FloorLab

FloorLab lets a warehouse owner try a better way to run their floor before moving a single shelf. Changing a layout or how a crew picks is expensive to get wrong, and small warehouses have no simulation team and often no warehouse system to connect. The owner describes the floor by answering questions, sending a photo or sketch, drawing it, loading an order file, or using their own words. FloorLab builds a simulated day with every item, cart and truck, and checks it against numbers the owner already knows, such as orders a day. If it can't match them, it shows the closest it got and labels results as unchecked. The owner picks a goal: the whole process, shorter walks, fewer labour hours, faster turnaround, lower cost per order, or the same work with fewer people. Meta's Muse then proposes changes over several rounds, sees how each one did, and tries again. A live chart adds a dot for each idea that passes, an x for each that fails, and a dashed line for the owner's aim. Every idea is tried on 8 trial days. The best go on to 20 test days held back from Muse. An idea must win on each kind of day (normal, busy, someone off sick, Monday backlog), stay within limits on late orders and turnaround, and pay back its one-off work. When nothing passes, FloorLab says so and shows what the best try really did and why it failed. The owner then plays their way and the new way side by side on the same day, and gets a plan: what changes, the one-off work, the payback, a move list, and a check to run two weeks later. On nine made-up example floors, recommended changes cut picker-hours 25-39% on the whole-process goal, each beating the current way on all 20 test days. All results are simulated.