Sweden’s forest industry plays a key role on the path to a circular bioeconomy, but of the annually harvested forest only 20% are transformed into products (sawn timber) with a long lifetime and a high rank in the material cascade. A main reason is the lack of integration in the information flow in the sawmill process, which is due to the unclear properties of the felled logs when they arrive at the sawmill. With the help of computed tomography (CT) it is possible to see the internal features of a log and optimise the process based of this data. We propose to harness modern machine-learning methods and wood-material science to create a virtual sawmill process, with CT-scanned sawlogs as the input data. This makes it possible to optimise which products will be produced, which quality they will have, and to which customer they will be sold – before making the first cut in the log. The project team consists of wood scientists, machine-learning researchers, and industry representatives.