fleetcore is a maritime technical operating system where AI is not an add-on — it is the foundation. Five autonomous agents are embedded across every core operational workflow: predictive maintenance runs a censoring-aware ML stack every 15 minutes against every equipment installation; a procurement agent closes the loop from inventory threshold to ranked supplier award automatically; an incident intelligence agent proposes predictive events 200–800 hours before a failure window; a compliance reporting agent pre-populates three report templates as AI-draft artifacts; and a conversational fleet intelligence agent answers any fleet question in plain language with 30+ maritime-domain handlers. Every agent action is gated by a three-tier human-approval governance model. The machine prepares and proposes — the crew and superintendents decide and approve.
On top of this agentic foundation, fleetcore centralizes OEM maintenance intelligence from 100+ maritime manufacturers — MAN B&W, Wärtsilä, Caterpillar, Kongsberg, ABB, Rolls-Royce, Alfa Laval — into a single cloud-native platform with SOLAS 2024 / MARPOL / ISM Code compliance built into the architecture.
Five Agents. Every Workflow Covered. Every maintenance cycle, procurement inquiry, incident alert, compliance report, and fleet conversation is governed by a purpose-built agent. Humans stay in command — the platform does the preparation.
Censoring-aware survival analysis runs continuously against every equipment installation using maintenance history, Equipment DNA operational embeddings, and cross-fleet federated priors. Outputs calibrated P05/P50/P95 Remaining Useful Life bands. Detects divergence from OEM baseline and proposes schedule interval adjustments bounded by criticality class (SOLAS-critical and important equipment have separate limits). Complements historical predictions with live streaming anomaly detection and third-party sensor feeds.
Human gate: All schedule and task mutations require role-gated approval — the agent never modifies operational records directly.
When inventory drops below the reorder threshold (computed from consumption history), the agent assembles a structured inquiry, dispatches it to all approved suppliers, parses inbound responses via AI to extract line items and prices, benchmarks offers against historical pricing data, and ranks them by price delta, lead time vs. RUL urgency, and supplier reliability. The ML-Procurement Bridge also triggers a pre-check automatically when a RUL estimate falls below twice the average supplier lead time for critical parts.
Human gate: Award decision is always manual. Initial inquiry drafts require procurement role review before dispatch.
At Tier 2 confidence, the agent proposes a predictive incident 200–800 hours before the failure window — bundled with a corrective maintenance task template as an atomic unit. Severity is mapped from RUL percentage (critical / high / medium / low). On root cause completion, cascades into a new preventive schedule proposal. Escalates to secondary approver if unresolved within the expiry window.
Human gate: Incident creation and corrective task assignment require operations role approval — predictive events are proposals, not automatic record mutations.
Three report templates are pre-populated automatically: weekly RUL summary (Monday 08:00 UTC) aggregating predictions across the fleet; anomaly alert report triggered by confidence threshold breach; and interval adjustment recommendation with full audit chain and regulatory justification. All are created as AI-draft artifacts. Promotion to submitted status requires explicit human action — never automated.
Human gate: Draft promotion is always manual — the machine prepares, the human submits.
Ask any fleet question in plain language: overdue tasks, inventory levels, RUL forecasts, procurement status, compliance certificates, crew records, fleet KPIs, and financial cost breakdowns. Multi-session memory maintains vessel context across conversations. Intent routing across 30+ domain-specific handlers. The conversational agent reads across all modules and surfaces the answer — no switching between screens.
Human gate: Read-only by default. Any write actions (task creation, approvals, record mutations) follow the same Confidence Score HITL model as the four operational agents.
→ Full AI Intelligence documentation
The predictive maintenance agent draws from three data layers fused into a single confidence-weighted composite RUL:
The entire procurement cycle runs automatically across six phases with one human gate — the award decision:
fleetcore centralizes maintenance intelligence from over 100 maritime equipment manufacturers into a single vendor-neutral source of truth. Equipment naming is automatically normalized across vessels — "CAT", "Caterpillar", and "Cat Engine" resolve to the same entity. One-click PMS import pre-loads manufacturer-verified maintenance schedules. Vessel onboarding from days or weeks to hours.
