Drooms on how artificial intelligence can improve due diligence

Machine learning can open the door to smoother real estate debt dealmaking, argues Rosanna Woods, UK managing director at Drooms.

This article is sponsored by Drooms

Artificial intelligence is increasingly being used in the real estate industry to improve repetitive but fundamental tasks, minimise problems and reduce costs. Drooms, a provider of digital data room solutions for real estate assets, is among AI’s proponents in the industry.

Drooms’ platform is designed to enable the digitisation and secure the exchange of relevant documents during the due diligence process, offering customised solutions for the entire lifecycle management of assets needed for real estate transactions and portfolio management. The company’s UK managing director Rosanna Woods tells Real Estate Capital Europe how artificial intelligence solutions are beginning to be used to further increase speed and accuracy, and to reduce cost, when exchanging the information required to carry out due diligence.

How has covid shaped the real estate industry’s approach to digitisation?

Rosanna Woods
Rosanna Woods

During the pandemic, we have seen an acceleration in the adoption of technologies. However, real estate is still far behind many other industries. In March, we conducted a survey of our clients to understand the barriers to change among European institutional investors.

Two main points stood out: 48 percent of respondents cited lack of knowledge of existing technologies, and 45 percent felt they did not understand the true benefits. However,
87 percent of respondents said that in the next two years they would be increasing their technology budgets. That shows our clients are seeing value in digital transformation and really beginning to understand the benefit.

Real estate has always been a very physical, tangible industry. But the ability to continue business as usual online became very important during the pandemic. In the past, many businesses did not have the budgets in place, or the understanding of workflow processes, to make that huge change. But covid was a unique situation which brought forward a transformation that we might have expected to see in five years’ time.

We saw increased uptake of our services, not only for due diligence, but also just to share confidential documents. Clients were looking for new means of sharing confidential data, delegating tasks, and putting workflows in place over an internet connection.

Why does the real estate industry need to adopt AI?

The benefits of AI are related to the vast amount of unstructured data real estate assets generate. In the asset management process, there are a vast number of data points generated by every tool used in workflows. Managers need to find ways of getting to the relevant information among those data points. In any due diligence process it is increasingly important that they do not just access data; they access information and do so as quickly as possible.

AI can help lead them to the right information, not just at the due diligence level, but across the whole real estate asset management process. Every real estate business has that same pain point: how to optimise repetitive tasks, minimising risk and cutting cost. The more elements of human interaction in the process, the higher the risk of mistakes being made, or not having all of the information at hand when you need it. Meanwhile, real estate businesses can also cut costs because the right AI technology is more efficient and improves productivity.

Which AI features can improve the real estate due diligence process?

The data room is standard practice for due diligence and has been for a while. The next revolution is making data rooms smart. There are three key AI features that we have adopted on our platform: auto-allocation, auto-naming and document translation.

Categorising huge amounts of unstructured data slows down the process and creates cost. No-one wins from having that task, and it is often left to junior staff. Where you have humans involved, you have the risk that they will make mistakes. Auto-allocation optimises data structuring. The user can dump unstructured data into the platform, and the AI will read the content and classify where it belongs in the index.

Another feature that builds on auto-allocation is auto-naming. Anyone who looks at their database will most likely see files classified under different naming conventions, or which are not named properly at all. With auto-naming, when the document is uploaded it is not only classified through auto-allocation; the AI also suggests what the document should be called. That convention operates as a template in the background, automatically producing a structured data room with a consistent naming convention as well.

The last feature is the document translation. When carrying out crossborder transactions, physical data rooms would have contained documents in several different languages. The translation AI function allows instant translation from one language to another.

It supports multiple languages, and the document retains its original format. It is not a legal translation, but it is very helpful when working on cross-border deals because different national teams can access information in their own language. It is trained with business and legal language, so the accuracy of the translation is excellent. It also removes the cost of having individual documents translated and is securer to do translations directly in the data room.

How can users be sure that the AI is reliable?

To achieve a high level of quality and accuracy the AI needs hundreds of thousands of training data points so that it can increase its reliability over time. A human checking element is still required. The platform auto-allocates or auto-names documents, then consults a human to find out if it has done the right thing. That is how the system learns, while at the same time building trust with the human user.

We train our platform with data that is available to the market, but then we also train specifically for our clients. That ‘self-training’ allows the person working with the system to train it to structure data in a way that suits how they like to work. As a provider we have to be very careful to meet the stringent requirements of the European General Data Protection Regulation, while ensuring that we are allowed to train our platform with this data.

How can the technology be applied in real estate lending without compromising security and confidentiality?

The use case of AI is very similar to the equity market. The process is always about sharing confidential documents as quickly and securely as possible. In debt, it is about giving access to data so that it can be reviewed by the different parties and further action can be taken. For example, when restructuring the finance for a real estate platform, a variety of different parties need access to the data: the investment banks, the lawyers, the advisers. It is not the typical due diligence process seen in a sale transaction, but ultimately it is also due diligence.

“I do not believe the due diligence process will ever be fully automated… The human element of judgement will re­main very important”

A digital solution can maintain confidentiality and security while giving everyone the correct access rights and allowing them to perform their tasks, and at the same time creating an audit trail on the actions that were taken.

To guarantee the same level of security when adopting AI, providers need to develop their own in-house AI solutions, so that the data does not leave the platform. Drooms has its own team of AI developers, who build everything users see on the platform. Other platforms sometimes have a connection to a third-party AI provider.

Can AI and technology help real estate firms tackle the ESG aspects of deals?

Our survey showed sustainability was ranked the top trend in European real estate. There is increasing pressure on businesses to put the right ESG processes in place, and also to demonstrate their credentials on a number of levels. That is why it is important for them to adopt a digital strategy for ESG measurement and reporting, gathering a large amount of additional data, and using AI to access that information in real time when it is needed.

If all the necessary data is online and structured, investors can check swiftly and accurately for ESG compliance.

For example, in the case of a green loan, a digital platform could be used to collect the information that proves that the loan is being made against an environmentally friendly asset, with all the certifications in place, and then give access to that data so that it can be checked. The lender can therefore be confident that everything is in place to grant the loan. Also, on an everyday level, using a digital platform is more eco-friendly because it saves paper. If all your documents are accessible online, you don’t have to travel to carry out administrative tasks.

What does the future hold for AI in this area?

We are still at the beginning of this transformation of the industry. The technology still has to be trained further. We would like to reach the point where we can say all the outputs of AI are accurate so the end user would not have to check them anymore.

I do not believe the due diligence process will ever be fully automated, though. The human element of judgement will remain very important. But we have already seen great improvements, and there is still a desire to go further. In the future there will be much greater use of data analysis to access crucial information to further speed up the process and ensure that market participants are ready to take advantage when they have only a short window of opportunity in which to secure a deal.