AI & Automation
PrismaWMS has six engines that look at your live warehouse data and suggest improvements: which pick face to refill, which item to move to a better bin, what to reorder, which pick route is shortest, who has too much work, and what looks wrong. They are rules and statistics that run in PrismaWMS itself — no data is sent to an outside AI service for them. Every suggestion explains why, and nothing changes until a person accepts it. The WMS Copilot is different: it is a chat assistant that uses an outside language model to answer questions.
How suggestions are made
The engines
| Engine | What it looks at | How it decides | Screen |
|---|---|---|---|
| Replenishment Planner | pick faces, their min / max, reserve stock, demand | the min/max rule + a reorder point with safety stock | Command Center, Replenishment |
| Slotting Optimizer | 90 days of picks per item and bin | pick frequency + ABC classes → the best bin nearer to dispatch | Slotting |
| Demand Forecaster | shipped quantities per day | simple exponential smoothing + moving average; accuracy measured by back-testing | Demand Forecast |
| Pick-Path Optimizer | open pick tasks of each wave | nearest-neighbour route + 2-opt improvement | Pick-Path |
| Labour Balancer | open tasks per worker, task timestamps | workers far above the crew average | Labour |
| Anomaly Detector | stock, expiry, receipts, counts, movements, order deadlines | rules + statistical outlier tests | Anomaly Detection |
The Model Center shows each engine, when it last ran, and how well its suggestions perform — measured from what users actually accepted or dismissed.
The confidence shown on a suggestion is a fixed level set by the rule that produced it (a certain rule — for example "this pick face has 0 available" — is 100 %, other rules are high, medium or low). It is not a prediction accuracy. Quality numbers such as acceptance rate or forecast error appear in the Model Center only once there is enough real data; until then it says Not enough data.
Which screen for what?
| I want to … | Screen |
|---|---|
| see all suggestions in one list and act on them | AI Command Center |
| ask a question in plain language | WMS Copilot |
| put fast movers in the best bins | Slotting Optimizer |
| see expected demand and what to reorder | Demand Forecast |
| give pickers a shorter route | Pick-Path Optimizer |
| balance work between people | Labour Optimizer |
| find and investigate things that look wrong | Anomaly Detection |
| switch engines on/off, set thresholds, see the decision log | Model Center |