How Do We Take a 138-Year-Old Distillery from Fixing Problems to Predicting Them?
Nearly 40% of repairs at the Girvan distillery were emergency responses — faults caught only after alarms sounded, leading to more downtime and stifled output . The goal was to shift from reactive maintenance to a predictive model that could get ahead of problems before they caused disruption.
Connecting Data to Catch Faults Early
Nexus Black plugged into every system at the distillery — including piping and instrumentation diagrams — and combined live sensor data with historic asset records . The system can flag a potential problem well before a sensor would trigger an alarm, and analyse how a specific fault will impact the entire distillery.
Putting the Right Tools in Engineers' Hands
Working on the factory floor revealed a simple truth: a text-based chatbot is no use to an engineer wearing protective gloves . The solution was built voice-first — letting technicians diagnose faults, report fixes, and capture sensor data just by talking. Those insights feed directly into accurate work orders, and everything learned goes back into the system to build future resilience .
£8.4 Million in Estimated Annual Savings
Once in business-as-usual mode, William Grant & Sons estimates savings of £8.4 million a year at Girvan . That figure is driven by higher first-time fix rates, less downtime, and increased overall output — proof that deploying AI at speed and scale delivers real, measurable value .