Jan 22, 2026
Continuous-process optimization for a glass furnace
Every shift ran the furnace slightly differently, and by the time monthly reports surfaced the drift, the margin was already gone.
Yield variability
↓↓
Duration
14 weeks
Team
5 people
Challenge
Operational drift between shifts was hard to detect in real time, and small deviations compounded into cost.
Solution
We built a shift-aware monitoring layer, defined control bands with plant engineers, and added AI assistance for anomaly detection rather than full automation.
Outcome
Variability tightened, shift handoffs improved and the plant recovered margin from efficiency that was previously leaking unseen.
What was happening
Shifts were running slightly differently. Nobody was wrong, but nobody was aligned — and the furnace paid for it.
By the time monthly reports surfaced the drift, the cost was already gone.
What changed
Engineers co-designed the control bands with us so the monitoring layer reflected the real process, not a theoretical model.
AI flagged anomalies early; humans decided how to respond.
- Shift-aware monitoring layer
- Control bands co-owned by plant engineers
- Anomaly detection feeding human judgment
AI Transformation, Process Optimization & Cost Efficiency