The challenge
A third-party logistics provider was drowning in exceptions: mismatched counts, missing labels, delayed carrier pickups. Supervisors spent hours each day reconciling data across the WMS, TMS and email — and customers often found problems before the provider did.
What we did
- We unified event data from the WMS, TMS and carrier feeds into a single, model-ready stream.
- Exception agents classify each event, gather evidence across systems and either resolve it directly or route it with a recommended fix.
- A supervisor workspace shows every agent decision, with one-click approval for higher-risk actions.
- Rollout followed a site-by-site playbook so each new warehouse went live in days.
The outcome
The exception backlog fell by almost two-thirds, supervisors regained hours every week, and customer escalations dropped as issues were fixed proactively.
