies that one can, uh, that one can use for this. Um, what follow-up questions could be asked, how should they possibly be answered? The whole thing, of course, then also with your, um, information, um, sprinkled in for, for example, questions that are regularly asked. Um, yes, you enter a farewell. There is a second announcement, uh, that you can, uh, that you can enter. So, in principle, you build this flow yourself. Um, how would you answer the phone, how would you speak with the tenants, and the AI can do that completely for you. Um, it can also forward directly to you upon request, so that you then, uh, receive this call immediately. Um, that is also simply possible. Yes, and that saves an extreme amount of time, and I, uh, yes, um, haven't looked again how expensive an account would be for me, um, to use it like this, but, um, I claim it's significantly cheaper than, um, yes, the working time that one otherwise spends on it. Therefore, Manageable AI is also one of those tools for me that does something that AI can do very well. Answering calls, but above all extracting information from them. Um, what it can do with a certain perfection and already incorporates. Um, I actually asked again yesterday here, the, uh, the voices here, they come from 11 Labs. 11 Labs is one, I think, the largest company, uh, that, um, yes, that offers this, um, this, this speech generation in the end, um, at the end. A lot will be added to this in the near future. You can probably do it with your own voice sooner or later, theoretically. Then the question is whether one wants to show the customer, um, that one is really on the phone or not. I honestly don't want to do that. I think everyone understands that an AI is on the line, um, with whom one can also speak, but, um, yes, you can probably adjust and rebuild a lot more in the future, but as it is now, perfect usability, um, to probably filter out 80% of the, um, inquiries first. You will also receive this presentation from me at the end. Um, I spoke with Tag about it and had a very nice conversation about it, which is why I promised many of you to include the link to appointment booking again. Um, if you are interested in Manageable AI, then book an appointment directly. Yes. Um, next in line, also something where I say, um, they understand what AI can do right now and how to integrate it into products very well and very quickly. Um, QuirPad has convinced me in the last, yes, one and a half, two years, um, primarily because new features and new modules are always being added, and always modules where I think, yes, that's exactly what we, um, what we need here, what is the first thing that makes absolute sense when integrating AI into our processes. Originally, um, what originally means AI-supported transaction platform is Quirepad. Let's take a look directly at the website. Um, that means I can say what my, uh, what my, um, what my acquisition profile is. Uh, what do I want to buy? What do I actually want to find? What do I actually want to see? Uh, others can also upload objects and say, um, they are for sale, and then it is matched very well by AI to see, um, whether this is an interesting object for me or not. That was the first, the first part of the whole thing. Um, what I honestly find more exciting is this, the exposé AI. We'll look at that in a moment too. This exposé AI, um, works like this: I simply forward an email or write a new email and simply attach the exposé there, um, there, there, and upload it. And as soon as I have done that, um, the exposé AI extracts all the information I need. From that, it checks whether it fits my acquisition profile or not, um, and simply gives me a very quick, very simple overview, um, of all these objects, um, that I have already looked at here. You can see it now. This is naturally more filled than, than, uh, my, uh, test that I have. And we'll come to the test now. Um, so yes, we'll first go into the, uh, the exposé reader AI. Uh, what I did here is simply sent an email with the, uh, exposé attached here to a specific address. I think exposé, uh, is the, then it can only work if I am also registered with my email address here. Clear. Uh, and now this exposé, it was again a completely fictitious exposé, has been extracted for me, uh, everything that I actually find in this exposé. That means, actually, I don't really need to read the exposé at this point. Um, and I immediately see, well, offer price 4.8 million, price per square meter, 70% rental apartments are planned. So, I see, in principle, first of all, everything that is really important to me, including the description, including, um, what else do we have here? Planning and building, uh, planning law situation. Yes, it's a plot of land. Uh, therefore 34 in this case, for example. Um, I see all of this at the push of a button. It also calculates what multiplier we would have here, what annual net, where at least according to the exposé, is assumed for rental units and and, um, then what comparable objects we have in the vicinity? This only works if I have uploaded more than one exposé. Clear, um, the whole thing is extended directly with real estate data. Uh, population growth relatively low for Berlin. Well, population growth, uh, which is included here. Uh, ESG criteria, yes, it's more of a vacant plot, I couldn't upload much there. Uh, and again, an overview, um, of