The shipping industry's pursuit of a single universal AI solver for fleet planning is a myth that should be abandoned, according to Natalia Liashenko, CEO of Marine Solver, in an exclusive opinion for Splash247. Liashenko argues that while the industry has progressed from data collection to analysis, the next step—exploring the decision space for fleet deployment—requires not a one-size-fits-all automation but a combination of specialized mathematical models and human-led decision-making.
The Myth of a Universal Solver
Liashenko contends that the idea of a universal, fully automatic fleet planning solver is a myth. Although many planning problems produce the same output—a vessel itinerary or sequence of voyages—the underlying mathematics differ fundamentally. "Identical inputs and outputs do not mean identical mathematics," she writes. A model does not describe data; it describes operational reality, and that reality varies with each business objective.
Four Distinct Planning Problems
To illustrate the diversity of mathematical structures required, Liashenko outlines four distinct planning scenarios:
| Planning Scenario | Objective | Mathematical Complexity | Human Role |
|---|---|---|---|
| Spot voyage | Maximize immediate returns while valuing future positioning | Simple routing but dynamic trade-offs | Set risk tolerance and negotiate terms |
| COA (Contract of Affreightment) | Multi-voyage scheduling within rigid contractual windows | Complex optimization of sequences and buffers | Manage contract compliance and exceptions |
| Trader's fleet | Manage entire corporate cargo portfolio plus open-market arbitrage | Portfolio optimization across multiple vessels and markets | Balance strategic vs. opportunistic moves |
| Part cargo planning | Stowage logic, volume-to-weight ratios, cargo compatibility | Combinatorial puzzle beyond routing | Physical cargo expertise and stowage decisions |
"Reducing commercial fleet planning to mere routing subtly replaces a highly complex operational reality with a simple problem of geometry," Liashenko warns.
The Human Element
Even with the correct mathematical model, Liashenko emphasizes that fleet planning cannot be fully automated. Given the same data, two operators may deliberately choose different strategies. One may pursue more aggressive commercial opportunities close to laycan boundaries, expecting minor deviations can be negotiated. Another may maintain larger time buffers to protect the entire downstream voyage chain against disruptions. One may extend vessel employment far into the future, accepting market uncertainty; another may release the vessel earlier for stronger opportunities. These are rational but different choices—no algorithm can replace that judgment.
Implications for Shippers and Operators
For freight forwarders, logistics managers, and ocean carriers, Liashenko's argument signals that digital transformation should not aim for a single AI 'black box' that plans everything. Instead, the next evolution of digital shipping is an architecture that combines specialized mathematical models with transparent, human-led decision-making. Shippers and operators should seek tools that allow them to understand the model's logic and override automated suggestions when operational reality demands it. The key takeaway: "Real intelligence does not begin with computation. It begins with choosing—and building—the right model."
Sources:
