Reruns the demand forecast against actual sales history and flags what has moved, so the planner spends time deciding, not recalculating.
The forecast gets rerun when someone remembers to, which usually means it's already out of date.
The planner reviews exceptions, not the whole forecast, every time.
Actual sales history is pulled for every item on a rolling basis.
The forecast model is recalculated against the latest actuals.
Seasonal pattern is weighted apart from one-off spikes and dips.
Items where actual demand has meaningfully diverged land in a ranked queue.
The planner accepts, overrides, or investigates each flagged item.
It's priced per engagement, based on SKU count and forecast frequency, plus the Copilot Credits it consumes at runtime. We size both before you commit.
New products with thin history, or items with erratic demand, get flagged as low-confidence rather than forecast with false precision.
Yes, it works from the sales history and item setup already in Business Central or F&O, it doesn't require a separate planning system.
Most demand planning agents go live within two to four weeks, since the model needs a look at enough sales history to be useful.
This agent is about the forecast itself; the replenishment agent executes purchase orders off it. They're built to run together, but either can run alone.
Tell us how demand planning works today. We will tell you honestly whether this is worth automating yet.
Talk through whether this fits your process as it actually runs.
Four minutes, no email required.