All solutions/Vertical·Manufacturing & Supply Chain

Returns processing

Inbound classification, disposition routing, reverse-logistics optimisation.

Shipped against
  • Big-box retailer
  • Off-price retailer
  • Specialty CPG
  • Electronics OEM aftermarket
  • Apparel marketplace

Anonymised — clients under NDA

+22%
margin recovered on returns
8 wks
to production MVP
5-person
squad with reverse-logistics ops

The context

Returns are a margin sink. Inbound classification is manual, disposition (resell, refurbish, recycle, scrap) is rule-of-thumb, and reverse-logistics routing is solved at the warehouse level with no view of the network.

Why it doesn't scale today

Most return management software focuses on the customer side — the RMA, the refund. The operations side, where the margin recovery actually lives, has been undertooled. Generic AI has the wrong sense of value: it optimises the wrong things.

What we ask in week one

  • iFor each of your product families, what is the actual margin recovery by disposition path today?
  • iiWhere can vision and document AI classify your inbound returns at line speed?
  • iiiHow do we route disposition against your live resale-channel demand and your current refurb capacity?
  • ivWhat does the override surface look like for your warehouse team — the people who'll catch the misclassifications?

What we build

Inbound classification by vision and document AI, dispositioning routed against live resale-channel demand and refurb capacity, reverse-logistics optimisation against the actual network. The warehouse team owns the override; the system improves recovery without taking decisions away.

Why we're the right squad

We have shipped returns operations for retailers and specialty CPG firms. The math behind disposition value is unforgiving — and we have done it before.

What you keepA system, not a prototype
  • Documentation

    Architecture, data contracts, deployment runbooks, eval criteria — written for the team that owns it next.

  • Infrastructure

    IaC, monitoring, rollback, on-call playbooks. Standard tooling, not bespoke.

  • Evals

    Eval harness + prompts + labelled fixtures + regression suite. Extends with your team.

  • Architect anchor

    A lead architect from our side stays in your Slack for at least 90 days post-handoff.

Want to scope this for your team?

We'll come back with a pod composition, a 2-week discovery sprint plan, and a price — measured against the KPI you care about, not against activity.

RelatedAdjacent solutions