One inbox for 70 marketplace channels

To answer one customer, a support rep opened that marketplace's back office. Then the next. Ten people were working across dozens of separate panels, none of them talking to each other, none of them showing how long a message had been waiting.

Before building anything we classified 957 tickets into 17 types. Returns were the largest bucket by a distance: one line alone accounted for 48.9% of everything, around 120 tickets a week. That is what put returns at the top of the roadmap, rather than intuition.

Mailbox centralises all of it, with an SLA on every row and the queue sorted by urgency by default, so the most pressing work is the first thing on screen without anyone sorting anything. The AI handles returns end to end, including generating the GLS return label through their API and creating the RMA in Odoo. After automating returns it covered around 70% of total cases.

It was built as an internal tool, and it went into real use before anyone asked whether it could be sold. The interesting part is who else has this problem: BigBuy's own B2B customers sell across the same marketplaces and answer their own customers the same way. A tool validated by real users inside the building, with the target market already in the customer base.