Blog

AI evaluation notes from the field.

Published writing on production AI quality, governance evidence, model reliability, and the operational work behind trustworthy deployment.

Jun 1, 20261 min

[Case Study] - Asia F&B Group Cuts Document Processing Time by 65% with AI

An Asia-based multinational F&B group validated an Intelligent Document Processing System using Knotest before committing. A one-month A/B test slashed processing time by 65.3%, boosted throughput by 188.5%, and cut error rates by 69.6%. Operations manager Mr. Lee: “Once you see the data, the choice is obvious. This isn’t just faster — it’s more reliable.”

Read article

Jun 1, 20261 min

[Case Study] - Hong Kong Darts Bar Sees 53.7% Stockout Rate Reduction with AI system

A Hong Kong darts bar where beer makes up 70% of revenue switched from manual forecasting to AI after validating it with Knotest. An eight-week A/B test cut stockouts by 53.7%, reduced overstock by 37.1%, and halved planning time. Manager Chan: “Now we’re not just guessing. We know what’s coming.”

Read article