Resources
Guides & one-pagers
Detailed briefs on how Jemba delivers on real production lines — the method, the numbers, and how each result was validated. Request any one and we will send it over.
Overview · one-pager
Industrial Machine Learning — the one-pager
Anomaly detection and process optimisation on the data your plant already produces. Two weeks to an answer, on a plant-level scope, with nothing to install on your network.
273 h downtime anticipated+9.5 pts yield2 weeks to an answer
Client case #1 · Anomaly detection
273 hours of downtime, anticipated from a log you already keep
Aerospace & defence supplier, 4 machines, regulated environment. From the run/stop event log alone — no new sensors, no intervention on the machines — Jemba anticipated 45% of the line’s total downtime, validated over a 12-month rolling backtest.
273 h · 45% of downtime0 new sensors12-month backtest
Client case #2 · Process optimisation
+9.5 points of yield, found in runs the line had already made
Rubber & polymer supplier, textile bobbin line. One export of 977 columns screened down to 10 levers; the recommended operating zone averages 78.1% yield vs 68.6% across the full 336-run history — measured on 33 past runs, not predicted.
+9.5 pts material yield977 → 10 leversno client man-days