Supported manufacturers include: MAN B&W (6S50MC-C, 6L70ME-C, 5G80ME-C), Wärtsilä (32, W20, W26, 20DF), Caterpillar (3516B, C32, 3406), Kongsberg (K-Chief, AutoChief C20), ABB (Azipod, turbochargers), Rolls-Royce, Alfa Laval, Wartsila (heat exchangers), and 90+ additional manufacturers across main engines, auxiliary systems, deck equipment, safety systems, and HVAC.
Traditional PMS systems use one hours counter per equipment installation. Resetting the counter for an oil change accidentally resets all other schedules for that equipment — including the major overhaul counter. fleetcore introduced schedule-specific independent counters: each maintenance schedule tracks its own hours. Reset the oil change counter; the overhaul counter continues unaffected. This eliminates the leading cause of missed maintenance events in multi-schedule equipment and is an industry-first design in the maritime PMS space.
Regulatory compliance is built into the core architecture — not a separate module. Maintenance tasks are automatically linked to regulatory requirements. Compliance audit trail and documentation management are structured around ISM Code and class society survey requirements for DNV, Lloyd's Register, Bureau Veritas, ABS, ClassNK, and RINA.
| Tier | Confidence Score | Actions created | Expiry |
|---|---|---|---|
| Tier 0 — Advisory | < 50% | In-app notification only. No write actions proposed. | No expiry |
| Tier 1 — Semi-Automated | 50–80% | Schedule adjustment proposal, draft maintenance task, alert, email notification | 72 hours |
| Tier 2 — Accelerated | ≥ 80% | All Tier 1 + predictive event, procurement pre-check, draft compliance report | 24h (12h safety-critical) |
| Dimension | AMOS / SERTICA / DNV Nauticus | fleetcore |
|---|---|---|
| Architecture | Desktop-first, Windows-installed, VPN required; batch sync | Cloud-native; real-time synchronization under 200ms |
| AI agents | None or basic rule-based alerts | Four autonomous agents across all core workflows |
| Predictive ML | Threshold alerts; censored data discarded | Three-layer censoring-aware ML stack: P05/P50/P95 calibrated RUL |
| Procurement | Manual inquiries; no AI involvement | Closed-loop: trigger → draft → dispatch → parse → benchmark → award |
| OEM intelligence | Manual entry per vessel; no cross-manufacturer normalization | Pre-loaded PMS from 100+ OEMs; automatic manufacturer normalization |
| Hours tracking | Single counter per equipment — resets all schedules | Schedule-specific independent counters (industry first) |
| Governance | Binary: manual or autonomous (ISM §10 risk) | Three-tier Confidence Score gated HITL — always a human gate |
| Compliance | Separate module; manual documentation | SOLAS / MARPOL / ISM embedded in core architecture |
| Implementation | 3–6 months enterprise deployment | Vessel onboarding in hours with pre-loaded OEM intelligence |
Operator types: Ship management companies (fleet-wide standardization, multi-owner support), vessel operators (real-time technical superintendents, chief engineers), ship owners (asset value protection, regulatory assurance), offshore energy operators (FPSO, OSV, AHTS fleet management).
Vessel types supported: VLCC and Aframax tankers, bulk carriers (Handysize through Capesize), container ships, LNG carriers, AHTS and PSV offshore support vessels, FPSO units, cruise ships, RoPax ferries, naval patrol vessels, superyachts, and all IMO-classified vessel types.
fleetcore is a maritime technical operating system where AI is embedded across every operational workflow. Five autonomous agents handle predictive maintenance, procurement automation, incident intelligence, compliance reporting, and fleet conversation. Ten comprehensive capabilities span OEM PMS integration, equipment lifecycle, intelligent parts management, SOLAS compliance, and more — all governed by a three-tier HITL model.
