Funding
Funding not disclosed
Founders
Product
Problem
Legacy optical design tools such as OpticStudio® and Code V require weeks of manual setup and costly labor to define merit functions, and their Monte Carlo tolerance analyses run for weeks or months while missing non‑linear manufacturing effects. This slows product development cycles for AR/VR, LiDAR, quantum photonics, and biomedical imaging systems.
Solution
Paraxial offers an AI‑native optical design platform that replaces traditional tools with a workflow built around differentiable ray tracing and hybrid AI tolerancing. Users specify performance goals in natural language, and the AI co‑pilot instantly generates merit functions and an initial lens layout. GPU‑accelerated automatic differentiation enables gradient‑based global optimization of the entire optical train, dramatically reducing design iteration time. For manufacturing analysis, surrogate‑model‑driven yield assessment combined with active learning delivers tolerance results 10–100× faster than brute‑force Monte Carlo while capturing complex, non‑linear sensitivities. The platform also provides seamless CAD export and API integration, eliminating downstream rework.
Target Audience
Primary customers are optical engineers and system designers in industries such as AR/VR displays, automotive and robotics LiDAR, quantum photonics, and biomedical imaging, as well as mechanical and systems engineers who need to contribute to optical design without extensive specialist training.
Features
- Natural‑language interface that creates merit functions and starting designs in minutes
- Differentiable ray tracing with GPU‑native backpropagation for gradient‑based global optimization of surfaces, coatings, and materials
- Hybrid AI tolerancing using Gaussian‑process surrogates and active learning to accelerate yield analysis 10–100× over traditional Monte Carlo
- Real‑time feasibility checks and trade‑off exploration across the full design space
- Direct integration with mechanical CAD tools and external APIs via the Model Context Protocol (MCP)
- Support for high‑performance optics in AR/VR, LiDAR, quantum photonics, and biomedical imaging applications