all the documents. Of course, I can also upload further documents, uh, which then, um, support this even more. And this extraction of data directly from the exposé and then the comparison of these, of these exposés that are sent to me, that naturally helps me if I receive 20, 30 exposés a week and actually have to spend two days, um, looking at them, then this helps me immensely that I directly have this match score. The exposé that I uploaded, unfortunately, did not fit the acquisition profile that I uploaded at all. But you directly have a match score and see how much, um, it actually fits what I am actually looking for and then know, yes, if we are at 30%, I don't need to look at it at all. If I am at 90%, um, then I'll click in here. Uh, I'll look at the documents again or look at everything that's in here first. I'll look at the document. It has just been downloaded. Um, and on we go. Yes, it saves an incredible amount of time. It saves you hours. You see all important KPIs in the overview. You can, of course, then also sort them accordingly. Um, and you also see whether the object has perhaps already been sent to you by another broker some time ago, and therefore you see whether there might be another proof, um, do I have other, um, information about it, or has the price perhaps been reduced? That also always says something. If you, yes, the whole thing, if you have uploaded something, several exposés, you can also display it in the deal map. Um, this is now the deal map of the objects that are generally available on the transaction platform, which we haven't looked at yet. Um, you see, there are 180 results in Germany, verified objects. Um, you can also look there to see if you might find one or the other suitable one. In the acquisition overview for my profile, there is nothing here. Um, that's why nothing is displayed here. But otherwise, um, I immediately see what is particularly relevant for me here. Of course, I can also sell directly through this. Um, and QuirPad is really one of those where I say, or a Prop, where I have to say, um, if there are simple, good solutions for, um, unautomated and undigitized processes for a developer or a holder, um, then it will probably be the case that Aquirbad is one of the first to recognize these processes and integrate them into very simple ways with AI and, and continue to work on it. I know a bit about what is still planned. Um, and that is already, um, quite a lot and extensive, and certainly something where you say, um, that, um, that will make a difference. Yes, exactly. And as the last tool of what we, uh, what we are looking at today, we naturally also have to, uh, also take time for it. Um, so if you don't know it yet, it is definitely worth, uh, to have an interview with us, to have the whole thing, um, shown to you for the individual use case. Um, yes, this is now the slight little advertising block of the whole thing, of course, but, um, I'll tell you honestly, I wouldn't work at Side if I didn't find this principle, this model, and this tool, um, so incredibly brilliant. Um, it has a strong intrinsic motivation, um, for me to be here. Um, Site, that is our website, offers, in principle, um, all data and potentials for an object, for a plot of land, um, at the push of a button, AI-based, um, at the push of a button and in real-time. That means, essentially, you only need to provide the address of an object, the address of a plot of land, and with this information, you receive all the data that is relevant for you in the first mile check, in project development, or also in the check as a broker. That means, first, you get the energy requirements, the energy efficiency classes for all residential buildings in Germany. Second, let's click through here, yes, that's another part. Second, you get all photovoltaic potentials for all existing buildings in Germany. You have the option to do a profitability calculation directly for each plot of land. We work very closely with Price. That means, rent and purchase prices from Priceable for all residential buildings in Germany are integrated. And with this residual value calculation, you then have the option to, um, directly assess whether the, um, plot of land or building is relevant for you or not. Uh, in three different scenarios, we consider existing, densification, and renovation and new construction. And, um, what is also included is renovation plans for all residential buildings. We need, or you need, only the address to build a renovation plan, um, for this object and to know exactly, okay, um, if I buy it now, for example, in the next 2 years, I will have to invest 250,000 € in energy-efficient renovation. Then the price is a bit too high for that. Then you calculate it with the profitability calculator and see, yes, I could offer 50,000, 100,000 € less, 100,000 € less, and it will pay off for me. Therefore, I would offer it now. Yes, so it's very fast. What we originally came from, um, and what, I think, caused the biggest impact 3 or 4 years ago, was the area of building potential. You get the corresponding building potential for every plot of land in Germany over time. Why? Because the AI is behind it, and because we can also go directly into the tool, I am on the website. So, let's just do that, and let's leave this half-theoretical part aside. Um, as I said, I can enter any address in Germany that