ECOSYSTEM
From Research to Industry
sgdl positions itself as a new infrastructure layer for AI operating in the physical world — from generative AI and synthetic cities to robotic teleoperation, digital twins, and low-altitude economy. Precise, frugal, and economical.
BUSINESS MODEL
Multiple Paths to Market
R&D Projects
Proving the technology works. sgdl participates in research programs funded by third parties (CERN, European agencies, industrial consortiums).
Industrial Partnerships
Demonstrating real economic value. sgdl brings technology and mathematical expertise; partners bring use cases, data, and industrial environments. Led by Michel Bernard, this track focuses on industrial junction and acceptability — proving that sgdl integrates within existing CAD, BIM, and OpenUSD workflows.
Technology Licenses
Exploiting and integrating sgdl proprietary technology (patents, algorithms, exolanguage) into products.
Platform (PaaS)
Making the Exolanguage Ariadne — a non-anthropomorphic language for spatial reasoning — a standard for complex structural problems. Becoming the mathematical infrastructure for Physical AI across 3D simulations, digital twins, and AI-driven workflows.
PARTNERSHIPS
World-Class Collaborations
Developed the ISS Canadarm2/Dextre on-orbit simulator for the Canadian Space Agency — pixel-exact rendering and real-time collision detection for the complete Mobile Servicing System, running on a single laptop with the entire ISS model under 2 MB. Validated in peer-reviewed IAC publication (2004) and used for astronaut training on Expeditions 7–9.
4 active projects with the world's leading particle physics laboratory:
- Spatial intelligence for volumetric brain imaging
- Structural anomaly detection for predictive maintenance
- European consortium with Mistral for AI in scientific infrastructures
- Reference validation
Real-time alignment between 3D preoperative and 2D intraoperative imaging. Precise, non-irradiating, with no calibration drift and no metallic artifacts.
Spinal Geodesy — from static snapshots to living coordinates
Current spine surgery navigation matches a preoperative CT to intraoperative fluoroscopy at a single moment — a fragile approach that assumes rigid anatomy, requires repeated radiation, and propagates errors into robotic screw placement.
SGDL's spinal geodesy uses the full spine as a biological reference axis, progressively replacing fluoroscopy with ultrasound and two body-mounted sensors that continuously track the patient in 4D. Registration becomes a living, self-correcting coordinate system that guides the robot with millimeter-level confidence throughout the entire procedure.
Since January 2026, sgdl is a member of the Alliance for OpenUSD (AOUSD), officially announced alongside Qualcomm, Booz Allen Hamilton, and other industry leaders following the landmark ratification of Core Specification 1.0. sgdl contributes expertise in non-anthropomorphic exolanguages for spatial reasoning, multidimensional data indexing, and geolocation.
"We view OpenUSD as the ideal framework to enhance interoperability in 3D simulations and AI-driven workflows. Our objective is to promote and advance open standards for complex geospatial and robotic simulation, fostering innovation across industries."
— Jérémie Farret, CTO of SGDL Innovation SA
Through strategic working groups, sgdl promotes open standards for complex geospatial and robotic simulation, with direct access to leading technology companies:
FLAGSHIP PROJECT
Thinking Crystal
Inventing new materials through mathematics and AI
The Concept
Where materials discovery is traditionally slow and empirical, Thinking Crystal uses a deterministic "Function to Form" approach — the same deductive direction that Noether's theorem uses to derive conserved quantities from symmetry. Instead of trial and error, sgdl computes the exact geometry needed to meet any functional constraint. The structural approach is exponentially more powerful than brute-force search: the ratio of effort between structure and search grows without bound as problem complexity increases.
Polyhedral Generator
sgdl can deterministically calculate the exact geometry to meet any functional constraint — generating precise polyhedral structures that satisfy target properties by construction, not by chance.
Accelerated Innovation
Innovation cycles of 18-36 months can be replaced with computational testing of thousands of structural variants in hours. This is a paradigm shift in materials science.
Connection to Nobel Chemistry 2025
Omar Yaghi's Metal-Organic Polyhedra (MOPs) demonstrate that materials can be programmable. The 230 crystallographic space groups — a direct application of Noetherian symmetry classification to matter — provide the structural taxonomy. sgdl inverts the traditional reasoning: from structure-to-properties to function-to-form, providing the mathematical engine to design materials by specification. sgdl has processed tens of thousands of crystallographic structures with exact spatial reasoning.
ROADMAP
What's Next
Technology Validation
R&D projects with CERN, benchmarking, patent maturation.
Industrial Validation
Partnership deployments, Thinking Crystal launch.
Industrialization
Technology licenses, Ariadne platform.
Scale
PaaS platform, spatial reasoning infrastructure for Physical AI.