The reorder that files itself.
A logistics company's ops team spent hours a day on routine reordering — same suppliers, same math, endless clicking. We automated the routine and kept humans on the judgment calls.
The problem
Reordering stock was predictable ninety percent of the time — check levels, check lead times, place the usual order. But the ten percent that wasn't predictable (supplier delays, price spikes, demand swings) meant a person had to babysit all of it. The routine work was holding the judgment work hostage.
What we built
An agent wired into their ERP that watches inventory, forecasts from actual order history, and places routine orders within rules the ops team wrote themselves — spend caps, approved suppliers, lead-time buffers. Anything outside the rules stops and asks a human, with its reasoning shown.
Trust was earned, not assumed: it ran in shadow mode for a month, drafting orders next to the team's real ones, until the drafts matched what the humans would have done. Only then did it go live — with an audit trail on every decision.
The outcome
Six months of hands-off routine ordering, roughly fifteen hours a week returned to the ops team, and — the part they care about — zero stockouts on automated SKUs. Humans handle exceptions and strategy now, which is what they were hired for.
Support copilot for a fintech
What's eating
your team's hours?
If it's predictable ninety percent of the time, it's probably automatable — safely. Let's find out with a two-week audit.