Credit: Shanghai International Port

Shanghai Port Unveils Four Layer AI System That Could Rewrite Terminal Operations

Estimated reading time: 4 minutes

Shanghai International Port Group has introduced a new artificial intelligence product system designed to connect port data, operational models, autonomous agents and terminal applications within a single technology framework.

The system was presented on July 19 during the 2026 World Artificial Intelligence Conference in Shanghai, where more than 100 representatives from ports, research institutions, technology companies and the wider maritime supply chain gathered to examine how artificial intelligence could reshape global port logistics.

The forum, held at the Shanghai World Expo Center, brought together senior figures including SIPG Chairman Yu Fulin, Tongji University Party Secretary and academician Zheng Qinghua, China Institute of Navigation Chairman Yan Xinping and China Ports and Harbours Association Chairman Gu Jinshan.

Millions of terminal records become AI training material

At the centre of the announcement was Fenghuolun, a specialised port data set built from tens of millions of terminal operating records, equipment status reports and other information collected from real port environments.

Port operating data is often spread across disconnected systems, equipment logs and departmental databases. SIPG said Fenghuolun converts that fragmented information into a structured knowledge base that can be used to train artificial intelligence systems for port related tasks.

In simple terms, artificial intelligence cannot learn how a terminal works without realistic examples. Fenghuolun is intended to provide those examples, functioning like a large collection of test questions and operational scenarios drawn directly from terminal activity.

Dual model platform targets port decision making

SIPG also presented the Lotus Seat capability platform, an artificial intelligence foundation developed specifically for ports.

The platform combines large language models with smaller specialised models. Large models can interpret complex instructions and broad operating information, while smaller models can focus on specific port tasks, equipment or processes.

An agent management platform supports the design, development, deployment and optimisation of artificial intelligence agents throughout their operational life cycle.

The system also includes a skills centre for storing port expertise and a connection centre designed to link artificial intelligence agents with existing business and terminal systems.

The objective is to create an open port technology environment in which different agents can exchange information, support decisions and carry out assigned tasks.

Natural language commands enter terminal operations

At the application level, SIPG subsidiary Nezha Smart Technology unveiled AITOS, which the company described as China’s first next generation terminal operating system built on an artificial intelligence native architecture.

AITOS contains groups of artificial intelligence agents covering terminal planning, equipment scheduling and operational execution.

It also allows frontline terminal employees to issue instructions using natural language rather than navigating multiple menus or entering commands through traditional software interfaces.

That could change how operators interact with terminal systems. Instead of adjusting several planning screens, an employee could describe the required action and allow the system to interpret the instruction, assess operational data and coordinate the relevant agents.

AITOS is designed around a process in which the terminal system detects operational information, artificial intelligence supports decision making and digital agents carry out the approved action.

The crucial question is whether natural language systems can operate reliably in environments where seconds matter and equipment movements must be precisely coordinated. SIPG said the architecture is intended to make production tasks more direct and easier for frontline staff.

White paper sets out global deployment framework

SIPG, Nezha Smart Technology, Shanghai Jiao Tong University and Tongji University also released a white paper on the development of port artificial intelligence agents.

The document provides a technical roadmap covering system design, data preparation, model development, agent management and practical implementation.

Its intended audience includes port operators and technology providers seeking to introduce similar systems at other terminals.

By publishing the framework, the organisations aim to turn individual research projects into technology solutions that can be copied, adapted and deployed across different port environments.

Shanghai targets lower cost and operational risk

SIPG described the product system as a four part structure combining data, models, agents and applications.

The port group said the technology could help terminals increase handling capacity and competitiveness while reducing operating costs and business risks.

Shanghai Port, the world’s largest container port by annual throughput, has been expanding automation and digitalisation across its terminal network. The new system signals a shift beyond conventional automated equipment towards terminals in which artificial intelligence increasingly supports planning, coordination and execution.

Speakers at the forum also discussed quantum computing, autonomous driving in port environments, international terminal standards and the development of digital world models for ports.

Participants called for greater international cooperation, broader collaboration between ports and technology companies and closer coordination across the maritime supply chain as artificial intelligence moves from experimental projects into daily terminal operations.


DISCLAIMER: “Breakbulk.News publishes editorial content, including news, features and press releases supplied by third‑party companies, institutions and PR agencies. Third parties who submit material to us are solely responsible for ensuring that all text, images, logos and other content they provide are accurate and that they hold all necessary rights, licences and permissions for news use. By submitting content to Breakbulk.News, contributors represent and warrant that their material does not infringe the rights (including copyright and related rights) of any third party and agree to indemnify Breakbulk.News respecting any claims arising from their submissions. human-edited, AI-assist. If you believe any content on our site infringes your rights, please contact us at info@breakbulk.news with full details and we will investigate promptly. Breakbulk.News is a Trademark of Breakbulk News & Media B.V. in The Netherlands.”

Get the Breakbulk.News app Google Play
×