I want. We already have all the data for all, um, buildings. What I always like to use. The building, um, is this one here, Fürsten 188 in Düsseldorf. Um, what we see here are 3D representations of the building. Um, these are so-called lidar data. We can also display the whole thing in so-called LOD models. And what site does differently than, um, actually all others, I believe, is that we use these LOD models, which come from the land surveying offices, they are accurate to about 10-15 cm in height, width, depth, and, um, but also use cadastral data simultaneously. They are accurate to the centimeter, um, for building outlines and plot outlines. And now we overlay these LOD models, which are accurate to 10-15 cm, onto the cadastral data. And this results in a new 3D model, and specifically a 3D model where we know the length and width exactly of all, um, walls and can display it precisely. And in height, we know the building with an accuracy of 10-15 cm. And honestly, that's still relatively little AI that lies behind it. The AI comes into play when it comes to building potential, because, um, through this precise data that we have for each object, we can say for each object exactly how much GFA it has, how high it is, what GRZ and GFZ we have here? Uh, we know the type of use of the building, and from that, um, because we know this for all buildings here in Germany and also here in the vicinity, um, we can then infer what we have in the vicinity and how, on this basis, we can presumably build the plot of land here that we have. What is the maximum possible buildability? In this case, it would be that we, um, yes, could not increase the building height further, but could significantly increase the GRZ. That means from 0.41, we also know 0.79, i.e., almost 400 m² of ground area, and that would result in a WGF of about 2250 square meters, and we know that for every plot of land in Germany. We can now, when you looked at Site two years ago, we didn't know that yet. We can now, for example, take this plot of land here, it's residential land, it's a relatively new building, so it doesn't make much sense to tear it down. But let's do it anyway. We can take the building and say, sorry, and say, yes, but the breakfast next door and the one next door also belong to it. And now I would like to know, please, how can I build this entire plot? If it's residential, about 14,000 m² GFA, if I wanted to change its use for some reason to industrial and commercial space, which we have less of in the vicinity, um, then the buildability would be significantly lower. Um, now with the height indication of the maximum possible height. But I also have the option to simply say, come on, I'll build now, I'll tear down the existing buildings here. I've just chosen a plot of land with relatively many existing buildings. For reasons of time, let's not take that, but let's take the one we had just now. I'll tear it all down. So. And so. Let the AI recalculate, because now we only have one breakfast here, the calculation won't be much different than what we had before. But I think we could have added a floor before. Um, let's check again, but, um, that's the consideration of the plot as completely empty, as completely free. Uh, and then I also have the option to think about how I would build it. And, for example, this simple bar closure would look like this for me. Um, then I pull the whole thing up to three, uh, uh, in 3D to about 20 m building height. Um, let me show the surroundings. So. And, uh, see then, okay, this is how it would fit in well. Grundfläche 244 m², with that I could generate a WGF of 1200 square meters here. Um, it's still not enough for me, because there's still some potential, and then I would simply continue to work on the plot. You receive a plot of land, an existing property, and with it you can very quickly determine at the push of a button what is possible on this plot of land. Should I actually look at it further or not? Therefore, the question of, um, yes, the initial check of the object, is a matter of 5 to 10 minutes to even assess whether I look at it further or not. And the same applies to existing properties, of course. Um, I already mentioned the renovation planner, you can see it here now too. So I can now simply say, please Wohnus 1, Wohners 2, they belong together now. And for this, I want to build a renovation plan. If I know a bit more about the building, for example, that the, uh, that the window proportion is medium to high, I can enter it here. We have an AI behind it that gives us the probability of the heating type. You can also display this. Um, that means these values are, um, very, very good in the right direction. Um, if there are a few small adjustments, you say, the usable area is actually 5 square meters larger, or we have perhaps only ten rental units, adjust it, yes, adjust it very quickly. Um, click on next, and then you can immediately build a renovation plan, and specifically through this measure slider here, that you say, what is now the most efficient, the most effective thing I should do? Well, here it makes the most sense to install a heat pump, and put the PV system on top. Then I'm at 130,000 €. Um, just for this energy-efficient renovation, but with it, I have achieved energy efficiency of A. That is one option. Alternatively, you can also simply remove all measures and