Real Battery Health. Not BMS Guesswork.

Standard BMS readouts can't capture non-linear degradation or dynamic internal resistance — and full-cycle lab testing can't scale. Sphere clears that bottleneck, estimating SOH with lab-grade accuracy from a single rapid charge, in minutes.

From Raw Data to Battery Intelligence

01
Pre-training

Trained on 4,000+ real-life cycling tests and >100M material configurations, combined with physics-based P2D models the model learns the correlation of cell-physics and aging.  

02
Fine-tuning

Augment model capabilities with historical operational data and additional use case-specific inputs. Develop a custom customer interface to monitor and estimate the SOH of independent assets.

03
Inference

Feed real-time operational data to obtain instantaneous SOH predictions and expected aging curves. Benchmark independent assets and unlock rich comparative performance insight.

One model. All degradation knowledge.

Our Solutions
Electric Vehicle Ecosystems
Few-Shot Field Calibration
Explainability Mechanisms

The model builds a firm baseline by comparing standard BMS readouts and vehicle mileage against historical fleet averages, then refines it using time-series data from a rapid, 10-minute dynamic voltage check.

Discover More

Decomposes capacity loss into its root causes — SEI growth, active material loss, and lithium plating — and flags degradation onset before it accelerates.
Drops operational SOH error from 5% to 2.5%, so warranty decisions and asset valuations rest on mathematically rigorous ground.

Discover More

Ensure interpretability of models by providing insights into model confidences and feature contributions

Discover More

See VETTA estimate SOH in real time

From raw charging data to lab-grade accuracy — in one rapid cycle.

Use Cases & Benefits

  • Faster Engineering Decisions

    Reliable lifetime predictions after just weeks of cycling data. No more waiting 12+ months for full test results before making design, sourcing, or certification decisions.

  • Reduced Test Matrix (DOE)

    Simulate untested combinations of temperature, C-rate, SOC window and usage profiles. AI identifies which experiments can be replaced — typically 30% fewer physical tests.

  • Automated P2D & ECM Parametrization

    Automatically parametrize Pseudo-2D and Equivalent Circuit Models from initial test data. Eliminates weeks of manual fitting — production-ready physical models in a fraction of the time.

  • Virtual Cell Qualification

    Simulate end-of-life behavior for new cell candidates before committing to full testing. Pre-screen suppliers and chemistries with AI-powered predictions from minimal data.

We are here to make your diagnostics faster

Schedule a call