Every capability we bring to industry, we bring to defence.
Defence assets are high in value, hard to reach, operate with intermittent or contested connectivity, and generate very little failure data. That is not a new problem for us. It is the environment our whole approach was built in.
The hard constraints are the same ones we already work under
We are not repositioning a commercial product for a defence audience. The conditions that make defence monitoring difficult are the conditions we designed for in the first place, offshore, where a technician cannot simply be sent out to look.
Access is the constraint
When reaching the asset is expensive, slow or occasionally impossible, the value of answering a question remotely goes up sharply, and so does the cost of a wrong answer.
Sparse and imperfect data
Few sensors, short records, gaps and events that are rare by design. Physics-anchored models are built precisely for the case where examples are the thing you do not have.
Connectivity you cannot assume
Intermittent, degraded or deliberately denied links. Computation at the sensor and store-and-forward in firmware are how we already operate, not a feature we would have to add.
Decisions that must be defensible
Results grounded in physical terms rather than in an opaque model, so an engineer with authority can check the reasoning and stand behind the conclusion.
Four things we do, wherever the asset sits
Structural health of platforms and structures
Operational modal analysis, modal expansion and physics-based models that turn measured motion into loads, fatigue and remaining life. Sparse instrumentation, large dynamically excited structures, and a change indicator referenced against an asset's own history and its peers.
Sensor trust and data integrity
A reliability judgement on the measurement itself, channel by channel, with a readable cause behind every flag. In degraded conditions the question of whether the data can be believed comes before any question the data is meant to answer.
Edge AI under intermittent or denied links
Computation at the sensor rather than streaming raw data rearward. The operator receives features and health flags, not waveforms, so the link carries a fraction of the volume and the system keeps working when the link does not.
Fleet-level prioritisation
The question at portfolio level is rarely about one asset. It is which asset needs attention first, on what evidence, and how confident that ranking is. Comparing an asset against its own history and against its closest peers is what makes that ranking possible.
The technical foundation
- Optical sensing
- Fibre Bragg grating transduction turns strain, temperature and acceleration into a shift in the wavelength of light. No electrical signal at the sensor, no metallic loop and no induced voltage, so electromagnetic interference is not a failure mode. We run optical in production today.
- Sensor-agnostic processing
- The algorithms are independent of sensor type. Fibre Bragg grating, MEMS, piezo and vibrating-wire data all fit, so existing instrumentation can be used rather than replaced.
- Physics-anchored models
- Physics supplies the fundamental behaviour and the available data supplies what is specific to the asset. This is what allows a defensible prediction from a short record, and what keeps the result explainable rather than opaque.
- Edge computation and resilient delivery
- Feature extraction on the device, buffering and store-and-forward in firmware, and time alignment across units. In our own production fleet this raises effective data delivery to 99 percent or better across links with roughly 90 percent connectivity uptime.
- Data handling
- Measurements from your assets are yours. Storage location, including whether data stays inside the European Union, is agreed before work begins, and non-disclosure and data processing agreements are standard rather than negotiated.
A Dutch engineering SME, in two roles
As a consortium partner
- Founded in 2020 in Brainport Eindhoven, owned entirely by its founders, funded through government and bank loans and research subsidies.
- Working relationships with TNO, TU Delft and TU Eindhoven, including a national research consortium on chip-scale optical interrogation.
- Used to being the specialist depth in a larger programme rather than the prime.
- Small enough that the people you meet are the people who do the work.
As a specialist supplier
- We bring algorithms and data capability, not hardware procurement. Your instrumentation and your integrators stay in place.
- We do not issue verdicts on whether an asset is fit for service. We report what changed and by how much; the authority to decide stays where it belongs.
- We are explicit about uncertainty. An honest range is more useful than false precision.
- Where a capability is not ready, we say so rather than let a proposal imply otherwise.
A conversation about your assets and your constraints.
The useful first step is usually narrow: one asset class, one question, and an honest read from us on whether our methods apply to it. We are equally comfortable joining a consortium and working directly as a specialist supplier.
