Smart logistics with a pricing engine and native AI.
When Aquiii started to grow beyond its first area, last mile became the bottleneck. The options: sign a single carrier (rigid and expensive), build an own fleet (capital-intensive), or build a quoting layer that orchestrates several carriers by cost, time and zone. We chose the third. But the real problem was not integrating carriers: it was protecting the margin without the operator doing manual sums on every order.
A Skydropx integration in production behind our own wrapper. The retail margin is applied server-side, so no customer ever sees internal quotes. An MCP server, safa-shipping, exposes quoting, label creation, tracking and carrier listing as tools, so the team and the Maaia agents work against the same source of truth. And a dedicated ops sub-agent inside Aquiii with a clear scope: logistics, pricing, Express.
When an internal service grows into a public product, most of the work is not features: it is discipline. Discipline to keep the margin server-side and out of the frontend. Discipline to version the MCP so a new agent does not break the operator's flow. Discipline to separate the retail rate (what the customer pays) from the carrier rate (what we pay). Done right once, that separation saves months of bugs.
If your business model depends on a margin between a carrier rate and a retail rate (logistics, marketplaces, brokerage, commission), we run that logic in production every day. And if you want AI agents in your operation, the MCP pattern we use for safa-shipping carries straight over to your vertical.
safa-shipping MCP running as the standard layerIf what we built here resonates with what you need, write to us. We reply within 48 hours with a discovery brief.