From a 2-person workflow to a single review per order.
Automated kit assignment and warehouse handoff across VeraCore, Shopify, and the existing warehouse workflow for a subscription box brand. No core systems replaced.
- 2 → 1 person per order
- Manual entry errors ↓
- Faster warehouse handoff
A subscription box brand was running every order through two people: one assigning the right kit (based on trimester, sizing, and customer history), one routing it to the warehouse and coordinating the handoff via spreadsheet. Errors were creeping in. The workflow couldn't keep up with growth.
The problem in one sentence
Kit assignment was deterministic — there were rules — but the routing layer was a spreadsheet, and the spreadsheet was a person.
What we built
- Rule-based kit assignment. AI automation applies the brand's existing kit-selection rules (trimester, size, customer history) to every incoming Shopify order, automatically.
- Order routing. Routes the assigned kit through VeraCore and the WMS, attaching all the context the warehouse needs to pick and pack — without the spreadsheet handoff.
- Warehouse handoff. A consistent, repeatable handoff workflow replaced the ad-hoc daily spreadsheet. Same warehouse staff, same physical workflow — just a cleaner upstream feed.
- Single human review. Each order surfaces to one operator for approval, not two — and only when the rules surface ambiguity.
The existing warehouse workflow stayed exactly as it was. We layered automation onto Shopify → VeraCore → WMS without touching how the warehouse actually picks, packs, or ships.
The outcome
- Two-person workflow collapsed to one review per order
- Manual order-entry errors materially dropped
- Warehouse processing sped up with a more consistent upstream handoff
- Headcount reallocated to growth work, not order-routing toil
What we deliberately didn't do
- We didn't replace VeraCore or the WMS. Same systems, fewer human bottlenecks between them.
- We didn't try to remove the human review entirely. Subscription kit assignment has enough customer-context nuance that one approval per order is the right level of guardrail — automating to zero would have been false economy.
- We didn't build a CRM. The customer history that drives kit rules already lived in the existing systems; we just made the rules executable instead of typed.