# Trust Layer for Operators
The customer is the **plant operator**, and the product is a **trust layer**: a system that continuously verifies whether a sensor signal can be believed.
> [!quote] The core problem, in the founder's words
> Today, operators "trust the data, whatever comes from the signal — there is nothing more we can do." They cannot verify a reading without physically going to the asset, which they do roughly **every 8 hours**. A small fault — e.g. a drop of oil on a sensor — silently degrades the reading, and the operator has no way to know between manual checks. "That is enough for a disaster to happen."
IK gives the operator a continuous, physics-grounded second opinion: the signal is cross-checked against physical law (input/output mass balance, pressure behaviour) so the operator knows *why* a reading is trustworthy or degraded. [[Four-State Trust Machine|Trust is the output]]; at this stage outputs are **read-only / advisory** because operators won't allow anything to disturb live control — see [[Deployment Path]].
> [!tip] The ROI argument
> Physical laws are not probabilistic — "physics is the brain that confirms you are right." If IK prevents **one** unplanned shutdown, the avoided cost dwarfs the price of the system. Context: in UK water, undetected integrity failures carry fines of **£50M+**, and models often "run on belief" — assumed flow rates and pipe states no one actually measures.
This framing — operator-facing, verification-first — is what differentiates IK from the telemetry vendors it sits above (see [[Layered Positioning]]).
---
Related: [[Integrity Kernel MOC]] · [[Four-State Trust Machine]] · [[Fail-Closed Trust Gate]] · [[Deployment Path]] · [[Layered Positioning]] · [[The OT Intelligence Layer]]