Turn a Buyer's Spreadsheet Into Clean Orders

Questions this answersQuestions this answersfaq

how do I turn a buyer's spreadsheet into orders without keying it in?how do I turn a buyer's spreadsheet into orders without keying it in?how do I map buyer lines onto catalog SKUs?

Most of it is matching, not AI. A script compares each line of their sheet against your catalog, plus a short table of that buyer's shorthand, and drops every line into one of three piles with a number for how sure it is: matched, needs review, or nothing in your catalog is close. A rep approves the draft before any order exists.Most of it is matching, not AI. A script compares each line of their sheet against your catalog, plus a short table of that buyer's shorthand, and drops every line into one of three piles with a number for how sure it is: matched, needs review, or nothing in your catalog is close. A rep approves the draft before any order exists.Fuzzy match buyer lines against the catalog deterministically, blending token overlap with string similarity and penalising a stated size that disagrees, then bucket by threshold into match, review and unknown with a per-line confidence. A runner-up within a small gap of the winner forces review even at a high score. The model, if used at all, sees review rows only, and a human approves before anything is created.

can AI understand a buyer who writes BD 3.5 and 2 cs?can AI understand a buyer who writes BD 3.5 and 2 cs?how are aliases and case quantities resolved?

The shorthand comes from a table you build out of past orders, not from the AI guessing: BD means Blue Dream, eighths means 3.5g. And 2 cs is not a number until you know the case size of the product that line matched, so the case size lives in your catalog export and the multiplication happens in code. Where the case size is missing, the line stops for a person.The shorthand comes from a table you build out of past orders, not from the AI guessing: BD means Blue Dream, eighths means 3.5g. And 2 cs is not a number until you know the case size of the product that line matched, so the case size lives in your catalog export and the multiplication happens in code. Where the case size is missing, the line stops for a person.Alias expansion runs from an operator-maintained table seeded from past orders, before scoring. Case quantities multiply by the catalog's case_size field, and a missing case size routes the line to review rather than inventing a number. Never hand the unit arithmetic to the model: it answers fluently and is right often enough that people stop checking.