Finch is generative AI for building design. Architects sketch constraints, and Finch produces optimized, rule-checked floor plans in real time, turning early-stage design from a slow manual craft into an interactive exploration.

Built by architects in Sweden, Finch sits in the feasibility and concept phases of a project, where the big value decisions happen. Draw a building footprint or adjust a site boundary and Finch instantly regenerates internal layouts, distributing units, cores, and corridors while respecting the rules you have configured. Change one parameter, the number of floors, a setback, a minimum apartment size, and every dependent metric updates live: unit mix, gross and net areas, efficiency ratios, daylight-relevant dimensions.

An immune system of rule checks flags code and playbook violations as you design rather than weeks later in review. Finch connects with the tools firms already use, including Revit, Rhino, and Grasshopper, so generated options flow into detailed design instead of dying in a silo.

Who is it for?

Architecture firms doing residential and mixed-use feasibility studies, and developers who want more options tested before committing capital. It is a professional tool; hobbyists designing a single house are not the target.

How much does Finch cost?

Custom, seat-based pricing for professional teams, typically arranged through a demo. Expect professional-software budgets rather than consumer subscriptions.

Our verdict

Finch is one of the most convincing applications of generative AI in architecture: it augments early design judgment instead of pretending to replace architects. If your firm bills hours on feasibility studies, it pays for itself quickly.