Products

AI solutions for advanced material analysis
Software that integrates benchtop NMR spectroscopy and Artificial Intelligence to support analysis and characterization of complex industrial materials.

Built for real materials, not synthetic demos

Industrial materials are not “standard”: they vary by process, supplier, operating conditions, and intrinsic variability. That is why generic models often fail when moving from the lab to real workflows.

Materials are not standard
Generic models rarely generalize reliably
AI must be built and verified on real experimental data
We work closely with customers to define metrics, datasets, and validation criteria

Discuss a pilot

The platform: Material Intelligence Platform

Material Intelligence Platform is platform designed to bring benchtop NMR and domain-specific AI into industrial workflows—with traceability and reproducibility as first-class requirements.

NMR instrument workflow

Acquisition, organization, and preprocessing of spectra.

Feature extraction

Reproducible pipelines designed for material families.

Model training & validation

Clear metrics and robust evaluation across datasets.

Interpretability & traceability

Comparable outputs over time and across campaigns.

Designed for a reliable path to industrialization: measurable validation, traceability, and a clear roadmap from pilot to production.

Overview

Two modules, one platform

Material Intelligence Platform is delivered through pilots and industrialization programs. The modules below are part of one platform—designed to be validated on your experimental data and to remain comparable over time.

Rombo AutoML

AutoML built for spectral data—validated on your experimental datasets.

Learn more

Module 1

NMR AI Analyzer

A module focused on benchtop NMR operations and on keeping analysis consistent over time. Designed for industrial labs where traceability and comparability matter.

End-to-end workflow

Measurement workflows, acquisition, organization.

Automated signal analysis

Preprocessing + extraction tailored to your context.

Material-family models

AI models specific to the material family under study.

Comparable outputs

Interpretable results consistent across campaigns.

Module 2

AutoML for material analysis

An advanced module to build custom AI models, designed for technical and R&D teams. It supports model selection, validation, and comparison—grounded in experimental datasets.

Feature selection and pipeline comparison
Training and validation with agreed metrics
Performance comparison across approaches/models
Support for adapting to new datasets (data quality + validation criteria)

How it’s delivered

Best for technical and R&D teams. We typically start with a structured pilot to define objectives, validate on your experimental data, and agree on success metrics before scaling to industrial workflows.

Use cases

Validated on your experimental datasets

Two examples of where the platform delivers measurable value. Each deployment starts with a structured pilot: clear objectives, success metrics, and validation on your data—then scales across batches, operating conditions, and sites.

Crude oil

Fingerprinting and comparison across blends, batches, and suppliers
Classification and similarity search with traceable features
Link spectral patterns to properties and process variability
Early detection of off-spec / drift with agreed thresholds

What you can expect

A validated workflow on your crude family and NMR setup: reproducible preprocessing, model performance on held-out data, decision criteria, and a path to scaling across new datasets and operating conditions.

Mineral oil (transformers)

Batch comparison and classification with reproducible indicators
Condition / aging proxies designed with technical teams
Trend monitoring over time and across maintenance cycles
Reporting outputs that remain comparable across campaigns

What you can expect

A pilot-defined set of indicators and validation metrics, tested on your samples; plus a scalable workflow to extend coverage to new oils, conditions, and sites without losing comparability.

How we work

Product-oriented pilots

We start with a structured pilot: clear objectives, agreed success metrics, and validation on your experimental data. You get a working solution and a roadmap to scale into operations.

1) Define objectives

What we measure and what we predict/classify.

2) Validation metrics

Accuracy, robustness, reproducibility, stability over time.

3) Data plan

Quality, quantity, and acquisition protocols.

4) Industrialization roadmap

Integration, governance, and model maintenance.

What you get: validated results, clear documentation of metrics and datasets, and an industrialization roadmap.