AI for inventory in Business Central

Stockout risk

Stock on hand alone answers almost nothing. The question that matters is whether what you have, plus what is arriving, covers what is committed and what you expect to sell before the next receipt. That means combining item ledger entries, open purchase orders with expected receipt dates, open sales orders and consumption history.

• Items falling below safety stock before the next receipt is due
• Items with no inbound order at all, separated from items where the order simply arrives late
• Exposure on items already promised on open sales orders — the expensive category
• Supplier reliability, where receipts consistently arrive after the confirmed date

That last point deserves emphasis. Planning against confirmed dates when a supplier is habitually eleven days late produces a plan that fails quietly. The history is in Business Central; nobody has time to compute it per supplier. Our article on reorder point and safety stock covers the underlying calculation.

Slow-moving and dead stock

The mirror problem, and usually the larger number. Capital sitting in items that have not moved is invisible on a P&L and painful on a balance sheet. Business Central holds everything needed to quantify it: last movement date, quantity, value and turnover per item.

Useful output ranks by money rather than by count. Four hundred slow-moving SKUs is a report nobody acts on; “€180,000 of stock has not moved in a year, and 60% of it is in eleven items” is a decision. The follow-up questions — is it obsolete, seasonal, or tied to a contract — are exactly what an assistant can answer in the same conversation.

Purchasing recommendations

Business Central has planning worksheets and requisition worksheets, and they work. Where AI adds value is in the reasoning around them: explaining why a suggestion appeared, flagging when a suggestion is based on unusual demand history such as a one-off large order, and combining the recommendation with supplier lead-time reality rather than the configured value.

Worth being direct about the boundary. If your item setup is properly maintained — reorder points, lead times, order modifiers current — the standard planning engine is good and AI should explain its output rather than replace it. If item setup has drifted, AI analysis will inherit the drift. Neither situation is improved by treating the model as a substitute for maintained master data.

Where inventory analysis gets subtle

Inventory value depends on your costing method, and analysis that ignores this produces numbers that disagree with the ledger. FIFO, average and standard behave differently, and figures pulled before cost adjustment has run are provisional. An assistant should report that caveat rather than present a provisional number as final. The mechanics are covered in inventory costing methods.

Locations add a second layer: stock that exists in the company but sits in the wrong warehouse is not available for a specific order. Analysis reported at company level when your business runs at location level is technically correct and operationally useless.

Inventory agents

An AI agent can extend this into action: creating purchase orders from approved suggestions, updating reorder points based on observed demand, or opening a supplier query when receipts drift. Because unposted purchase orders are recoverable, this is one of the areas where unattended creation with review before release is a reasonable starting position — more so than in finance.