What needs you, not a wall of green

The panel you open first

Every vessel page leads with a "needs attention" panel: current alarm state, high-priority open worklist items, and trend flags drawn from the 2-sigma anomaly counts in the daily rollup. Under it sits a readiness score, 0-100, with the component breakdown that produced it. The score is meant to be argued with, not believed. The fleet overview is a league table ranked by readiness, and every row clicks through to the ship. All of it is vessel-level: the crew aboard sees the same picture shore sees, and LOOP does not score individuals.

• Priority "needs attention" panel first on every vessel page, before any chart
• Readiness dial 0-100 with a component breakdown, so you can see which part moved the number
• Fleet league table ranked by readiness; rows link straight to the vessel page
• This-ship-vs-class charts on each vessel page — today the class baseline is four demo sisters, and they are labelled demo
• Stakeholder lenses — the same underlying data presented for different readers

The two tables that answer the email you were about to send

The compare section carries a rounds table and a worklist table across the hulls. Rounds: compliance for today and the last seven days, completed against scheduled. Worklist: open items and high-priority open items per vessel. Alongside them sit about 20 headline trends with 30-day series and 2-sigma anomaly counts, and the consumption rates the rollup carries daily. Two things to hear before you look: rounds are counted per vessel and never per person — there is no named per-user login aboard today, and per-reading attribution is a roadmap item — and only one hull is sending real data.

• Rounds comparison across hulls: compliance for today and the last seven days, completed against scheduled
• Worklist comparison: open items and high-priority open items per vessel
• About 20 headline trends per vessel with 30-day series and 2-sigma anomaly counts
• Consumption rates carried daily — fuel t/day, cylinder LO L/day
• One hull sends real daily rollups today; the others are clearly labelled demo, cross-hull deviation figures are withheld while the fleet is mixed, and sister-ship deviation bands need at least three real hulls in build

Photos arrive with the item. Approvals happen in your inbox.

Up to eight compressed issue photos ride in with the daily rollup and land in a grid on the vessel page. The intent is to turn "leak at number two" into something a contractor can scope before the ship berths — that has not been measured, so take it as the design goal, not a result. Worklist items carry priority, photos, and pending-review and pending-approval states, with soft delete and restore. Remote approval by email already works: shore replies APPROVE or REJECT and the item moves.

• Up to 8 compressed issue photos per vessel per day, shown as a grid on the vessel page
• Worklist items carry priority, photos, pending-review and pending-approval states, with soft delete and restore
• Email approval works today: reply APPROVE or REJECT, no portal login, no app
• Watch notes with photos and carry-forward, so what the 04-08 watch saw does not evaporate at turnover
• Weekly reports with AI summaries, and TSV export when you want the numbers in your own spreadsheet

Built for a per-GB link, and for no link at all

The ship does not depend on you being reachable. All data is local — Android tablets, a Windows hub, a Raspberry Pi engine-room display, an Isar database aboard — and the AI assistant runs on the ship with no internet. Sync is opportunistic and small: about two documents per vessel per day, typically under 5 KB, plus up to eight compressed issue photos. Raw history and full-resolution media stay aboard.

• Every core function runs with zero connectivity: rounds, inline alarm evaluation, trends, worklist, the Pi display, the assistant
• Daily rollup carries rounds compliance (today and 7 days), worklist stats, alarm state, about 20 headline trends, consumption rates, plant changeovers, equipment status, recent notes, a readiness score with component breakdown and an AI context digest
• If the link is down, shore simply goes without that day's summary — the ship keeps rounding, evaluating alarms and recording on its own
• Live AIS positions for the fleet on a Windy weather map with 24-hour dead-reckoned tracks; three of five hulls have been seen reporting real AIS positions
• The shore dashboard is password-protected, and the crew aboard sees the same data it shows

What is next, in the order it matters to you

None of this section exists yet — it is the build order, not the product. The defect lifecycle is next: shore-acknowledgement fields on each defect and an exportable open-defect register, so a surveyor or an owner can be handed a register instead of a re-typed list. The record shape — defect recorded, dated, acknowledged by shore — is the shape SIRE 2.0 rewards; SIRE 2.0 is a tanker instrument and does not bind a US-flag container ship, so it is cited here as good practice, never as compliance. After that, the contractor handoff pack. When the NS5 bridge lands it will be read-only: LOOP will not create a work order.

• Defect lifecycle with shore-acknowledgement fields and an exportable open-defect register in build
• Contractor / riding-squad handoff pack: open items, photos, trends and notes assembled before they board in build
• NS5 (ABS Nautical Systems) read-only bridge — one-way in, human in the loop, no work orders created by LOOP in build
• True-copy PDF exports and incident or survey evidence bundles in build
• Append-only, signed, tamper-evident record store and named per-user login aboard — until both land, the shipboard record layer is mutable and capacity-capped and there is no per-reading attribution, so today's records are not an evidence trail in build

What is shipped, and what is not

This table is the point of the page. It is written to be read by a sceptic, and it is kept honest deliberately — every claim here is checkable against the running system.