This interview was the occasion to look at the impact of tech and data on real estate, an industry quite different from the others. Real estate has a strong written culture, with a need for contractual documents and a lack of standardization. To extract specific information from the mass of documents handled, it takes a lot of time and effort for asset managers to do so. And it’s where AI in real estate comes into play. To take over such a time-consuming task and make it easier for people (and the industry as a whole) to perform.
Discover the benefits of Artificial Intelligence as we sum up the main points discussed during this interview.
Starting with what is probably the main question. Is the industry already being disrupted by AI?
If you’re looking for a positive answer, you are in the wrong. Real estate and AI are still at the premises of their relationship. And Robin was prompt to remind us why. AI is not a magical tool and before turning to AI solutions, the industry first needs to work on its data. This data will in turn be used to train models which could then take over some human tasks to accomplish them faster and with larger datasets.
And it all starts with being able to extract this needed data and benefit from the full potential of AI. But this written culture we touch upon at the beginning makes the tasks harder, with scattered data and masses of documents to go through. This is where the industry stands today. At the early stages of data extraction, waiting to get that down to get to the next step of training models. Before being able to rely on AI to take better decisions.
And how do we extract this data? Well, not with AI. This is where we get to talk about Natural Language Processing, usually just known as NLP. And this interview was a good opportunity to dive a little bit into how it all works.
Let’s take an example. When you read through a document with specific formatting, numbered titles, and some specific information, you will know what kind of document it is. Because you know what a contract looks like. But an AI does not know that. At first at least. And so we use NLP and human supervision to train models, turning implicit into explicit. NLP is about teaching a machine how to read and understand written information. In our example, to automatically understand that a specific document is a contract and where the main information is located. At first, the machine will make mistakes but will keep on learning and getting better each time, continuously improving. And then, it will know which key data to look for.
Now that we talked about how it works, it is time to delve into the heart of the topic: benefits.
Robin and Andrew went over several benefits AI can bring to real estate, which we’ll now go over.
The first benefit is, according to Robin, that AI brings in the opportunity to make the market more liquid. Document classification is not a rewarding job but one that is necessary to collect and check all the documents needed in case of arbitration. Technology can do just that perfectly well. Stonal has actually developed an AI to collect, classify and sort through all the documents to create a ready-to-sale dataroom. And thus, reducing the time to market an asset. In France, it usually takes 3 weeks from the data collection to the arbitration. With Stonal, it gets to only 24 hours. That sure makes the ability to trade real estate a lot easier.
Another use of AI would be for the operational phase. We’re talking about maintenance and operation. It is one of the most promising uses of AI, once we are able to collect the data to train the models. Thanks to AI, asset managers will be able to gauge the amount of capital expenditure needed (CAPEX) to, for example, be aligned with current and future energy efficiency regulations. Or with the evolution of the rental market. A significant financial gain and a clear view of what is to come.
And last but not least, energy. Stonal already uses its data extraction capabilities to improve ESG reporting, as we are able to use the data extracted from documents and auto-fill ESG questions. The idea is to go beyond that and move towards a system that is no longer just declaratory. But with proper proof of each ESG-related action. And properly measure carbon and energy efficiency to head towards the net zero carbon targets.
So, AI definitely brings in a lot of benefits. But, as the interview comes to an end, a final topic was raised. What could make AI adoption slower? Data issue aside.
The first one will be managerial, with real estate professionals needing to integrate AI into their daily work. But once again, AI will help improve things as it will take on redundant and low-value tasks, letting asset managers focus on their job. And with the rise of models such as ChatGPT and their gain in popularity, we see people starting to use AI every day. It brought AI into our daily lives so why not use it professionally as well?
The second one could be either a price or privacy issue. For the former, we know that it will need to be competitive in terms of pricing. This is why we have developed our own tool based on open-source models, proving better results than any popular Large Language Models everyone’s talking about and able to handle millions of documents at a competitive price. For the latter, we have seen that these popular models, such as ChatGPT, store data in different continents or sold data to third-parties. Thankfully, ChatGPT is not the only tool available for the industry. Stonal has one that was specifically created and trained for real estate, answering the actual needs of the industry. While protecting your data. How good does that sound?