All earlier work ↗
Historical engagement record
About this record
This is an earlier engagement or reference design, preserved as authored. Reported metrics and delivery status have not been independently revalidated for this preview; source dates are not established here. Modeled figures and design targets are not realized results. This record does not substantiate outcomes for the proposed product.
Explore Tensile today ↗
Most plant-analytics projects die on the data: placeholder job codes and inconsistent entries that quietly corrupt every metric downstream. So we built the unglamorous core first: an engine that resolves the plant’s messy entries onto a composite business key, then a warehouse and dashboards on top. The result moves the plant from monthly manual sheets toward shift-level decisions on furnace utilization, pack efficiency and downtime. It’s in UAT now; the efficiency-gain targets are still being proven.
The challenge
Daily plant data lived in spreadsheets with no shift-level view of efficiency, and the raw data was messy: placeholder and inconsistent job entries that defeated naive ingestion. The plant needed engineered efficiency metrics it could act on by shift, not a monthly retrospective.
The approach
A Python ingestion path (Sheets API) into a Postgres warehouse with row-level security, feeding web dashboards.
A custom dirty-data job-mapping engine that resolves placeholder and inconsistent entries via a composite business key, so KPIs sit on clean joins.
Engineered efficiency KPIs (furnace capacity utilization, glass pack efficiency), defect/downtime tracking at shift granularity, and a shift-approval app.
Artifacts and handoff
An ingestion → warehouse → dashboard stack with the job-mapping engine, shift-level defect/downtime tracking and a shift-approval app, currently in UAT.
Reported measures and design targets
In UAT: piloting with plant operations
Shift-level granularity: defect and downtime tracked per shift, not per month
Engineered KPIs: furnace capacity utilization, glass pack efficiency
Dirty-data engine: composite-key job-mapping resolves placeholder/garbled entries
Explore the failure mechanisms, the tests they need and the evidence that establishes the customer outcome.
AI reliability, from production to customer outcome.