Cartesien moves three-dimensional generation to solid bodies, surpassing polygonal approximation and integrating manufacturing pathways and building modeling

Three-dimensional generative design tools are moving toward surpassing the stage of approximate visual models and non-manufacturable polygonal meshes, a shift that Formas pursues with its new Cartesien tool for architecture and product design. The system, developed between Singapore and Tokyo, focuses on addressing the most prominent weakness in embodied AI applications, namely the generation of 3D shapes that rely on surface-polygon approximations rather than solid geometric bodies. The tool aims to enable the construction of precise, topologically clean models based on NURBS surfaces and clear edges that are measurable, editable, and ready for direct industrial production.

The tool’s technical architecture relies on the principle of maintaining full editability for every element within the model without needing to redraw the entire scene. Designers can input voice or text commands alongside an initial sketch, photograph, or scan, specifying fixed elements they wish to preserve precisely, such as a restaurant’s row of seats or interior void walls, while the model constructs and updates the remaining composition based on geometric relationships between parts. The tool also incorporates a structure compatible with building information modeling, linking metadata to spatial geometry and defining element names and explicit relationships, thereby providing an organized pathway for integration into engineering design software.

Surpassing the traditional constraints of costly desktop design software licensesIt forms a pillar in the distribution plan, as the system enables direct native export to the Reno and SketchUp formats, with plans to support open AutoCAD and IFC files for building information modeling. This allows engineering teams to produce precise solid bodies ready for measurement, cutting, 3D printing, and operation via CNC machines, with toolpaths and manufacturing tolerances calculated for structural panels, ranging from furniture components and residential space design to complex structural frameworks, stadiums, bridges, and large-scale urban planning.

For engineering consultancy offices and real-estate development firms in the Gulf, Egypt, and the Levant, which face compressed timelines on major infrastructure and urban-masterplan projects, this type of tool demands a radical reordering of workflows. Instead of spending hours redrawing visual AI outputs and manually converting them into solid bodies that meet construction requirements, these architectures allow the initial conceptual design phase to be shortened and reduce the cost of specialized CAD desktop licenses for early-stage teams. The approach also shifts engineers’ skill requirements in the region from routine hand drawing toward expertise in verification, adjusting engineering constraints, and integrating data within building-modeling environments.

The company opens experimental access to the Cartesien tool on 18 September 2026, prioritising the platform’s foundational studio network, a step that serves as a practical test of how well generative models can replace entrenched solid-modeling environments. The technical bet here does not stop at shape-generation speed; it hinges on the system’s accuracy in detecting construction-engineering conflicts and its ability to address actual manufacturing machines without errors, a decisive criterion that will separate conceptual tools from true engineering production systems.