Three patterns where AI could help in classification work
Three areas where AI orchestration could sit alongside Class NK’s existing work. Each rests on the same orchestration substrate; each could stand alone.
These pages document patterns. None of them assume Class NK does anything. They exist so people who weren’t in the meeting can follow what was demonstrated, and so people who were can return and read at their own pace.
Pattern 1 — A knowledge layer
A natural-language search agent over a corpus of rules. A surveyor or rule editor asks a question; the agent returns relevant clauses with citations to chapter, paragraph, and source document.
For the meeting demonstration the corpus includes 10,000+ pages of Class NK rules (converted to markdown), plus SOLAS, MARPOL, and public IMO source documents. A natural extension adds a Rule Change-Tracker chain on top — surfacing IMO amendments and the Class NK rules they touch.
Pattern 2 — Validation at scale
An agent team that handles the routine parts of validating client data submissions. The agents do the cross-checking, the anomaly detection, the casefile management, the audit logging. The surveyor stays at every judgment seat.
This sits naturally alongside ZETA — which already handles GHG / CII / MRV / FuelEU compliance for 5,000+ ships, with NAPA Fleet Intelligence integration since July 2025. The orchestration substrate proposed here would add an agent layer on top of that pipeline, not parallel to it.
Pattern 3 — Department agent teams
The same orchestration pattern, applied per department. Each Class NK department gets an agent team that watches for relevant events, suggests what to do next, and drafts the first version of the work product. Humans approve, edit, decide.
This sits naturally on top of ShipDC / IoS-OP — the neutral data exchange Class NK already owns through its wholly-owned subsidiary, with NYK / MOL / K-Line / ONE already connected. The data substrate exists; the agent layer would sit above it.
Before any of these patterns — the context we bring
We have been operating on Class NK data for several years, on the ship-operator side. We scrape the Class NK portal for ship-operator clients. We track Class NK certificate lifecycles. We coordinate around survey windows and follow defects to closure with Class NK surveyors.
That existing operational context is what shaped these three patterns. They are not abstract proposals; they came from watching Class NK’s processes daily, from the customer side, for the owners we serve.
If a conversation continues
These pages document patterns and demonstrations. Whether any of them would be worth pursuing further is for Class NK to decide.
If something does resonate, our preference would be to follow patterns Class NK already uses for engaging with technology — Innovation Endorsement, Technology Qualification, the Digital Twin Project consortium — rather than propose anything new.