Kiln Scheduling with Real-Time Energy Pricing
The Challenge
Manual scheduling across 12 kilns had no connection to live energy price feeds. Decisions were made on static plans — causing significant over-spend during peak-rate grid windows with no ability to re-sequence in response to intraday fluctuations or kiln anomalies.
What We Built
An agentic workflow integrates live energy pricing APIs, kiln health telemetry and production demand signals into a single autonomous system. Agents re-order kiln batches during price spikes, flag anomalies in real time and pass full operational context across shifts automatically — no operator intervention required.
Operational Impact
18% reduction in energy cost per tonne sustained in production across all 12 kilns. 94% schedule adherence achieved. Deployed into existing plant control infrastructure in 6 weeks from kickoff.
Technology Used
Python
LangChain
FastAPI
Azure
PostgreSQL
