Turning unlabelled rig sensor data into actionable intelligence – using AI to extract meaningful patterns from raw IoT signals without the need for costly, time-consuming labelling projects.  

From sensor signals to crew-activity models – combining domain expertise with AI and coding agents to identify and understand activity patterns across drilling operations.  

Keeping AI honest without labelled data – exploring a human-in-the-loop autoresearch approach that uses engineering knowledge to guide research, validate emerging patterns and avoid misleading machine-learning outputs.  

Making future labelling faster and cheaper – using AI-generated clusters to structure the data and create a foundation for targeted labelling and more scalable machine learning.  

Unlocking new operational insights from the rig – demonstrating how unlabelled sensor data can evolve into crew-activity intelligence and support better understanding of drilling performance.