DAVIS raises €4.7 million to automate the upstream phase of real estate development

For decades, the digital transformation of real estate has focused on asset management, marketing, and 3D visualization. Yet the most economically critical phase — the one that takes place before the very first shovel hits the ground, has remained largely artisanal. Feasibility studies, regulatory constraints, volumetric analysis, financial trade-offs, architectural design: the upstream phase of real estate development still operates as a fragmented chain heavily dependent on successive human interventions.

It is precisely this phase that Davis, a Franco-American startup founded only a few months ago by MEHDI RAIS and AMINE CHRAIBI, is now attempting to industrialize. Davis has announced a €4.7 million pre-seed funding round led by Heartcore Capital and Balderton Capital, with participation from Evantic, Yellow VC, and Entrepreneurs First. Behind this transaction lies a broader ambition than architectural automation alone: drastically reducing real estate development timelines and transforming a historically sequential industry into an infrastructure driven by generative models.

Davis claims it can reduce certain upstream phases from several months to just a few days through a combination of artificial intelligence and human expertise. The company centralizes regulatory, technical, and economic data to generate directly actionable feasibility studies and architectural concepts, which are then reviewed, adjusted, and validated by architects before delivery.

With Gaudi-1, the company says it has developed an approach distinct from traditional diffusion models used in image generation. Instead of operating in pixel space, the model reportedly functions within a spatial environment, directly manipulating architectural components (rooms, walls, circulation flows, and layouts) to produce configurations compatible with the regulatory and economic constraints of real-world projects.

This specialization of models is likely one of the defining trends of today’s AI ecosystem. After a phase dominated by general-purpose models, many startups are now developing architectures trained on highly specific industry constraints. In real estate, AI cannot simply generate a convincing image; it must incorporate zoning regulations, technical requirements, financial ratios, local standards, and highly practical operational considerations.

The company says it is already working with several leading property developers and plans to handle several hundred projects by the end of 2026.