Offshore marine and engineering group Seatrium requires proof of concept AI projects to demonstrate business value within 3 months. After this deadline, the project is either moved to the implementation stage or stopped.
According to Senior Vice President of Digital Transformation Lee Wey Lii, the goal is not to avoid failure, but to prevent AI projects from lingering indefinitely in the testing phase and becoming a “graveyard of pilots.”
A successful example is an AI application combined with OCR to cross-reference thousands of equipment codes with technical drawings. A process that used to take up to 50 person-days now takes only about 2 person-days.
Due to the specific nature of technical drawings where text positions are not fixed, previous OCR-only solutions did not achieve the desired accuracy. AI has significantly improved the efficiency of this process.
For CEO Chris Ong, the key performance indicator (KPI) for the AI program is not the number of experimental projects, but the number of projects actually put into operation. Projects that stop at the testing phase are not counted toward KPIs.
In late 2024, Seatrium established an AI Council consisting of 15 senior leaders, chaired by the CEO, to select and monitor AI projects from conception to practical implementation.
Seatrium stated that its AI transformation stems from a digitization strategy started in 2018, when the company established an initial digital transformation team of just 3 employees to build data infrastructure and digital platforms.
Another key project is building a digital twin for the entire shipyard, helping to simulate space, resources, and project progress. AI uses data from the digital twin to suggest spatial allocation plans when a new project arrives, after which humans evaluate and make the final decision.
Heavy AI tasks are processed on GPUs located right at the shipyard instead of the cloud to ensure performance and data security.
Currently, Seatrium has about 100 people, including employees and partners, participating in digital initiatives. The company collaborates with the Singapore Institute of Technology and the Institute of Technical Education to train human resources combining shipbuilding expertise and AI. Despite promoting AI, Seatrium stated that cybersecurity remains its top concern and all technologies must pass a safety review process before being put into operation.
📌 Seatrium is applying a very pragmatic approach to AI: every project must prove value within 3 months or be stopped. Instead of chasing the number of pilots, the company measures success by the number of projects put into actual operation. With applications such as reducing processing time from 50 person-days to 2 days and building a digital twin for the shipyard, Seatrium shows that AI is only considered successful when it creates clear business value, while still placing high priority on cybersecurity and data quality.
