Florent Poux Launches 3D Scan-to-BIM Automation Course

Florent Poux Launches 3D Scan-to-BIM Automation Course

October 13, 2026 Rachel Kim – Technology Editor Technology

Florent Poux, founder of the 3D Geodata Academy, announced the release of a practical instruction guide focused on automating 3D Scan-to-BIM workflows. Poux details a system designed to reconstruct floors, ceilings, and walls directly from point cloud data and export the results into Industry Foundation Classes (.IFC) files.

The Tech TL;DR:

  • Florent Poux released a 3D Scan-to-BIM automation guide detailing automated wall, floor, and ceiling reconstruction from point clouds.
  • The framework centers on exporting processed spatial data directly into standardized .IFC files for building information modeling.
  • The author reports that the core methods rely on 15 years of R&D focused on 3D spatial artificial intelligence problems.

Fifteen Years of Spatial AI Research Culminates in Automated IFC Exports

Poux writes that the preceding three months marked the most exciting period of his career, during which long-standing spatial geometry problems finally yielded results. Having spent 15 years doing research and development on 3D spatial artificial intelligence—specifically targeting point cloud datasets—he indicates that his cumulative work has unified into a cohesive methodology. He refers to this approach as the Brain-to-Deploy method, designed to let human operators build alongside AI agents.

The core utility of the newly published material centers on removing friction from the building industry’s traditional Scan-to-BIM pipeline. By applying automated spatial reconstruction algorithms, the workflow extracts structural elements from classified office scans and automatically writes the corresponding .IFC wall data.

Open Knowledge Sharing for AI-Assisted Spatial Workflows

Rather than reserving the methodology for closed client teams, Poux states an intention to share the core science as open knowledge. He argues that the current pace of development, accelerated by AI agents operating alongside human coders, is like nothing seen before.

# Example conceptual pipeline for point cloud parsing and IFC generation
import open3d as o3d
import ifcopenshell

def process_point_cloud_to_ifc(file_path):
    pcd = o3d.io.read_point_cloud(file_path)
    # Spatial segmentation and planar extraction for walls
    print(f"Loaded {len(pcd.points)} points. Parsing structural planes...")
    # Generate IFC entities
    ifc_file = ifcopenshell.file()
    return ifc_file

The released course material serves as an instructional entry point—dubbed Scan-to-BIM Automation 101—aimed at documenting the repeatable steps required to turn raw spatial coordinates into interoperable building models.

More on this