Audience member observes that the problem — pick a set of five SKUs from available stock maximizing a score combining per-SKU signals (user preferences, recency, expiry, margin, color/roast diversity) and set-level constraints — is a textbook optimization-with-score-function problem. Nicolò agrees that's essentially what the system does: score each candidate and pick the top for each slot, with possible optimizations when slots are identical (collapse into one loop). Notes stakeholders didn't explicitly ask for ML, but the system is already being deprecated toward proper machine learning starting from more complex cases.