All products
Product · Live

GeoSolv

Find safer, higher-conductivity Na-ion electrolytes — at 105 screened per pilot, with calibrated confidence on every shortlist.

Pipeline3D mixture → property prediction
3D POINT CLOUDmixture as set of atomsSET-SE(3) MESSAGE PASSINGrotation- + translation-equivariantPHYSICS-CONSTRAINED OUTPUT5 mS/cmσ(T), η — Arrhenius-valid
0.00

Held-out generalization

10⁵

candidates

Per Na-ion shortlist

0.0

mS/cm

Best predicted

Built withPyTorchJAXCheMixHubAIMEMS NRT
The problem

ML for mixtures collapses on unseen chemistries.

01

2D screens promote candidates that fail in cells.

Fingerprint-based ML scores molecules independently. Mixtures aren't separable — pairwise solvation governs whether ions move at all. Two-formulation differences invisible in 2D ruin a cell-build at week 12.

02

Small composition shifts flip viscosity and stability.

Property surfaces over composition aren't smooth — they have ridges where the model needs to be sharpest. Grid search on a 6-component blend explodes; off-grid points are exactly where the gains live.

03

Solvation shells govern transport. We learn the shell.

Ionic conductivity and viscosity track shell geometry, not connectivity. Set-SE(3) message passing puts the 3D shell into the model — that's why GeoSolv generalises to molecules it has never seen during training.

How it works

A physics-constrained pipeline, end to end.

01

3D point cloud encoding

=ϕ(x)ϕ(R · x + t)
02

Set-SE(3) message passing

5 mS/cm1/T →log σσ = A · exp(−Eₐ / k T)
03

Differentiable Arrhenius readout

5.4mS/cm shortlist
04

10⁵ candidate screening

What you get

+22 R2 over the prior best.

Arrhenius conductivity trace

top Na-ion candidate
5 mS/cm — viability thresholdInverse temperature (1/T) →log σ (mS/cm)σ(T) = A · exp(−Eₐ / k T)

CheMixHub leave-molecules-out

R² · higher = better
Random Forest (ECFP)R² = 0.42
Chemprop (2D GNN)R² = 0.58
MIX-MPNNR² = 0.61
GeoSolvoursR² = 0.83
How a pilot runs

Eight weeks. Fixed scope. Real deliverables.

We engage as a paid 8-week design partnership — not a one-off API call, not a consulting hand-wave. You bring the chemistry; we return a ranked, physics-validated shortlist plus the model head. Prediction IP stays with you.

Week 0 · You bring

Your chemistry, your target.

  • A SMILES list or sketch of the formulation space
  • Target property + threshold (e.g. > 5 mS/cm at -10 °C)
  • Optional: any in-house measurements you can share
Week 8 · You get back

A shortlist you can build.

  • Top-50 candidates ranked by composite objective
  • Calibrated confidence intervals on every prediction
  • MD-grounded sanity report on the top 10
  • Synthesis notes + a fine-tuned model head you keep
Throughout · How we handle data

Your data, your namespace.

  • Isolated tenant — separate buckets, IAM, KMS keys
  • TLS 1.3 in transit, AES-256 at rest, hardware KMS
  • Wiped on engagement end or 30 days, whichever first
  • You keep IP on every derived prediction and shortlist
Work with us

Have a Na-ion formulation problem?