say, yes, due to, uh, monument protection of the facade, for example, I cannot renovate the facade at all. Uh, heat pump will also be difficult. Therefore, I only want to know, if I replace the windows, insulate the attic, insulate the basement ceiling, where do I get to? You get to D. Um, with an annual saving of ancillary costs of 5,000 € and a payback period of over 30 years, at 350,000 €. Um, the whole thing, so you can adjust much more. Prices, costs, you can adjust. You can, um, you can adjust the insulation standard. I'll show you that again here, for example, to KW40, and then it calculates accordingly. Uh, it's very simple, very accessible, very fast. Um, and you always only need the address, and that's the core of Site. And by the way, as the penultimate point, very quickly. Um, all the data we have is also searchable backwards. That means, if you now say, I'm looking for, um, possibilities to add floors for residential land in, um, Düsseldorf, then you will find these objects in a very short time. We want at least an addition of, um, 100 m², so that it pays off somewhere, and, um, so, and only for buildings, um, where we can add at least one full floor, maximum three, and where we can really do it safely, and these buildings should be from these construction years. Sorry, I know this is going very fast now. The GRZ doesn't matter to us, but the buildings should have at least a ground area of 200 m² and in the existing stock already at least 600 BGF. Um, and a poor energy efficiency class, for example. Then we have 90 matching objects in Düsseldorf, which are now displayed, and let's go to, for example, here and see now, yes, this is a building in the backyard, but it is 21.3 m high, the one next door is 24.5 m high. That means the AI correctly indicates here that 2.60 m would be possible, and it would also be a full floor. That means here one would probably consider whether to, um, the attic, um, yes, rebuild somewhat. Of course, we already have apartments in there, so it will be a bit more difficult in this case, but, um, yes, the AI has done what it should, and given us exactly these objects. In the last minute, very briefly, what Site also does, we create complete architect analyses. This is no longer AI. This, um, is completely done manually by our architects, architect analyses, potential analyses, feasibility studies, whatever you want to call it. Um, we can deliver, um, architect services within a few working days, um, which otherwise might take two weeks or cost 10,000-15,000 €. We are significantly cheaper. You may have just seen the 1900 € that it costs now. We manually check the entire building law of a plot of land. Um, we check what repurposing is possible, what other asset classes we might be able to build on here, and then provide you with a concept with mass models. It's exactly the same as what an architect does. Um, we are perhaps sometimes a bit more efficient, I would say. Yes, I'll gladly send you this as well, if you want. Um, this, this, uh, sample analysis, then you can take a look at the whole thing. So, and at 12:00, I still have to give you a reminder about the next webinar, where we will not primarily look at Site, or not primarily Site, but rather, um, other automation possibilities for our, um, for our everyday life. This can be ChatGPT, it can be with Google Gemini, it can also be that we really build workflows, um, that, um, really represent automation. And, um, this here, the next webinar is an introduction to this automation. That means, um, we look at what is possible and what we can build as easily as possible with a practical example. December 4th, it's time. At 11 am, and this is the QR code. If you scan it, um, you will immediately go to the registration page. If I have done everything correctly, I hope so. Uh, then you will immediately go to the registration page and can, um, register for the webinar. Also, of course, free of charge. Um, yes, and if you then say, is now the next slide, if you then say, um, man, with Mr. Neuenkamper, with colleagues, um, with us, of course, I absolutely want to speak to Site, then you can secure my contact details by scanning this, um, entering your email address. I will receive a notification about it, but you will not enter an automatic funnel or flow, um, but you will simply receive an email with my contact details, and you can then feel free to contact me. Um, what I would recommend is to generally follow on LinkedIn, on LinkedIn, perhaps also follow me. Every now and then, um, we show there, yes, what is possible with AI. Um, I try to regularly publish articles with a very practical use case, um, that can be done with other AI or with Site. Um, and, uh, yes, feel free to check in there from time to time. Um, there are definitely great, good things there about what can be done with AI. So, um, yes, 12:02 pm. Uh, we have reached the end of this webinar. As I said, feel free to send criticism, discussion, wishes, dreams, ideas for the next webinars. Um, send them to me by email, send them to me on LinkedIn, send them to us by email, very gladly, and, um, then I look forward to welcoming you all again next time. You will receive an email with the presentation, you will receive an email with the recording as soon as it is available, and, um, yes, until then, I would say, thank you for your participation at this point. So, bye.