Presenting Project c|AI|sson at AIMCT sparked a level of interest that highlighted a growing reality across the infrastructure sector: artificial intelligence has the potential to unlock value from data that already exists within asset management programmes.
While AI is often associated with high-profile applications and emerging technologies, applying it to the challenge of managing offshore and marine infrastructure was not necessarily what delegates expected. However, once attendees saw the model applied to a real asset, the potential quickly became clear.
“Some people were surprised to see AI applied to a problem like this. It is not where most people would expect to find it — caissons, and inspection data that is already there. But once they saw it working on a real asset, the questions came quickly. And some of those conversations continued at our Tech Day the next day.”
The response reinforced a key message: organisations already possess valuable inspection data, but there is often an opportunity to extract significantly more insight from it.
Turning Insight Into Better Decisions
Inspection programmes generate large volumes of information about asset condition. Traditionally, this data is used to understand the state of an asset at a single point in time and to support immediate maintenance decisions. However, when analysed over longer periods, the same datasets can provide much deeper insights into degradation patterns and future risk.
Project c|AI|sson has been developed to help clients move beyond simple condition reporting and towards predictive asset management. By understanding where deterioration is occurring, how quickly it is progressing, and which areas remain stable, operators can make more informed decisions about maintenance, inspection frequency, and investment priorities.
“Inspection data can do much more than report the condition on the day it was collected. It can show an operator where the loss is concentrating, so they look at five metres instead of thirty. It can show how fast it is moving, so they know which year to act. And it can show which sections are safe to leave, which is just as useful and much harder to argue without evidence.”
This ability to transform historical inspection records into actionable intelligence has significant implications. More targeted interventions can reduce costs, focus resources on areas of genuine concern, and provide greater confidence in asset management decisions.
Extending the Value of Every Inspection
At the heart of Project c|AI|sson is a simple but powerful concept: every inspection should continue delivering value long after the survey report has been issued.
The model uses historical survey data to forecast future wall thickness on a cell-by-cell basis, helping asset owners understand how assets are likely to perform in the years ahead. During development, the team tested the model’s predictive capability by withholding one survey entirely from the training dataset. The results demonstrated a high level of accuracy, providing confidence in the model’s ability to forecast future conditions.
“Project c|AI|sson uses the surveys already carried out on an asset and forecasts the wall thickness cell by cell for future years. We tested it by keeping one survey away from the model. It never saw that survey, and the forecast matched 96%. Once the model is trained on one caisson, it can be used on others of the same type, calibrated with a single survey. So, the value of an inspection does not stop when the report is finished.”
The ability to apply a trained model across similar assets, requiring only minimal calibration, also presents an opportunity to scale these benefits across portfolios of infrastructure. Rather than starting from scratch with each asset, operators can build on existing knowledge and accelerate the adoption of predictive maintenance strategies.
Looking ahead
The conversations generated at AIMCT continued during our Technology Day in Doha, and demonstrated a strong appetite for practical AI solutions that solve real industry challenges. Rather than replacing existing inspection programmes, Project c|AI|sson enhances the value of the data already being collected, helping operators make smarter, evidence-based decisions about their assets.
As the industry continues to seek more efficient and sustainable approaches to asset management, the combination of engineering expertise, inspection data, and artificial intelligence has the potential to deliver a new level of insight, confidence, and value.