go to homepage

Collaboration for a smarter future in real estate

DNB Eiendom and DNB, together with Itera, have explored the potential business benefits of using generative AI. The result was the development of the broker assistant named 'Emma', and now DNB Eiendom is in a better position to both experiment with and incorporate generative AI into the company.

About DNB Eiendom

DNB Eiendom is Norway's leading real estate brokerage company, offering sales of used and new homes, as well as leisure properties. They are Norway's largest broker for project management and new construction, with over 900 employees across over 100 offices. The company is owned by DNB ASA.

The use of generative AI (Gen AI) is among the most significant technology trends we have seen since the internet, and it is already beginning to disrupt many industries.

The project's main purpose was to provide practical learning and experience with Gen AI for DNB Eiendom and DNB, as well as develop a prototype that could solve a specific challenge for the company.

A GPT-based chatbot was chosen to assist brokers with internal procedures and rules related to property viewings, post-processing, and sales. In an industry with high competition, the efficiency created by Gen AI is crucial to gain a competitive advantage.

 

Difficulty finding the right information in the quality handbook

DNB Eiendom is Norway's leading real estate brokerage company with over 500 brokers involved in pre-processing, viewings, post-processing, and sales of properties. This process involves many routines, regulations, procedures, and laws that must be followed.

DNB Eiendom uses an internal quality handbook consisting of industry-specific standards, regulations, and quality requirements to ensure good quality.

The content in the quality handbook is of high quality, but brokers may find it difficult to locate and interpret the information contained within it.

Itera conducted a quantitative survey among all brokers and several qualitative studies to identify brokers' challenges. It turned out that only one-third of brokers found it easy to find the right information in the quality handbook.

The survey also revealed that brokers often contacted responsible experts to discuss their challenges even though the answers were available in the quality handbook.

This placed a burden on these experts, who had to answer the same questions multiple times while not adhering to the quality handbook, which could have significant consequences for DNB Eiendom, such as fines or loss of trust from customers.

Gen AI is designed to ensure compliance with regulations.

In collaboration with Itera, DNB Eiendom wanted to explore the possibility of using Gen AI to ensure that the regulations described in their quality handbook are followed in the daily work of brokers.

The goal is to reduce the burden on responsible experts, allowing them to focus on other important tasks.

In partnership with Itera, it was decided to create a custom GPT-based chatbot that brokers can use to seek help regarding procedures and rules described in the quality handbook.

During a 4-week sprint, Itera developed a prototype that a selected group of employees tested to ensure the accuracy of the provided answers.

All processes and work conducted to achieve these results were documented for maximum learning throughout the organisation.

The solution is built on deep insights and expertise from a multidisciplinary team.

The prototype was developed by a multidisciplinary team consisting of business consulting, service development, and AI development to ensure clear goals, KPIs, and business priorities for the solution.

Itera focused on user-friendliness and accurate responses to encourage brokers to use the solution in their daily work.

Based on this foundation, Itera created a prototype for a Gen AI-driven chatbot that brokers could use to access information from the quality handbook.

The prototype underwent repeated testing in collaboration with representatives from DNB Eiendom to ensure that the answers provided were helpful and useful for brokers.

The domain knowledge of real estate brokerage held by DNB Eiendom was essential in creating the best possible prototype.

Meet Emma

Emma is tailored to help brokers find routines and processes to ensure quality in their work. With each answered question, Emma provides sources from the quality handbook so that brokers can be confident about the information's origin.

Emma's minimalist design makes it easy for brokers to explore the solution.

Emma is built using components from the Microsoft Azure cloud platform. Resources such as Azure Cognitive Search and Azure OpenAI have played a vital role in ensuring optimal functionality for Emma.

Although Emma is available as a web application, early prototypes for adapting it to other platforms, such as mobile and sidebar, have also been explored. Much of the development time has been spent experimenting with parameters like temperature, system messages, and scoring profiles to ensure that Emma can assist brokers effectively.

Feedback on Emma during early testing phases conducted by Itera with brokers, responsible experts, and risk representatives has been very positive. User tests indicate that Emma adheres to the quality handbook and does not fabricate information.

Feedback suggests that the solution will not require extensive training, and most employees will be able to use it after a brief introduction.
Nancy Twang

– Our close collaboration throughout the project period has made DNB Eiendom more confident that the solutions we develop within generative AI maintain the necessary quality requirements while not posing a risk of violating privacy regulations.

Nancy Twang

Head of Risk Management, Privacy and operational risk at DNB Eiendom

– Our close collaboration throughout the project period has made DNB Eiendom more confident that the solutions we develop within generative AI maintain the necessary quality requirements while not posing a risk of violating privacy regulations.

Nancy Twang, Head of Risk Management, Privacy and operational risk at DNB Eiendom

Through the project, DNB Eiendom and DNB have learned a lot about managing a project involving Gen AI. This rapidly changing field has required understanding how generative AI can provide business value.

Itera has written a comprehensive report describing everything from issues and results from qualitative and quantitative surveys to processes, methods, solutions, and outcomes. This report has been widely distributed to provide maximum learning benefits to relevant environments in DNB Eiendom and DNB.

 

The way forward for DNB Eiendom

After the end of the project, DNB Eiendom is in a better position to both experiment with and incorporate generative AI into the company.

The issue with the quality handbook is just one of several important use cases where DNB Eiendom sees Gen AI as a possible solution. Other use cases are related to streamlining the central tasks performed by brokers daily, including sales presentations.

The prototype developed by Itera is designed to be expandable to other use cases. Generative AI offers vast possibilities that DNB Eiendom wants to explore in order to become an even stronger player in the brokerage industry.

Let's talk about data and AI

Let's talk about data and AI
Vishal Kaura
Head of Data & AI

Vishal Kaura