oriented models
Data compatible
Architecture search
Smaller models
models
Build accurate models in record time
Shorten iteration loops from design exploration to validated behavior.
Reliable results with fewer samples
Works when data is scarce, expensive, or slow to collect.
Built for demanding engineering domains
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?
One compact, physics-grounded model that stays usable from design to operations — and can be improved through software updates, without hardware redesign.
Learn from CAE/FEA/CFD or parametric sweeps to explore the design space faster.
Concentrate tests on the most informative runs, and update the model as results come in.
Operate compact models for real-time monitoring, virtual sensors, and decision support on existing hardware.
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 & prevent battery failures before they happen
Digital twins that learn normal battery behavior and catch anomalies before they escalate.
From predicting failures to prescribing actions
Models that recommend the right corrective action and optimal timing to extend asset life.
Autonomous antenna recalibration by software
On-board models that detect drift and recalibrate antenna parameters automatically.
Maximize throughput, minimize waste
Lightweight digital twins on production lines to optimize parameters and reduce downtime.
Fewer tests, faster iteration
Accurate surrogates from limited runs to cut test campaigns and improve design coverage.
Extend lifetime by software
Keep improving performance through software upgrades — delay replacement cycles.
"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."
