Your property’s schema is the single set of fields every customer’s data eventually maps into — set it up once on the Definition page.

Uploading a sample

Click Analyse and pick a representative CSV or XLSX file (yours, or a typical customer’s). The app samples the header row and a few data rows, then asks AI to propose a schema from them.

Core fields vs. custom fields

• Core fields are a fixed, built-in set (reference, origin/destination address pieces, scheduled date/time, quantity, weight, description, customer name, status, and a few more) that AI matches your columns against when there’s a strong semantic fit.
• Custom fields are anything that doesn’t fit a core field — AI invents a field for it instead of discarding it. Most properties end up with a few of these particular to their own business.
• Unmapped columns are ones AI couldn’t confidently place anywhere; you can still address these through chat.

Refining with chat

The chat panel below the proposed schema lets you correct anything — rename a field, remap a column, remove a false match, ask it to reconsider an unmapped column. Every change replaces the whole schema, so you can ask for several changes across several messages without losing earlier ones.

Confirming

Once it looks right, click Confirm & Save Schema. Customers can only be mapped (Customer Field Mapping) once your property’s schema is confirmed.

The Export tab

Alongside Core fields / Custom fields / Unmapped, the Export tab controls what the unified Orders table and every export format actually show — field order, which fields are included, and (if you turn it on) splitting each order into a header row plus one row per item by quantity.