Logistics and freight broker
Transportation, 240 employees
Challenge
Two AI workflows shipped in the last year were quietly degrading. The vendor that built them had moved on, model accuracy had drifted, and incidents only surfaced when a customer complained.
What we did
Three-month managed ops engagement covering monitoring, weekly tuning, monthly delivery of one to two new workflows, and a one-page board report that ties AI to dollars and SLA performance.
Result
Both legacy workflows back inside accuracy targets within four weeks. Six new workflows shipped over the first six months, with a single monthly bill replacing the planned hire of an in-house ML engineer.
6
New workflows in 6 months
4 weeks
Accuracy recovery
1 ML hire
Headcount avoided
