Background

The client, a global leader in cement and mining equipment, was shifting from a hardware-centric model to a services-and-insights ecosystem, a platform where plants could benchmark performance and receive asset recommendations. However, this ambition relied on high-fidelity data that could explain why an asset behaved a certain way, not just that it did.

While the platform could already ingest automated sensor data, sensors cannot capture what an operator sees, decides, or executes. That critical layer of context such as incidents, observations, and maintenance actions, lived in paper logbooks or stayed in people’s heads.

Service technicians work in demanding, high-mobility settings and consequently, manual logging was slow, inconsistent, and unstructured, resulting in data that was nearly unusable for downstream analysis and highly vulnerable to loss during shift changes.

This led to the insight that reorganized the entire project: machine data provides the baseline, but only production personnel possess the context that explains it. Our mandate was to capture tacit operator knowledge at the exact moment and context of creation, in a structured format the platform could immediately compute. The solution had to be effortless for a user with their hands full, yet clean enough for backend data science.