Beyond discovery
Autonomous translation, not just autonomous discovery
Materials-discovery SDLs ask which candidate is promising. We build laboratories that answer the next two questions: can it actually be made in a real unit process — and can it be scaled?
The loop
Anatomy of a self-driving process lab
One closed loop, repeated: recipes go in, processability comes out.
- Machine-readable recipes — drive automated unit processes — mixing and dispersion, coating, drying — with the full process history recorded.
- In-line analytics — read the raw materials, the intermediate product — slurry, ink, wet film, where failure actually happens — and the final layer; vision-based sensing even reads viscosity straight from video.
- Decision AI — plans the next experiment — Bayesian optimization over physics-informed models with calibrated uncertainty.
The deliverable
Processability maps, with failure as first-class data
The output is not one good sample — it is a calibrated map of where the process works, where it fails, and how confidently we know.
- Probabilistic quality windows — Gaussian-process quantile regression turns scattered experiments into statistically defined operating windows.
- Negative results are data — failure conditions are recorded and learned from, not discarded — the least-managed data layer in manufacturing R&D.
- Scale-up factors, measured small and fast — rheology, drying behavior, and process history feed physics-based models and digital twins — instead of robotizing the pilot plant.
Roadmap
Toward autonomous R&D for liquid-based manufacturing
We are building modular SDL cells around our own unit processes — and the interfaces that let them work as one system.
- Module by module — automated mixing and dispersion first, then slot-die coating, then drying — reconfigurable cells with automated sample transport.
- Physics + AI, not black boxes — conservation laws and domain models anchor scarce, noisy process data, so the loop stays trustworthy at scale-up.
- The destination — laboratories that judge whether a new material can be made and scaled — closing the gap between discovering a material and manufacturing it.