Signal-integrity layer
“Is my position real?”
GNSS / AIS / network spoofing + jamming detection. RF-domain, cyber-physical. The 2024–2025 spoofing incidents (Beirut, MSC Antonia, Front Eagle) made it investable.
Platform · Future-proofing
The software-defined vessel — and the discipline of what not to build.
The vessel is becoming a software-defined platform — every subsystem a structured data layer that the others can read. The future-proof move is not to build more layers. It is to be the cleanest, most consumable version of exactly one layer: physical-asset condition. This page states, plainly, where Hullproof sits and what it refuses to build.
The layers of a software-defined vessel
NTNU's Maritime AI institute (Trondheim, 2026) frames the future ship as software-defined. In that frame, the vessel is a stack of structured data layers. Here is the stack — and exactly one of these is Hullproof.
Not Hullproof
OT / machinery layer
“Is my equipment behaving?”
Time-series anomaly detection on sensors, engines, pumps, heat exchangers. NVIDIA NV-Tesseract, Cognite, sensor-fusion platforms.
Condition-integrity layer
“Is my physical asset sound?”
Hull, coating, structure, biofouling, corrosion, weld defects — visual condition that no sensor reads and, on a crewless vessel, no human sees. This is Hullproof.
Navigation + autonomy layer
“Where do I go and how?”
Route planning, collision avoidance, autonomous control. Kongsberg, Massterly, the autonomy-system integrators.
Vessel decision support
“Given all of the above — what do I do?”
The layer that cross-references signal, OT, condition, and navigation into one operational picture. Hullproof is a structured INPUT to this — never the layer itself.
How far to build — in plain numbers
The honest answer to “should Hullproof build toward GNSS, OT, and vessel decision support?” — mostly no, and the no is deliberate.
GNSS / signal integrity
Build: 0%. Interoperate: 100%.RF-domain spoofing detection is a deep specialist field. Building it would be scope collapse. Instead: CoatingPassport is a clean structured input that a signal-integrity-aware decision layer can cross-reference. We stay out of the RF domain entirely.
OT / machinery time-series
Build: 0%. Interoperate: 100%.Time-series anomaly detection is the NV-Tesseract / Cognite space. Hullproof already exports to Cognite CDF so condition findings sit beside OT anomalies on one Asset hierarchy. Complementary channel, not a competing build.
Vessel decision support
Build: input layer only.Hullproof is the physical-condition input to vessel decision support — never the decision layer itself. The decision layer is navigation + safety + cyber territory. Our REST + MCP + JSON-Schema surface is already built so any decision-support system can consume CoatingPassport.
Condition intelligence
Build: 100%. This is the whole company.Footage-agnostic ingest, structured findings with confidence + lineage, drift detection, VQA, the VLA seam, edge-deployable inference. Everything compounds here. Nothing is spent outside it.
Why the discipline IS the future-proofing
Decision layers come and go. The vessel-decision-support system of 2026 will not be the one of 2032. The cyber- integrity vendor, the autonomy-system integrator, the OT- anomaly model — all of those will be replaced, merged, or re-platformed across the next decade.
What does not get replaced is the need for a structured, audit-traceable, machine-readable record of physical-asset condition. Every decision layer — this generation and the next — needs that input. By being the cleanest version of that one layer, and refusing to dilute into navigation or cyber or OT, Hullproof stays relevant to whatever sits above it. A company that is one excellent layer survives re-platforming. A company that is a mediocre four layers gets unbundled.
The technology curve — and why CoatingPassport rides it
The model generation changes. The data contract does not.
Today — VLM
Vision-Language-Models read footage and emit structured findings. CoatingPassport is produced by a VLM call.
Next — VLA
Vision-Language-Action: a finding triggers an autonomous action (closeup capture, robot dispatch, work order). The VLAProvider seam is already in the codebase.
After — embodied / physical AI
Inspection robots that see, reason, and act in one loop. CoatingPassport remains the structured intermediate object they read and write — the data contract outlives the model generation.
VLM today, VLA next, embodied / physical AI after that. Through all three, CoatingPassport is the structured object that gets produced, queried, diffed, and acted on. The seam (VLMProvider, VLAProvider) lets the backend swap without touching the data contract. That is the mechanism of future-proofing — not predicting the model, but owning the contract.
For vessel-decision-support builders
Building a vessel-decision-support system, a Remote Operations Center, or a software-defined-vessel platform? Hullproof is the physical-condition data layer you plug in — REST, MCP, JSON Schema, OSDU, CDF, all already built. You own the decision layer; we own the condition input. Clean seam, no overlap.