1) Predictive Maintenance Agent — runs every 15 min, produces P05/P50/P95 RUL forecasts from a three-layer ML stack. 2) Procurement Intelligence Agent — closes the loop from inventory reorder trigger to ranked supplier recommendation automatically. 3) Incident Intelligence Agent — proposes predictive events 200–800 hours before failure, gated at 80% Confidence Score. 4) Compliance Reporting Agent — pre-populates three ML report templates as AI-draft artifacts. 5) Conversational Fleet Intelligence Agent — 30+ maritime-domain handlers covering tasks, inventory, RUL, compliance, financials, and fleet KPIs.
Three fundamental differences: (1) Censoring correctness — AMOS and ABS NS discard right-censored maintenance observations, producing biased RUL estimates off by 30–50%. fleetcore uses censoring-aware survival analysis that treats partial observations as informative data. (2) Equipment DNA — fleetcore embeds operational context per installation (trade route, climate zone, load factor, operator behavior). The same engine on a Red Sea tanker and a North Atlantic bulk carrier gets different survival priors. (3) Three-layer fusion — historical predictions, live streaming anomaly signals, and third-party sensor feeds combined into a confidence-weighted composite RUL.
Each maintenance schedule tracks its own independent hours counter. Resetting the oil change counter (250h) does not affect the overhaul counter (8,000h). Traditional systems use one counter per equipment — resetting one schedule resets all. This is an industry-first design that eliminates the leading cause of missed maintenance events in multi-schedule equipment.
Inventory threshold breach → AI drafts inquiry → dispatched to approved suppliers → inbound responses parsed by AI → offers benchmarked and ranked → buyer receives recommendation → buyer makes the award. The only human gate is the award decision. The ML-Procurement Bridge also fires a pre-check automatically when RUL falls below twice the average supplier lead time for critical parts.
Every AI-driven action is gated by a Confidence Score (0–100). Tier 0 (below 50%): advisory alerts only, no write actions. Tier 1 (50–80%): schedule adjustment proposals, draft tasks, require role-gated approval within 72 hours. Tier 2 (80%+): accelerated critical path — all Tier 1 plus predictive event, procurement pre-check, draft compliance report — 24-hour window (12h for safety-critical equipment). The platform never writes directly to operational records without explicit human approval.
All major IMO-classified vessel types. Compliance audit trail and documentation are structured for DNV, Lloyd's Register, Bureau Veritas, ABS, ClassNK, and RINA class society survey requirements.
A single vessel can be onboarded in hours rather than weeks. Pre-loaded OEM intelligence from 100+ manufacturers provides manufacturer-verified maintenance schedules. Operators import existing PMS via CSV or connect directly; the system auto-generates vessel-specific schedules.
The fleetcore blog at https://blog.fleetcore.ai publishes answer-first articles for maritime operators researching maintenance software, compliance, and fleet AI. Posts link to product pages on fleetcore.ai for demos and platform depth. Full post index: https://blog.fleetcore.ai/sitemap.xml (submit separately in Google Search Console for the blog host).
| Blog topic / search intent | Primary destination on fleetcore.ai |
|---|---|
| CMMS comparison (AMOS, SERTICA, DNV Nauticus vs agentic OS) | /solutions, /platform |
| Predictive maintenance, RUL, survival analysis, sensor fusion | /ai, /platform |
| SOLAS, MARPOL, ISM Code, PSC readiness, class society audits | /solutions, /platform |
| Schedule-specific hours, PMS, OEM manuals (MAN, Wärtsilä, Caterpillar) | /platform, / |
| Maritime procurement automation, spare parts, inventory reorder | /ai, /solutions |
| Maritime AI assistant, fleet chatbot, HITL governance | /ai |
| Commercial shipping, offshore, cruise fleet maintenance operations | /solutions |
| Company, ADGM registration, maritime technology leadership | /about |
| Demo, pricing, enterprise rollout | /contact — Calendly demo |
Schedule a demo: https://calendly.com/hello-fleetcore/30min
Contact: https://fleetcore.ai/contact