The odd stuff is the whole problem.

The problem we kept hearing described the same way

In warehouse after warehouse, the knowledge of how to pack the difficult items lives in one person's head. They know which box takes the awkward bracket, how much void fill a ceramic piece needs before a cross-country lane, when a crate beats a carton. None of it is written down.

Then they leave. The queue does not. What is left is a team guessing, a damage rate that climbs quietly, and a materials spend nobody can explain.

We did not arrive at that from a market report. It came out of conversations with operators — re-commerce and refurb floors, a live-event production company, a specialty crating and freight business — who each described the same pattern without being prompted, and who all confirmed the same thing: there is reasoning software for routing, for rating, for warehouse management, and effectively none at the point where somebody decides what to actually put the item in.

Two roads, and we are only on one of them

If you ship the same case a thousand times a day, this is not for you. That problem is standardized, thin-margin, and already solved by people who solved it years ago.

Custom and variable logistics is the territory: items that differ every day, arrive without original packaging, and have no template to fall back on. That is where the tribal knowledge sits, where the damage claims come from, and where an engine that can reason about a specific item and a specific route is worth something.

How the engine is built to be trusted

An operator will not use a tool that is confidently wrong about a shipment they are responsible for. So the Materials Advisor is built to three rules, and they are constraints on the engineering rather than claims in the marketing:

• Deterministic, not generative. Sizing runs against a live catalog of real boxes. The same inputs return the same plan every time — no drift between one run and the next.
• It shows its reasoning. Every pick carries the logic behind it and the packing steps that follow from it. If you disagree with the output, you can see exactly where it decided.
• It halts rather than guesses. When the engine cannot answer within its catalog and its rules, it says so. A confident wrong answer costs an operator a shipment; an honest "not this one" costs them thirty seconds.

The same principle governs this site. The packing guides give general, clearly-caveated guidance — ranges and defaults — and never state a specific box size or material quantity as the answer for your item. That precision only exists inside the tool, where it is computed rather than written down.

Where we are

Early, and working directly with a small number of design partners rather than selling broadly. The free tool is genuinely free and fully featured — metered by volume, never crippled by feature — because an operator who cannot get real value from it on day one has no reason to tell a peer about it.