Knowing the materials embedded in buildings well before demolition or major renovation opens the door to reuse, repair, self-sufficiency and more resource-efficient decision-making. At VTT, an AI-based tool currently under development enables this by making use of building owners’ own data.
At VTT, a new tool is being developed that has the potential to change how the material stock of the built environment is utilised. In the recently completed LiveCol project, an initial demo focusing on the use of building portfolio data demonstrated that property owners’ project banks contain a vast, partly unexploited data asset. By combining drawings, building geometry data and archival information, AI enabled the extraction of material-level information, which quickly awaked growing interest among real estate and construction sector stakeholders.
Traditional urban mining approaches aim to assess the material stock of the built environment by relying on external data sources of varying quality, such as registers or generic building typologies. Portfolio mining turns this logic around by utilising building owners’ own primary data, such as drawings, building models and archival project information stored in their project banks. This data is already available and sufficiently accurate for material reuse, renovation planning and maintenance needs. At the same time, it creates visible value for companies and cities through improved anticipation and resource efficiency. Understanding the material stock is not only a circular economy issue, but also a core element of supply security for both public and private actors.
Customer interest is growing and use cases are expanding
The first demo confirmed technical feasibility: AI was able to read 2D drawings, geometry-based structural models and archival data, and to respond to user queries in natural language. This laid the foundation for the current tool development phase, which aims at more advanced data extraction, including AI models that combine image and text-based materials.
Discussions with cities and companies have repeatedly highlighted the same message: material data is critical for circular economy solutions, building maintenance, renovation, and planning for self-sufficiency.
Towards the strongest material data platform on the market
Project preparation at VTT is being carried out jointly by several teams to ensure that expertise in construction, data and artificial intelligence is fully integrated. The goal is to develop a tool that is accurate, transparent and genuinely useful in the everyday operations of companies and cities. Preparatory work is progressing in three parallel tracks: defining the technical solution, designing pilot projects, and expanding use cases.
At the core of the technical solution is the development of a unified terminology and data validation logic, enabling reliable integration of material information from multiple sources. Pilots will ensure practical functionality and gather user feedback, which will later be used to further expand use cases to better meet the needs of both companies and cities.
A future growth path is already taking shape, and we invite industry actors to co-develop the solution with us. This collaboration will help build a broadly applicable and practically distinctive material data platform. Companies will have the opportunity to influence the tool’s content and benefit from the results of the development work at an early stage – join us now!
LiveCol was a Business Finland (NextGen EU) funded development project carried out by VTT and its partners, focusing on renewing collaboration and planning practices in the construction sector through real-time 3D data and digital tools. One of the project demos explored how data stored in building owners’ project banks could be utilised more effectively. In this context, the first Portfolio Mining demo was developed together with partners, enabling the compilation of material-level information from existing buildings. https://www.tuni.fi/fi/tutkimus/livecol
