Managing documents with the help of AI: docunite ensures greater efficiency with Azure Cognitive Services

The challenge: The sheer number of files, systems, and stakeholders makes managing documents difficult

A property’s lifecycle is made up of a series of different stages: planning, construction, sale, management, and then at a certain point either renovation or demolition. Over that time, all kinds of documents are created: surveys and leases, completion and condition certificates, ground plans and land records. In the case of large properties, there can be up to 500,000 documents per building. This poses an array of challenges, particularly to large real estate companies, which might own or manage hundreds or even thousands of properties. Patrick Penn, CEO and Founder of docunite, knows this only too well: “In Germany, real estate companies are often regulated by BaFin, the German Federal Financial Supervisory Authority. They must always be in a position to provide information about their portfolio, generate assessment reports at the touch of a button, and prepare sales extremely quickly.”

In practice, however, this is often hard to achieve due to the rather fragmented nature of the industry. To coordinate renovation projects or find new tenants, real estate companies work with external stakeholders, who require access to those documents relevant to their role. Files are then exchanged via a range of systems like email or SharePoint. Even within the real estate company itself, this process involves various departments such as Asset Management or Legal and Finance. To make matters even more complicated, each of these stakeholders has their own requirements for the documents, which must also be provided in the appropriate version. “This results in elaborate file storage systems with up to 800 folders per property, of which 70 percent are empty and 10 percent are no longer accessible through the online structure,” Penn says. He knows all about the dilemma companies are facing. “The number of duplicate files is skyrocketing. On average, some 60 percent of documents exist in more than one place. At a certain point, this becomes neither scalable nor manageable.”

In the past, many real estate companies relied on generic digital document management systems (DMSs), adapting them to their specific requirements and spending a lot of money to keep them up-to-date. “That meant each real estate company was creating its own solution, even though they all had the same requirements. There simply wasn’t a standardized DMS that the entire industry could use,” Penn says—and that was the idea behind founding his company: “In 2017, we launched docunite to provide real estate companies with just such a standardized DMS.” From day one, the new company chose Microsoft Azure as the technology foundation from which to continuously grow and scale its operations. “Since our customers’ data stocks grow really quickly, a high degree of scalability is essential. For us, Azure was the solution that is flexible, invoiced according to demand, offers high system security for our sensitive customer data, and with which we were able to start directly after our foundation without any great investment,” Penn says. “That's why this technology was clear to us right from the beginning.”

The solution: Automated management and classification of documents with Azure Cognitive Services

Today, the cloud platform Azure provides docunite customers with automated and standardized document management in real time: when documents are uploaded, the system is capable of recognizing around 70 different file formats and converts all these to PDF. To ensure efficiency and scalability, the files are processed in containers using Azure Kubernetes Service and Azure Virtual Machines. At the same time, Azure Cognitive Services extract blocks of text from the files. “Our Machine Learning model uses certain key terms to identify whether a document is, say, a lease or a survey, and suggests the appropriate document classification,” Penn explains. “Once a human has checked and approved this, the document is automatically classified, integrated into the structure, and made available online to all participants in their chosen version at precisely the right point in the process—even in multiple languages.” What’s more, the solution’s integration into Power BI and Microsoft 365 opens up simultaneous collaboration and straightforward visualization in reporting. Microsoft Entra ID and Microsoft Defender for Cloud keep the solution secure. “Thanks to Azure and Azure Cognitive Services, we’ve established a foundation and a solid data stock that will allow us to reap the vast benefits of AI in the future,” Penn summarizes.

Penn and his team have also launched a pilot project to roll out Azure OpenAI Service at docunite: “In the future, it should be possible to pose questions relating to a specific property and receive a fully formulated response. Customers will be able to see exactly what document content the response is based on.” For instance, when it comes to selling a property, the system could automatically generate suitable FAQs. “So instead of having to manually trawl through every document looking for the answers to investors’ questions, employees can have the system automatically provide the answers to a ready-made list of standard questions.” Through this pilot project, Penn has been able to show how to make the whole process much more efficient: “We’re immediately seeing where our documents and data might be incomplete, and how we can shorten the transaction process from between six and twelve months to just a few weeks.”

What’s more, the flexibility, scalability, and compatibility of Azure offer docunite and its customers the option of being able to tweak structures and processes—at any time and without much administrative effort—to reflect any sudden changes in regulations or to integrate new software solutions. In this way, Penn is safeguarding the future viability and competitiveness not only of docunite but also of its customers. “Thanks to the scalability of our solution, we can put even small real estate companies in a position where they can offer the same level of performance and quality as larger companies. No matter how many properties they have in their portfolio, they will always have a transparent and well-ordered digital environment.”

And that’s exactly what Penn plans to further enhance in the future because the volume of data and the number of companies docunite supports are growing all the time. “At the moment, we’re exploring other potential applications of Azure OpenAI Service and ChatGPT. We’re also working to ensure that those docunite customers who collaborate with one another on an interdepartmental level can draw on the same data stock—naturally with appropriate access safeguards. Azure will help us there, too."