The Last Manual Process in a Digital Fleet

Maritime incidents are multi-causal. No single factor explains a vessel fire or a cargo loss event. But as the Allianz Safety and Shipping Review (2024) and annual TT Club incident data consistently show, container shipping carries among the highest large-loss exposure of any cargo segment, and the trend in loss severity has moved in only one direction over the past decade.

The growing density of high-risk cargo is a significant driver. Lithium batteries, undeclared dangerous goods, and complex hazardous material mixes now feature prominently in cargo incident investigations globally. Precise container placement is no longer a productivity question alone. It is a safety-critical discipline.

Yet in many liner operations today, stowage planning remains the last major operational decision made through manual, experience-driven processes, in an environment where everything else, from network optimisation to predictive maintenance, has been systematically digitised.

The risk is structural. It does not trigger alerts. It doesn't appear on operational dashboards. It lives in the gap between the complexity of the vessel and the capacity of the planning process to match it.

Consider a controlled scenario: three experienced planners independently produce a stowage plan for the same vessel, the same cargo mix, and the same port rotation. All three plans comply with regulation. All three would be signed off without issue. Below the surface, the plans diverge materially: overstow ranging from 3.2% to 5.1%, stability margins from +8% to +15% over minimum, crane balance from balanced to highly imbalanced, and additional moves required ranging from 62 to 104.

This is not a failure of competence. It is a structural property of manual planning: outcomes are person-dependent, experience-dependent, and time-pressure-dependent. Without systematic benchmarking, no one in the organisation can know whether the submitted plan is the best achievable, or merely the most recently completed.

At fleet scale, even modest planning inefficiencies become material liabilities. Industry benchmarks from Drewry's Port Benchmarker place the cost of a single additional vessel-hour at berth in the range of $8,000 to $15,000, depending on port and vessel size.

For a fleet of 20 vessels operating 10-port rotations across six annual sailings per vessel, a conservative one-hour planning-driven delay per port call amounts to $9.6M annually, at the low end of the cost range, before accounting for downstream schedule recovery, fuel consumed in acceleration steaming, or reputational exposure from chronic port delays.

Beyond berth time, avoidable overstow represents a recurring, largely invisible drag on terminal and vessel productivity. And fuel penalties from suboptimal trim and stability configurations compound over the full voyage. These costs rarely surface individually. They are absorbed into operational variance, attributed to congestion or weather, and accepted as the baseline cost of doing business. They are not.

Modern enterprise operations demand that consequential decisions be measurable, repeatable, auditable, and continuously improvable. Manual stowage planning is structurally resistant to all four.

When outcomes vary by planner, terminal, region, and experience level, the organisation has no baseline against which to measure performance. There is no audit trail of trade-offs considered or constraints evaluated. And there is no institutional learning: when a senior planner retires, their expertise walks out with them.

The question being asked at board level is increasingly direct: how do we know our plans represent the best available decision, consistently, across every vessel, every rotation, every year?

This is not a novel challenge. Aviation reached a structurally identical point when load planning complexity exceeded what operations teams could reliably manage manually. Energy trading faced it when derivatives portfolio risk could no longer be held in human working memory. Financial services institutionalised algorithmic decision support not to replace traders, but because the interaction variables exceeded cognitive bandwidth.

In each case, the response was the same: institutionalise the decision logic. Build systems that can evaluate vastly more configurations, enforce constraints with perfect consistency, and produce outcomes that are documented, benchmarked, and continuously refined.

Container shipping is at that same inflection point now. The organisations that recognise it first will define the operational standard for the next decade.

SONATA is an AI-powered stowage planning copilot built specifically for the operational complexity of modern liner fleets. It is not a compliance checker. It is not a visualisation layer. It is an optimisation engine, designed to do what manual planning cannot: evaluate the full solution space, enforce every constraint simultaneously, and surface the best achievable plan for every sailing.

SONATA operates as a force multiplier for the planning team: planners retain control and apply their expertise, while SONATA handles the combinatorial complexity that exceeds human cognitive bandwidth.

Every rotation carries hidden cost from planning variability. Intelligent planning makes it visible, and recoverable.