oriented models
Data compatible
Architecture search
Smaller models
models
Build accurate models in record time
Fully automatic : one button to build the model. No trial and error required.
Accurate models from small datasets
Model structure shaped by the physics, enough to learn, not enough to overfit.
The physics is in the architecture
Aerospace, manufacturing, energy, and more — where physical meaning matters.
Efficient output for embedded systems
Compact models suited for real-time monitoring, virtual sensors, and control.
No AI expertise required
A guided workflow for engineers: no hyperparameter tuning, no AI plumbing.
What happens when you need to change something late in the project?
Build parsimonious digital twin from design to operations on one physics-grounded platform. During operations, deploy compact models that improve your system, through a software update with no hardware redesign.
Learn from CAE/FEA/CFD or parametric DoE to extensively explore the design space in a split second.
NeurEco will help to select the most relevant experiments, letting your build the best model with minimal data.
Deploy compact models on existing hardware to power real-time control, virtual sensing, and decision support.
NeurEco Tabular
Build predictive models mapping inputs to continuous physical outputs — ideal for surrogate models of simulation results.
NeurEco Tabular
Supervised classification for discrete outputs — fault detection, regime identification, quality control.
NeurEco Tabular
Autoencoders compressing high-dimensional data (stress–strain fields, spectra) into compact latent representations.
NeurEco Dynamic
Recurrent neural models for time-series and real-time control — handles quasi-chaotic behavior with long-term accuracy.
NeurEco PFS
Surrogate modeling of frequency-domain responses, S-parameters, and antenna characteristics over parametric sweeps.
NeurEco Grid
Spatial field prediction from geometry or mesh data — ideal for stress, temperature, and flow field surrogates.
Predict failures before they happen
Active battery protection.
From predicting failures to prescribing actions
Prescribe maintenance actions to maximize aircraft availability.
Remedy partial failures through software updates
Keep array antennas and radars operational.
Higher throughput, less waste
Improve production quality and reduce waste in chemical process.
Fewer tests, faster iteration
Reduce very expansive crash tests by 50%
Pro-active protection, hardware cost reduction
Preserve engine shafts and turbocharger blades from fatigue and overload
"I’ve worked with Adagos on several projects using their NeurEco software: drift detection, fast meta-models, and virtual sensors. Their responsiveness, adaptability, and data-efficient approach — with no hyperparameter tuning — make them a key partner for future control system applications."
Stéphane Lagarde
"We were very satisfied with our collaboration with Adagos. With limited data, their NeurEco Dynamic toolbox accurately modeled our system’s dynamics. Their integration of cell aging into anomaly detection and strong involvement made the project especially effective."
"The stakes are high in embedded systems. We closely monitor innovation through top universities and start-ups. Yet the breakthrough came from Adagos, with a truly unique approach. We see them as a game changer in this field."
Michel Rochette
"The application of NeurEco to our predictive maintenance project was very interesting. These results are remarkable because the data is rather scarce."
