LotLens — Quality Drift Investigator

Manufacturing quality teams need to connect a changing defect trend to the production lot and evidence behind it. LotLens turns timestamped inspection readings into an explainable drift alert, identifies the affected line, lot, and time window, and presents the supporting readings alongside synthetic procedure excerpts. The detector calculates its baseline and alert locally using a transparent sustained-change rule. When AMD model access is configured, an AMD-hosted open model turns the calculated finding and supplied procedure excerpts into a concise investigation brief with source-linked possible checks and visible uncertainty. The model does not calculate the alert or determine a root cause. A quality lead reviews the finding and decides what to do. LotLens also includes an inventory-impact workflow that maps an alerted lot to a synthetic SKU and order snapshot. It compares affected reservations and unallocated orders with unaffected stock, then presents a human-reviewed allocation suggestion. It does not change inventory or orders. The demo uses synthetic readings and synthetic procedure excerpts, including a factory-derived synthetic fixture that is not an independent production export. LotLens has no live factory or inventory connection and is not validated for industrial quality control.