Design is a multidisciplinary task where a solution must be found that fulfills several competing requirements. Artificial intelligence (AI) supported prediction and simulation help both designers and decision makers. The use of AI streamlines design work and enables new services designers can deliver to the entire built environment value chain.
Ramboll – machine learning supporting design of sustainable societies
Ramboll is an international design and consulting company. The company employs 16,000 experts globally and 2,500 in Finland.
AI development
AI is already used quite extensively at Ramboll. Several experts in different countries develop AI solutions, usually in small teams as part of a business unit. AI is also used in various projects and tasks in different functions of the organization.
The organization is decentralized, which complicates developing uniform practices and standards. Nevertheless, a purely centralized organizational model might not work due to differences in customer segments and local needs in different countries. For this reason, Ramboll deploys a hybrid model using local and centralized functions in the development work.
Aha!
Did you notice how the story of Ramboll reflects what we discussed in Chapter 1? There is no one right way to organize for AI projects – instead, a constant balancing act is needed to find the most appropriate model.
Local businesses are responsible for the ideation, design, and prioritization of AI solutions according to customer needs. Thus, development work takes place from the bottom up. The most successful ideas can enter the Ramboll Innovation Accelerator, which provides additional resources to promote selected projects.
Ramboll uses partners to design and implement various solutions, although most tasks are handled in-house. The company strives to follow the same policies and processes regardless of who implements the solutions so that they can also be used more widely in the organization.
Business AI solutions
Ramboll already has several AI-based solutions in daily use. For example, AI supports functions such as financial and human resource management. However, the most important applications are new services and solutions, which provide the greatest benefit to Ramboll's customers.
The company's AI solutions create new opportunities primarily in two areas:
Developing and streamlining processes
Offering completely new services to customers. With these services, data is collected in new ways and from various sources, with information flowing from maintenance back to design.
The impact on business
Using AI challenges existing business models, meaning traditional ways of working may be permanently changed to something new. This highlights the importance of interaction, as we learned in Chapter 1.
Ramboll’s business model has traditionally been based on selling expertise, often on an hourly basis. Different AI solutions can challenge this model by removing the need for work steps previously done manually. In this situation, time is freed up for higher value-added work, such as tasks that require creativity, strategic thinking and interaction with the customer.
The following example illustrates how previously laborious and inaccurate modeling of traffic flows can now be automated and done at a very precise level using AI.
Example application – simulating traffic usage with Brutus
The Brutus solution simulates the movement of people in different vehicles from one place to another. For example, the simulation model helps to assess how the construction of a new bridge at a specific location affects the use of buses and the amount of private car transport. The solution can work at either a city or a regional level.
Ramboll acquired Brutus as part of an acquisition, and the app is widely used to support traffic planning. The solution is a cloud service licensed by customers on a software as a service (SaaS) model.
Advanced simulation models require huge amounts of data. Access to information often proves to be a bottleneck in development projects, as has been shown in the case of Brutus. Ensuring the availability of data is labor intensive and special attention must be paid to the protection of individuals in the movement data. Simulations also require significant computing power.
It is possible that Generative AI can augment the capabilities of Brutus e.g. automating the creation of simulation reports by using a GenAI agent for discussing and fine-tuning requirements from the customer.
Lessons learned so far
In simulation models, uncertainty is an issue. Users may highlight individual cases where the prediction model has clearly failed and try to use these to prove the solution doesn't work. This can happen even if the accuracy of the predictions produced by the solution is high on most of the cases.
Managing expectations in AI projects is important. If users’ expectations for an AI solution are too high, they will easily be disappointed. When using a forecast model, the message should be that it's only part of a bigger process. The solution provides information that supports the work of the experts, but does not replace their own judgment.
Brutus – Brutally detailed transport modelling
Read Ramboll's presentation about Brutus.
Future developments
Above all, Brutus helps to plan the allocation of investments. A sustainable development perspective is also linked to the solution. For example, the construction of a cycle path has environmental impacts that can't be assessed by economic indicators alone. Ramboll is continually developing Brutus and related services.
Sign up to solve exercises
Granlund – energy optimization of real estate
Granlund Oy is a group of experts in the construction and real estate industry. It employs more than 1,000 professionals in 27 locations in Finland.
Example application – AI Energy Mapping
In its data, analytics, and AI operations, Granlund focuses on energy and responsibility – both of which are strongly linked to the company's core strategy. Granlund’s focus is especially on new services and products, so the use of AI is intended to generate new business.
The internally developed AI Energy Mapping service is Granlund's first productized and commercial AI-based solution. AI Energy Mapping enables the identification of energy saving potential from large properties by offering concrete measures. The starting point is a machine-readable energy certificate for an individual building.
These are the different stages of the AI Energy Mapping service. The starting point for the service is a machine-readable energy certificate for the building, which is enriched with the company's own databases, such as the cost database, the CO2 database, and real estate data (“simulation” and “optimization”). Based on the data, the system is able to produce construction-specific action proposals (“outcome”).
AI service development path
Granlund built the service in six months from idea to actual use. The tool was initially intended for internal use, but during the project it became apparent the solution had clear business potential. This insight arose primarily because an employee who worked with customers was involved in the development work from the beginning.
Aha!
In AI projects, it’s important that the end user of the solution or service is involved in the development work from the very beginning. This greatly increases the chances of the project succeeding.
The development was the responsibility of a small team within the company for two reasons. First, the implementation of the solution required an understanding of the customer and the field. Second, the application was built on top of Granlund’s existing Energy Simulation tool.
The development team ensured the functionality of the energy mapping by building a minimum viable product in a few months, the results of which were compared with previous, manually performed analyses.
Scalable professional services
AI Energy Mapping makes professional services scalable. The new tool can be used to analyze and simulate large real estate properties as a whole rather than just individual buildings.
The simulation tool included in the service uses several databases. The energy certificate mentioned earlier will be enriched with, among other things, a cost database, a CO2 database, and real estate data. All databases belong to Granlund and are self-maintained.
The cost database contains information on the costs of various energy saving measures, such as the installation of a heat pump of a certain size or changes to window and insulation types. Similarly, the CO2 databank reports the emissions of potential investment measures and suggests the best option.
Property data includes information on the type of building, its purpose, area, volume, location, and the age of technical systems. Based on the data, the system is able to produce building-specific proposals for measures like the possible renewal of structures, the installation of heat pumps, or the optimal dimensions of solar panels. Everything is aimed at minimizing the building's lifecycle costs, ultimately leading to savings.
The service is aimed especially at property owners who have a significant number of residential and office buildings. In the future, Granlund intends to develop the service to include other properties such as hospitals, shopping malls, and industrial properties. This will require further data and additional resources.
A successful AI project often leads to new ideas, and this happened at Granlund. Project seeds are created at an accelerating pace and are presented in-house during monthly data and AI webinars.
Granlund intends to further develop the AI Energy Mapping tool and to replicate the concept for the company's other services.
It is probable that Granlund may in future investigate the possibility of using GenAI to further scale their professional services e.g. a GenAI based agent can orchestrate the entire knowledge based business process from energy mapping, question-answering, recommendation and optimizations, creating renovation proposals to making sales orders.
